go/src/reflect/value.go

2526 lines
72 KiB
Go
Raw Normal View History

// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package reflect
import (
"math"
"runtime"
"unsafe"
)
runtime: replace GC programs with simpler encoding, faster decoder Small types record the location of pointers in their memory layout by using a simple bitmap. In Go 1.4 the bitmap held 4-bit entries, and in Go 1.5 the bitmap holds 1-bit entries, but in both cases using a bitmap for a large type containing arrays does not make sense: if someone refers to the type [1<<28]*byte in a program in such a way that the type information makes it into the binary, it would be a waste of space to write a 128 MB (for 4-bit entries) or even 32 MB (for 1-bit entries) bitmap full of 1s into the binary or even to keep one in memory during the execution of the program. For large types containing arrays, it is much more compact to describe the locations of pointers using a notation that can express repetition than to lay out a bitmap of pointers. Go 1.4 included such a notation, called ``GC programs'' but it was complex, required recursion during decoding, and was generally slow. Dmitriy measured the execution of these programs writing directly to the heap bitmap as being 7x slower than copying from a preunrolled 4-bit mask (and frankly that code was not terribly fast either). For some tests, unrollgcprog1 was seen costing as much as 3x more than the rest of malloc combined. This CL introduces a different form for the GC programs. They use a simple Lempel-Ziv-style encoding of the 1-bit pointer information, in which the only operations are (1) emit the following n bits and (2) repeat the last n bits c more times. This encoding can be generated directly from the Go type information (using repetition only for arrays or large runs of non-pointer data) and it can be decoded very efficiently. In particular the decoding requires little state and no recursion, so that the entire decoding can run without any memory accesses other than the reads of the encoding and the writes of the decoded form to the heap bitmap. For recursive types like arrays of arrays of arrays, the inner instructions are only executed once, not n times, so that large repetitions run at full speed. (In contrast, large repetitions in the old programs repeated the individual bit-level layout of the inner data over and over.) The result is as much as 25x faster decoding compared to the old form. Because the old decoder was so slow, Go 1.4 had three (or so) cases for how to set the heap bitmap bits for an allocation of a given type: (1) If the type had an even number of words up to 32 words, then the 4-bit pointer mask for the type fit in no more than 16 bytes; store the 4-bit pointer mask directly in the binary and copy from it. (1b) If the type had an odd number of words up to 15 words, then the 4-bit pointer mask for the type, doubled to end on a byte boundary, fit in no more than 16 bytes; store that doubled mask directly in the binary and copy from it. (2) If the type had an even number of words up to 128 words, or an odd number of words up to 63 words (again due to doubling), then the 4-bit pointer mask would fit in a 64-byte unrolled mask. Store a GC program in the binary, but leave space in the BSS for the unrolled mask. Execute the GC program to construct the mask the first time it is needed, and thereafter copy from the mask. (3) Otherwise, store a GC program and execute it to write directly to the heap bitmap each time an object of that type is allocated. (This is the case that was 7x slower than the other two.) Because the new pointer masks store 1-bit entries instead of 4-bit entries and because using the decoder no longer carries a significant overhead, after this CL (that is, for Go 1.5) there are only two cases: (1) If the type is 128 words or less (no condition about odd or even), store the 1-bit pointer mask directly in the binary and use it to initialize the heap bitmap during malloc. (Implemented in CL 9702.) (2) There is no case 2 anymore. (3) Otherwise, store a GC program and execute it to write directly to the heap bitmap each time an object of that type is allocated. Executing the GC program directly into the heap bitmap (case (3) above) was disabled for the Go 1.5 dev cycle, both to avoid needing to use GC programs for typedmemmove and to avoid updating that code as the heap bitmap format changed. Typedmemmove no longer uses this type information; as of CL 9886 it uses the heap bitmap directly. Now that the heap bitmap format is stable, we reintroduce GC programs and their space savings. Benchmarks for heapBitsSetType, before this CL vs this CL: name old mean new mean delta SetTypePtr 7.59ns × (0.99,1.02) 5.16ns × (1.00,1.00) -32.05% (p=0.000) SetTypePtr8 21.0ns × (0.98,1.05) 21.4ns × (1.00,1.00) ~ (p=0.179) SetTypePtr16 24.1ns × (0.99,1.01) 24.6ns × (1.00,1.00) +2.41% (p=0.001) SetTypePtr32 31.2ns × (0.99,1.01) 32.4ns × (0.99,1.02) +3.72% (p=0.001) SetTypePtr64 45.2ns × (1.00,1.00) 47.2ns × (1.00,1.00) +4.42% (p=0.000) SetTypePtr126 75.8ns × (0.99,1.01) 79.1ns × (1.00,1.00) +4.25% (p=0.000) SetTypePtr128 74.3ns × (0.99,1.01) 77.6ns × (1.00,1.01) +4.55% (p=0.000) SetTypePtrSlice 726ns × (1.00,1.01) 712ns × (1.00,1.00) -1.95% (p=0.001) SetTypeNode1 20.0ns × (0.99,1.01) 20.7ns × (1.00,1.00) +3.71% (p=0.000) SetTypeNode1Slice 112ns × (1.00,1.00) 113ns × (0.99,1.00) ~ (p=0.070) SetTypeNode8 23.9ns × (1.00,1.00) 24.7ns × (1.00,1.01) +3.18% (p=0.000) SetTypeNode8Slice 294ns × (0.99,1.02) 287ns × (0.99,1.01) -2.38% (p=0.015) SetTypeNode64 52.8ns × (0.99,1.03) 51.8ns × (0.99,1.01) ~ (p=0.069) SetTypeNode64Slice 1.13µs × (0.99,1.05) 1.14µs × (0.99,1.00) ~ (p=0.767) SetTypeNode64Dead 36.0ns × (1.00,1.01) 32.5ns × (0.99,1.00) -9.67% (p=0.000) SetTypeNode64DeadSlice 1.43µs × (0.99,1.01) 1.40µs × (1.00,1.00) -2.39% (p=0.001) SetTypeNode124 75.7ns × (1.00,1.01) 79.0ns × (1.00,1.00) +4.44% (p=0.000) SetTypeNode124Slice 1.94µs × (1.00,1.01) 2.04µs × (0.99,1.01) +4.98% (p=0.000) SetTypeNode126 75.4ns × (1.00,1.01) 77.7ns × (0.99,1.01) +3.11% (p=0.000) SetTypeNode126Slice 1.95µs × (0.99,1.01) 2.03µs × (1.00,1.00) +3.74% (p=0.000) SetTypeNode128 85.4ns × (0.99,1.01) 122.0ns × (1.00,1.00) +42.89% (p=0.000) SetTypeNode128Slice 2.20µs × (1.00,1.01) 2.36µs × (0.98,1.02) +7.48% (p=0.001) SetTypeNode130 83.3ns × (1.00,1.00) 123.0ns × (1.00,1.00) +47.61% (p=0.000) SetTypeNode130Slice 2.30µs × (0.99,1.01) 2.40µs × (0.98,1.01) +4.37% (p=0.000) SetTypeNode1024 498ns × (1.00,1.00) 537ns × (1.00,1.00) +7.96% (p=0.000) SetTypeNode1024Slice 15.5µs × (0.99,1.01) 17.8µs × (1.00,1.00) +15.27% (p=0.000) The above compares always using a cached pointer mask (and the corresponding waste of memory) against using the programs directly. Some slowdown is expected, in exchange for having a better general algorithm. The GC programs kick in for SetTypeNode128, SetTypeNode130, SetTypeNode1024, along with the slice variants of those. It is possible that the cutoff of 128 words (bits) should be raised in a followup CL, but even with this low cutoff the GC programs are faster than Go 1.4's "fast path" non-GC program case. Benchmarks for heapBitsSetType, Go 1.4 vs this CL: name old mean new mean delta SetTypePtr 6.89ns × (1.00,1.00) 5.17ns × (1.00,1.00) -25.02% (p=0.000) SetTypePtr8 25.8ns × (0.97,1.05) 21.5ns × (1.00,1.00) -16.70% (p=0.000) SetTypePtr16 39.8ns × (0.97,1.02) 24.7ns × (0.99,1.01) -37.81% (p=0.000) SetTypePtr32 68.8ns × (0.98,1.01) 32.2ns × (1.00,1.01) -53.18% (p=0.000) SetTypePtr64 130ns × (1.00,1.00) 47ns × (1.00,1.00) -63.67% (p=0.000) SetTypePtr126 241ns × (0.99,1.01) 79ns × (1.00,1.01) -67.25% (p=0.000) SetTypePtr128 2.07µs × (1.00,1.00) 0.08µs × (1.00,1.00) -96.27% (p=0.000) SetTypePtrSlice 1.05µs × (0.99,1.01) 0.72µs × (0.99,1.02) -31.70% (p=0.000) SetTypeNode1 16.0ns × (0.99,1.01) 20.8ns × (0.99,1.03) +29.91% (p=0.000) SetTypeNode1Slice 184ns × (0.99,1.01) 112ns × (0.99,1.01) -39.26% (p=0.000) SetTypeNode8 29.5ns × (0.97,1.02) 24.6ns × (1.00,1.00) -16.50% (p=0.000) SetTypeNode8Slice 624ns × (0.98,1.02) 285ns × (1.00,1.00) -54.31% (p=0.000) SetTypeNode64 135ns × (0.96,1.08) 52ns × (0.99,1.02) -61.32% (p=0.000) SetTypeNode64Slice 3.83µs × (1.00,1.00) 1.14µs × (0.99,1.01) -70.16% (p=0.000) SetTypeNode64Dead 134ns × (0.99,1.01) 32ns × (1.00,1.01) -75.74% (p=0.000) SetTypeNode64DeadSlice 3.83µs × (0.99,1.00) 1.40µs × (1.00,1.01) -63.42% (p=0.000) SetTypeNode124 240ns × (0.99,1.01) 79ns × (1.00,1.01) -67.05% (p=0.000) SetTypeNode124Slice 7.27µs × (1.00,1.00) 2.04µs × (1.00,1.00) -71.95% (p=0.000) SetTypeNode126 2.06µs × (0.99,1.01) 0.08µs × (0.99,1.01) -96.23% (p=0.000) SetTypeNode126Slice 64.4µs × (1.00,1.00) 2.0µs × (1.00,1.00) -96.85% (p=0.000) SetTypeNode128 2.09µs × (1.00,1.01) 0.12µs × (1.00,1.00) -94.15% (p=0.000) SetTypeNode128Slice 65.4µs × (1.00,1.00) 2.4µs × (0.99,1.03) -96.39% (p=0.000) SetTypeNode130 2.11µs × (1.00,1.00) 0.12µs × (1.00,1.00) -94.18% (p=0.000) SetTypeNode130Slice 66.3µs × (1.00,1.00) 2.4µs × (0.97,1.08) -96.34% (p=0.000) SetTypeNode1024 16.0µs × (1.00,1.01) 0.5µs × (1.00,1.00) -96.65% (p=0.000) SetTypeNode1024Slice 512µs × (1.00,1.00) 18µs × (0.98,1.04) -96.45% (p=0.000) SetTypeNode124 uses a 124 data + 2 ptr = 126-word allocation. Both Go 1.4 and this CL are using pointer bitmaps for this case, so that's an overall 3x speedup for using pointer bitmaps. SetTypeNode128 uses a 128 data + 2 ptr = 130-word allocation. Both Go 1.4 and this CL are running the GC program for this case, so that's an overall 17x speedup when using GC programs (and I've seen >20x on other systems). Comparing Go 1.4's SetTypeNode124 (pointer bitmap) against this CL's SetTypeNode128 (GC program), the slow path in the code in this CL is 2x faster than the fast path in Go 1.4. The Go 1 benchmarks are basically unaffected compared to just before this CL. Go 1 benchmarks, before this CL vs this CL: name old mean new mean delta BinaryTree17 5.87s × (0.97,1.04) 5.91s × (0.96,1.04) ~ (p=0.306) Fannkuch11 4.38s × (1.00,1.00) 4.37s × (1.00,1.01) -0.22% (p=0.006) FmtFprintfEmpty 90.7ns × (0.97,1.10) 89.3ns × (0.96,1.09) ~ (p=0.280) FmtFprintfString 282ns × (0.98,1.04) 287ns × (0.98,1.07) +1.72% (p=0.039) FmtFprintfInt 269ns × (0.99,1.03) 282ns × (0.97,1.04) +4.87% (p=0.000) FmtFprintfIntInt 478ns × (0.99,1.02) 481ns × (0.99,1.02) +0.61% (p=0.048) FmtFprintfPrefixedInt 399ns × (0.98,1.03) 400ns × (0.98,1.05) ~ (p=0.533) FmtFprintfFloat 563ns × (0.99,1.01) 570ns × (1.00,1.01) +1.37% (p=0.000) FmtManyArgs 1.89µs × (0.99,1.01) 1.92µs × (0.99,1.02) +1.88% (p=0.000) GobDecode 15.2ms × (0.99,1.01) 15.2ms × (0.98,1.05) ~ (p=0.609) GobEncode 11.6ms × (0.98,1.03) 11.9ms × (0.98,1.04) +2.17% (p=0.000) Gzip 648ms × (0.99,1.01) 648ms × (1.00,1.01) ~ (p=0.835) Gunzip 142ms × (1.00,1.00) 143ms × (1.00,1.01) ~ (p=0.169) HTTPClientServer 90.5µs × (0.98,1.03) 91.5µs × (0.98,1.04) +1.04% (p=0.045) JSONEncode 31.5ms × (0.98,1.03) 31.4ms × (0.98,1.03) ~ (p=0.549) JSONDecode 111ms × (0.99,1.01) 107ms × (0.99,1.01) -3.21% (p=0.000) Mandelbrot200 6.01ms × (1.00,1.00) 6.01ms × (1.00,1.00) ~ (p=0.878) GoParse 6.54ms × (0.99,1.02) 6.61ms × (0.99,1.03) +1.08% (p=0.004) RegexpMatchEasy0_32 160ns × (1.00,1.01) 161ns × (1.00,1.00) +0.40% (p=0.000) RegexpMatchEasy0_1K 560ns × (0.99,1.01) 559ns × (0.99,1.01) ~ (p=0.088) RegexpMatchEasy1_32 138ns × (0.99,1.01) 138ns × (1.00,1.00) ~ (p=0.380) RegexpMatchEasy1_1K 877ns × (1.00,1.00) 878ns × (1.00,1.00) ~ (p=0.157) RegexpMatchMedium_32 251ns × (0.99,1.00) 251ns × (1.00,1.01) +0.28% (p=0.021) RegexpMatchMedium_1K 72.6µs × (1.00,1.00) 72.6µs × (1.00,1.00) ~ (p=0.539) RegexpMatchHard_32 3.84µs × (1.00,1.00) 3.84µs × (1.00,1.00) ~ (p=0.378) RegexpMatchHard_1K 117µs × (1.00,1.00) 117µs × (1.00,1.00) ~ (p=0.067) Revcomp 904ms × (0.99,1.02) 904ms × (0.99,1.01) ~ (p=0.943) Template 125ms × (0.99,1.02) 127ms × (0.99,1.01) +1.79% (p=0.000) TimeParse 627ns × (0.99,1.01) 622ns × (0.99,1.01) -0.88% (p=0.000) TimeFormat 655ns × (0.99,1.02) 655ns × (0.99,1.02) ~ (p=0.976) For the record, Go 1 benchmarks, Go 1.4 vs this CL: name old mean new mean delta BinaryTree17 4.61s × (0.97,1.05) 5.91s × (0.98,1.03) +28.35% (p=0.000) Fannkuch11 4.40s × (0.99,1.03) 4.41s × (0.99,1.01) ~ (p=0.212) FmtFprintfEmpty 102ns × (0.99,1.01) 84ns × (0.99,1.02) -18.38% (p=0.000) FmtFprintfString 302ns × (0.98,1.01) 303ns × (0.99,1.02) ~ (p=0.203) FmtFprintfInt 313ns × (0.97,1.05) 270ns × (0.99,1.01) -13.69% (p=0.000) FmtFprintfIntInt 524ns × (0.98,1.02) 477ns × (0.99,1.00) -8.87% (p=0.000) FmtFprintfPrefixedInt 424ns × (0.98,1.02) 386ns × (0.99,1.01) -8.96% (p=0.000) FmtFprintfFloat 652ns × (0.98,1.02) 594ns × (0.97,1.05) -8.97% (p=0.000) FmtManyArgs 2.13µs × (0.99,1.02) 1.94µs × (0.99,1.01) -8.92% (p=0.000) GobDecode 17.1ms × (0.99,1.02) 14.9ms × (0.98,1.03) -13.07% (p=0.000) GobEncode 13.5ms × (0.98,1.03) 11.5ms × (0.98,1.03) -15.25% (p=0.000) Gzip 656ms × (0.99,1.02) 647ms × (0.99,1.01) -1.29% (p=0.000) Gunzip 143ms × (0.99,1.02) 144ms × (0.99,1.01) ~ (p=0.204) HTTPClientServer 88.2µs × (0.98,1.02) 90.8µs × (0.98,1.01) +2.93% (p=0.000) JSONEncode 32.2ms × (0.98,1.02) 30.9ms × (0.97,1.04) -4.06% (p=0.001) JSONDecode 121ms × (0.98,1.02) 110ms × (0.98,1.05) -8.95% (p=0.000) Mandelbrot200 6.06ms × (0.99,1.01) 6.11ms × (0.98,1.04) ~ (p=0.184) GoParse 6.76ms × (0.97,1.04) 6.58ms × (0.98,1.05) -2.63% (p=0.003) RegexpMatchEasy0_32 195ns × (1.00,1.01) 155ns × (0.99,1.01) -20.43% (p=0.000) RegexpMatchEasy0_1K 479ns × (0.98,1.03) 535ns × (0.99,1.02) +11.59% (p=0.000) RegexpMatchEasy1_32 169ns × (0.99,1.02) 131ns × (0.99,1.03) -22.44% (p=0.000) RegexpMatchEasy1_1K 1.53µs × (0.99,1.01) 0.87µs × (0.99,1.02) -43.07% (p=0.000) RegexpMatchMedium_32 334ns × (0.99,1.01) 242ns × (0.99,1.01) -27.53% (p=0.000) RegexpMatchMedium_1K 125µs × (1.00,1.01) 72µs × (0.99,1.03) -42.53% (p=0.000) RegexpMatchHard_32 6.03µs × (0.99,1.01) 3.79µs × (0.99,1.01) -37.12% (p=0.000) RegexpMatchHard_1K 189µs × (0.99,1.02) 115µs × (0.99,1.01) -39.20% (p=0.000) Revcomp 935ms × (0.96,1.03) 926ms × (0.98,1.02) ~ (p=0.083) Template 146ms × (0.97,1.05) 119ms × (0.99,1.01) -18.37% (p=0.000) TimeParse 660ns × (0.99,1.01) 624ns × (0.99,1.02) -5.43% (p=0.000) TimeFormat 670ns × (0.98,1.02) 710ns × (1.00,1.01) +5.97% (p=0.000) This CL is a bit larger than I would like, but the compiler, linker, runtime, and package reflect all need to be in sync about the format of these programs, so there is no easy way to split this into independent changes (at least while keeping the build working at each change). Fixes #9625. Fixes #10524. Change-Id: I9e3e20d6097099d0f8532d1cb5b1af528804989a Reviewed-on: https://go-review.googlesource.com/9888 Reviewed-by: Austin Clements <austin@google.com> Run-TryBot: Russ Cox <rsc@golang.org>
2015-05-08 01:43:18 -04:00
const ptrSize = 4 << (^uintptr(0) >> 63) // unsafe.Sizeof(uintptr(0)) but an ideal const
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// Value is the reflection interface to a Go value.
//
// Not all methods apply to all kinds of values. Restrictions,
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// if any, are noted in the documentation for each method.
// Use the Kind method to find out the kind of value before
// calling kind-specific methods. Calling a method
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// inappropriate to the kind of type causes a run time panic.
//
// The zero Value represents no value.
// Its IsValid method returns false, its Kind method returns Invalid,
// its String method returns "<invalid Value>", and all other methods panic.
// Most functions and methods never return an invalid value.
// If one does, its documentation states the conditions explicitly.
//
// A Value can be used concurrently by multiple goroutines provided that
// the underlying Go value can be used concurrently for the equivalent
// direct operations.
//
// Using == on two Values does not compare the underlying values
// they represent, but rather the contents of the Value structs.
// To compare two Values, compare the results of the Interface method.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
type Value struct {
// typ holds the type of the value represented by a Value.
typ *rtype
// Pointer-valued data or, if flagIndir is set, pointer to data.
// Valid when either flagIndir is set or typ.pointers() is true.
ptr unsafe.Pointer
// flag holds metadata about the value.
// The lowest bits are flag bits:
// - flagStickyRO: obtained via unexported not embedded field, so read-only
// - flagEmbedRO: obtained via unexported embedded field, so read-only
// - flagIndir: val holds a pointer to the data
// - flagAddr: v.CanAddr is true (implies flagIndir)
// - flagMethod: v is a method value.
// The next five bits give the Kind of the value.
// This repeats typ.Kind() except for method values.
// The remaining 23+ bits give a method number for method values.
// If flag.kind() != Func, code can assume that flagMethod is unset.
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
// If ifaceIndir(typ), code can assume that flagIndir is set.
flag
// A method value represents a curried method invocation
// like r.Read for some receiver r. The typ+val+flag bits describe
// the receiver r, but the flag's Kind bits say Func (methods are
// functions), and the top bits of the flag give the method number
// in r's type's method table.
}
type flag uintptr
const (
flagKindWidth = 5 // there are 27 kinds
flagKindMask flag = 1<<flagKindWidth - 1
flagStickyRO flag = 1 << 5
flagEmbedRO flag = 1 << 6
flagIndir flag = 1 << 7
flagAddr flag = 1 << 8
flagMethod flag = 1 << 9
flagMethodShift = 10
flagRO flag = flagStickyRO | flagEmbedRO
)
func (f flag) kind() Kind {
return Kind(f & flagKindMask)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// pointer returns the underlying pointer represented by v.
// v.Kind() must be Ptr, Map, Chan, Func, or UnsafePointer
func (v Value) pointer() unsafe.Pointer {
if v.typ.size != ptrSize || !v.typ.pointers() {
panic("can't call pointer on a non-pointer Value")
}
if v.flag&flagIndir != 0 {
return *(*unsafe.Pointer)(v.ptr)
}
return v.ptr
}
// packEface converts v to the empty interface.
func packEface(v Value) interface{} {
t := v.typ
var i interface{}
e := (*emptyInterface)(unsafe.Pointer(&i))
// First, fill in the data portion of the interface.
switch {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
case ifaceIndir(t):
if v.flag&flagIndir == 0 {
panic("bad indir")
}
// Value is indirect, and so is the interface we're making.
ptr := v.ptr
if v.flag&flagAddr != 0 {
// TODO: pass safe boolean from valueInterface so
// we don't need to copy if safe==true?
c := unsafe_New(t)
typedmemmove(t, c, ptr)
ptr = c
}
e.word = ptr
case v.flag&flagIndir != 0:
// Value is indirect, but interface is direct. We need
// to load the data at v.ptr into the interface data word.
e.word = *(*unsafe.Pointer)(v.ptr)
default:
// Value is direct, and so is the interface.
e.word = v.ptr
}
// Now, fill in the type portion. We're very careful here not
// to have any operation between the e.word and e.typ assignments
// that would let the garbage collector observe the partially-built
// interface value.
e.typ = t
return i
}
// unpackEface converts the empty interface i to a Value.
func unpackEface(i interface{}) Value {
e := (*emptyInterface)(unsafe.Pointer(&i))
// NOTE: don't read e.word until we know whether it is really a pointer or not.
t := e.typ
if t == nil {
return Value{}
}
f := flag(t.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(t) {
f |= flagIndir
}
return Value{t, e.word, f}
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// A ValueError occurs when a Value method is invoked on
// a Value that does not support it. Such cases are documented
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// in the description of each method.
type ValueError struct {
Method string
Kind Kind
}
func (e *ValueError) Error() string {
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
if e.Kind == 0 {
return "reflect: call of " + e.Method + " on zero Value"
}
return "reflect: call of " + e.Method + " on " + e.Kind.String() + " Value"
}
// methodName returns the name of the calling method,
// assumed to be two stack frames above.
func methodName() string {
pc, _, _, _ := runtime.Caller(2)
f := runtime.FuncForPC(pc)
if f == nil {
return "unknown method"
}
return f.Name()
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// emptyInterface is the header for an interface{} value.
type emptyInterface struct {
typ *rtype
word unsafe.Pointer
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
// nonEmptyInterface is the header for a interface value with methods.
type nonEmptyInterface struct {
// see ../runtime/iface.go:/Itab
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
itab *struct {
ityp *rtype // static interface type
typ *rtype // dynamic concrete type
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
link unsafe.Pointer
bad int32
unused int32
fun [100000]unsafe.Pointer // method table
}
word unsafe.Pointer
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
// mustBe panics if f's kind is not expected.
// Making this a method on flag instead of on Value
// (and embedding flag in Value) means that we can write
// the very clear v.mustBe(Bool) and have it compile into
// v.flag.mustBe(Bool), which will only bother to copy the
// single important word for the receiver.
func (f flag) mustBe(expected Kind) {
if f.kind() != expected {
panic(&ValueError{methodName(), f.kind()})
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
}
// mustBeExported panics if f records that the value was obtained using
// an unexported field.
func (f flag) mustBeExported() {
if f == 0 {
panic(&ValueError{methodName(), 0})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
if f&flagRO != 0 {
panic("reflect: " + methodName() + " using value obtained using unexported field")
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
}
// mustBeAssignable panics if f records that the value is not assignable,
// which is to say that either it was obtained using an unexported field
// or it is not addressable.
func (f flag) mustBeAssignable() {
if f == 0 {
panic(&ValueError{methodName(), Invalid})
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// Assignable if addressable and not read-only.
if f&flagRO != 0 {
panic("reflect: " + methodName() + " using value obtained using unexported field")
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
if f&flagAddr == 0 {
panic("reflect: " + methodName() + " using unaddressable value")
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
}
// Addr returns a pointer value representing the address of v.
// It panics if CanAddr() returns false.
// Addr is typically used to obtain a pointer to a struct field
// or slice element in order to call a method that requires a
// pointer receiver.
func (v Value) Addr() Value {
if v.flag&flagAddr == 0 {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect.Value.Addr of unaddressable value")
}
return Value{v.typ.ptrTo(), v.ptr, (v.flag & flagRO) | flag(Ptr)}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Bool returns v's underlying value.
// It panics if v's kind is not Bool.
func (v Value) Bool() bool {
v.mustBe(Bool)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return *(*bool)(v.ptr)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Bytes returns v's underlying value.
// It panics if v's underlying value is not a slice of bytes.
func (v Value) Bytes() []byte {
v.mustBe(Slice)
if v.typ.Elem().Kind() != Uint8 {
panic("reflect.Value.Bytes of non-byte slice")
}
// Slice is always bigger than a word; assume flagIndir.
return *(*[]byte)(v.ptr)
}
// runes returns v's underlying value.
// It panics if v's underlying value is not a slice of runes (int32s).
func (v Value) runes() []rune {
v.mustBe(Slice)
if v.typ.Elem().Kind() != Int32 {
panic("reflect.Value.Bytes of non-rune slice")
}
// Slice is always bigger than a word; assume flagIndir.
return *(*[]rune)(v.ptr)
}
// CanAddr reports whether the value's address can be obtained with Addr.
// Such values are called addressable. A value is addressable if it is
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// an element of a slice, an element of an addressable array,
// a field of an addressable struct, or the result of dereferencing a pointer.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// If CanAddr returns false, calling Addr will panic.
func (v Value) CanAddr() bool {
return v.flag&flagAddr != 0
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// CanSet reports whether the value of v can be changed.
// A Value can be changed only if it is addressable and was not
// obtained by the use of unexported struct fields.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// If CanSet returns false, calling Set or any type-specific
// setter (e.g., SetBool, SetInt) will panic.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) CanSet() bool {
return v.flag&(flagAddr|flagRO) == flagAddr
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Call calls the function v with the input arguments in.
// For example, if len(in) == 3, v.Call(in) represents the Go call v(in[0], in[1], in[2]).
// Call panics if v's Kind is not Func.
// It returns the output results as Values.
// As in Go, each input argument must be assignable to the
// type of the function's corresponding input parameter.
// If v is a variadic function, Call creates the variadic slice parameter
// itself, copying in the corresponding values.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Call(in []Value) []Value {
v.mustBe(Func)
v.mustBeExported()
return v.call("Call", in)
}
// CallSlice calls the variadic function v with the input arguments in,
// assigning the slice in[len(in)-1] to v's final variadic argument.
// For example, if len(in) == 3, v.CallSlice(in) represents the Go call v(in[0], in[1], in[2]...).
// CallSlice panics if v's Kind is not Func or if v is not variadic.
// It returns the output results as Values.
// As in Go, each input argument must be assignable to the
// type of the function's corresponding input parameter.
func (v Value) CallSlice(in []Value) []Value {
v.mustBe(Func)
v.mustBeExported()
return v.call("CallSlice", in)
}
reflect, runtime: fix crash in GC due to reflect.call + precise GC Given type Outer struct { *Inner ... } the compiler generates the implementation of (*Outer).M dispatching to the embedded Inner. The implementation is logically: func (p *Outer) M() { (p.Inner).M() } but since the only change here is the replacement of one pointer receiver with another, the actual generated code overwrites the original receiver with the p.Inner pointer and then jumps to the M method expecting the *Inner receiver. During reflect.Value.Call, we create an argument frame and the associated data structures to describe it to the garbage collector, populate the frame, call reflect.call to run a function call using that frame, and then copy the results back out of the frame. The reflect.call function does a memmove of the frame structure onto the stack (to set up the inputs), runs the call, and the memmoves the stack back to the frame structure (to preserve the outputs). Originally reflect.call did not distinguish inputs from outputs: both memmoves were for the full stack frame. However, in the case where the called function was one of these wrappers, the rewritten receiver is almost certainly a different type than the original receiver. This is not a problem on the stack, where we use the program counter to determine the type information and understand that during (*Outer).M the receiver is an *Outer while during (*Inner).M the receiver in the same memory word is now an *Inner. But in the statically typed argument frame created by reflect, the receiver is always an *Outer. Copying the modified receiver pointer off the stack into the frame will store an *Inner there, and then if a garbage collection happens to scan that argument frame before it is discarded, it will scan the *Inner memory as if it were an *Outer. If the two have different memory layouts, the collection will intepret the memory incorrectly. Fix by only copying back the results. Fixes #7725. LGTM=khr R=khr CC=dave, golang-codereviews https://golang.org/cl/85180043
2014-04-08 11:11:35 -04:00
var callGC bool // for testing; see TestCallMethodJump
func (v Value) call(op string, in []Value) []Value {
// Get function pointer, type.
t := v.typ
var (
fn unsafe.Pointer
rcvr Value
rcvrtype *rtype
)
if v.flag&flagMethod != 0 {
rcvr = v
rcvrtype, t, fn = methodReceiver(op, v, int(v.flag)>>flagMethodShift)
} else if v.flag&flagIndir != 0 {
fn = *(*unsafe.Pointer)(v.ptr)
} else {
fn = v.ptr
}
if fn == nil {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect.Value.Call: call of nil function")
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
isSlice := op == "CallSlice"
n := t.NumIn()
if isSlice {
if !t.IsVariadic() {
panic("reflect: CallSlice of non-variadic function")
}
if len(in) < n {
panic("reflect: CallSlice with too few input arguments")
}
if len(in) > n {
panic("reflect: CallSlice with too many input arguments")
}
} else {
if t.IsVariadic() {
n--
}
if len(in) < n {
panic("reflect: Call with too few input arguments")
}
if !t.IsVariadic() && len(in) > n {
panic("reflect: Call with too many input arguments")
}
}
for _, x := range in {
if x.Kind() == Invalid {
panic("reflect: " + op + " using zero Value argument")
}
}
for i := 0; i < n; i++ {
if xt, targ := in[i].Type(), t.In(i); !xt.AssignableTo(targ) {
panic("reflect: " + op + " using " + xt.String() + " as type " + targ.String())
}
}
if !isSlice && t.IsVariadic() {
// prepare slice for remaining values
m := len(in) - n
slice := MakeSlice(t.In(n), m, m)
elem := t.In(n).Elem()
for i := 0; i < m; i++ {
x := in[n+i]
if xt := x.Type(); !xt.AssignableTo(elem) {
panic("reflect: cannot use " + xt.String() + " as type " + elem.String() + " in " + op)
}
slice.Index(i).Set(x)
}
origIn := in
in = make([]Value, n+1)
copy(in[:n], origIn)
in[n] = slice
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
nin := len(in)
if nin != t.NumIn() {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect.Value.Call: wrong argument count")
}
nout := t.NumOut()
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// Compute frame type.
frametype, _, retOffset, _, framePool := funcLayout(t, rcvrtype)
// Allocate a chunk of memory for frame.
var args unsafe.Pointer
if nout == 0 {
args = framePool.Get().(unsafe.Pointer)
} else {
// Can't use pool if the function has return values.
// We will leak pointer to args in ret, so its lifetime is not scoped.
args = unsafe_New(frametype)
}
off := uintptr(0)
// Copy inputs into args.
if rcvrtype != nil {
storeRcvr(rcvr, args)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
off = ptrSize
}
for i, v := range in {
v.mustBeExported()
targ := t.In(i).(*rtype)
a := uintptr(targ.align)
off = (off + a - 1) &^ (a - 1)
n := targ.size
addr := unsafe.Pointer(uintptr(args) + off)
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
v = v.assignTo("reflect.Value.Call", targ, addr)
if v.flag&flagIndir != 0 {
typedmemmove(targ, addr, v.ptr)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*unsafe.Pointer)(addr) = v.ptr
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
off += n
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// Call.
call(frametype, fn, args, uint32(frametype.size), uint32(retOffset))
reflect, runtime: fix crash in GC due to reflect.call + precise GC Given type Outer struct { *Inner ... } the compiler generates the implementation of (*Outer).M dispatching to the embedded Inner. The implementation is logically: func (p *Outer) M() { (p.Inner).M() } but since the only change here is the replacement of one pointer receiver with another, the actual generated code overwrites the original receiver with the p.Inner pointer and then jumps to the M method expecting the *Inner receiver. During reflect.Value.Call, we create an argument frame and the associated data structures to describe it to the garbage collector, populate the frame, call reflect.call to run a function call using that frame, and then copy the results back out of the frame. The reflect.call function does a memmove of the frame structure onto the stack (to set up the inputs), runs the call, and the memmoves the stack back to the frame structure (to preserve the outputs). Originally reflect.call did not distinguish inputs from outputs: both memmoves were for the full stack frame. However, in the case where the called function was one of these wrappers, the rewritten receiver is almost certainly a different type than the original receiver. This is not a problem on the stack, where we use the program counter to determine the type information and understand that during (*Outer).M the receiver is an *Outer while during (*Inner).M the receiver in the same memory word is now an *Inner. But in the statically typed argument frame created by reflect, the receiver is always an *Outer. Copying the modified receiver pointer off the stack into the frame will store an *Inner there, and then if a garbage collection happens to scan that argument frame before it is discarded, it will scan the *Inner memory as if it were an *Outer. If the two have different memory layouts, the collection will intepret the memory incorrectly. Fix by only copying back the results. Fixes #7725. LGTM=khr R=khr CC=dave, golang-codereviews https://golang.org/cl/85180043
2014-04-08 11:11:35 -04:00
// For testing; see TestCallMethodJump.
if callGC {
runtime.GC()
}
var ret []Value
if nout == 0 {
memclr(args, frametype.size)
framePool.Put(args)
} else {
// Zero the now unused input area of args,
// because the Values returned by this function contain pointers to the args object,
// and will thus keep the args object alive indefinitely.
memclr(args, retOffset)
// Copy return values out of args.
ret = make([]Value, nout)
off = retOffset
for i := 0; i < nout; i++ {
tv := t.Out(i)
a := uintptr(tv.Align())
off = (off + a - 1) &^ (a - 1)
fl := flagIndir | flag(tv.Kind())
ret[i] = Value{tv.common(), unsafe.Pointer(uintptr(args) + off), fl}
off += tv.Size()
}
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
return ret
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// callReflect is the call implementation used by a function
// returned by MakeFunc. In many ways it is the opposite of the
// method Value.call above. The method above converts a call using Values
// into a call of a function with a concrete argument frame, while
// callReflect converts a call of a function with a concrete argument
// frame into a call using Values.
// It is in this file so that it can be next to the call method above.
// The remainder of the MakeFunc implementation is in makefunc.go.
runtime, cmd/gc, cmd/ld: ignore method wrappers in recover Bug #1: Issue 5406 identified an interesting case: defer iface.M() may end up calling a wrapper that copies an indirect receiver from the iface value and then calls the real M method. That's two calls down, not just one, and so recover() == nil always in the real M method, even during a panic. [For the purposes of this entire discussion, a wrapper's implementation is a function containing an ordinary call, not the optimized tail call form that is somtimes possible. The tail call does not create a second frame, so it is already handled correctly.] Fix this bug by introducing g->panicwrap, which counts the number of bytes on current stack segment that are due to wrapper calls that should not count against the recover check. All wrapper functions must now adjust g->panicwrap up on entry and back down on exit. This adds slightly to their expense; on the x86 it is a single instruction at entry and exit; on the ARM it is three. However, the alternative is to make a call to recover depend on being able to walk the stack, which I very much want to avoid. We have enough problems walking the stack for garbage collection and profiling. Also, if performance is critical in a specific case, it is already faster to use a pointer receiver and avoid this kind of wrapper entirely. Bug #2: The old code, which did not consider the possibility of two calls, already contained a check to see if the call had split its stack and so the panic-created segment was one behind the current segment. In the wrapper case, both of the two calls might split their stacks, so the panic-created segment can be two behind the current segment. Fix this by propagating the Stktop.panic flag forward during stack splits instead of looking backward during recover. Fixes #5406. R=golang-dev, iant CC=golang-dev https://golang.org/cl/13367052
2013-09-12 14:00:16 -04:00
//
// NOTE: This function must be marked as a "wrapper" in the generated code,
// so that the linker can make it work correctly for panic and recover.
// The gc compilers know to do that for the name "reflect.callReflect".
func callReflect(ctxt *makeFuncImpl, frame unsafe.Pointer) {
ftyp := ctxt.typ
f := ctxt.fn
// Copy argument frame into Values.
ptr := frame
off := uintptr(0)
in := make([]Value, 0, int(ftyp.inCount))
for _, typ := range ftyp.in() {
off += -off & uintptr(typ.align-1)
addr := unsafe.Pointer(uintptr(ptr) + off)
v := Value{typ, nil, flag(typ.Kind())}
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(typ) {
// value cannot be inlined in interface data.
// Must make a copy, because f might keep a reference to it,
// and we cannot let f keep a reference to the stack frame
// after this function returns, not even a read-only reference.
v.ptr = unsafe_New(typ)
typedmemmove(typ, v.ptr, addr)
v.flag |= flagIndir
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
v.ptr = *(*unsafe.Pointer)(addr)
}
in = append(in, v)
off += typ.size
}
// Call underlying function.
out := f(in)
numOut := ftyp.NumOut()
if len(out) != numOut {
panic("reflect: wrong return count from function created by MakeFunc")
}
// Copy results back into argument frame.
if numOut > 0 {
off += -off & (ptrSize - 1)
if runtime.GOARCH == "amd64p32" {
off = align(off, 8)
}
for i, typ := range ftyp.out() {
v := out[i]
if v.typ != typ {
panic("reflect: function created by MakeFunc using " + funcName(f) +
" returned wrong type: have " +
out[i].typ.String() + " for " + typ.String())
}
if v.flag&flagRO != 0 {
panic("reflect: function created by MakeFunc using " + funcName(f) +
" returned value obtained from unexported field")
}
off += -off & uintptr(typ.align-1)
addr := unsafe.Pointer(uintptr(ptr) + off)
if v.flag&flagIndir != 0 {
typedmemmove(typ, addr, v.ptr)
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*unsafe.Pointer)(addr) = v.ptr
}
off += typ.size
}
}
}
// methodReceiver returns information about the receiver
// described by v. The Value v may or may not have the
// flagMethod bit set, so the kind cached in v.flag should
// not be used.
// The return value rcvrtype gives the method's actual receiver type.
// The return value t gives the method type signature (without the receiver).
// The return value fn is a pointer to the method code.
func methodReceiver(op string, v Value, methodIndex int) (rcvrtype, t *rtype, fn unsafe.Pointer) {
i := methodIndex
if v.typ.Kind() == Interface {
tt := (*interfaceType)(unsafe.Pointer(v.typ))
if uint(i) >= uint(len(tt.methods)) {
panic("reflect: internal error: invalid method index")
}
m := &tt.methods[i]
if !m.name.isExported() {
panic("reflect: " + op + " of unexported method")
}
iface := (*nonEmptyInterface)(v.ptr)
if iface.itab == nil {
panic("reflect: " + op + " of method on nil interface value")
}
rcvrtype = iface.itab.typ
fn = unsafe.Pointer(&iface.itab.fun[i])
t = m.typ
} else {
rcvrtype = v.typ
ut := v.typ.uncommon()
cmd/compile, etc: store method tables as offsets This CL introduces the typeOff type and a lookup method of the same name that can turn a typeOff offset into an *rtype. In a typical Go binary (built with buildmode=exe, pie, c-archive, or c-shared), there is one moduledata and all typeOff values are offsets relative to firstmoduledata.types. This makes computing the pointer cheap in typical programs. With buildmode=shared (and one day, buildmode=plugin) there are multiple modules whose relative offset is determined at runtime. We identify a type in the general case by the pair of the original *rtype that references it and its typeOff value. We determine the module from the original pointer, and then use the typeOff from there to compute the final *rtype. To ensure there is only one *rtype representing each type, the runtime initializes a typemap for each module, using any identical type from an earlier module when resolving that offset. This means that types computed from an offset match the type mapped by the pointer dynamic relocations. A series of followup CLs will replace other *rtype values with typeOff (and name/*string with nameOff). For types created at runtime by reflect, type offsets are treated as global IDs and reference into a reflect offset map kept by the runtime. darwin/amd64: cmd/go: -57KB (0.6%) jujud: -557KB (0.8%) linux/amd64 PIE: cmd/go: -361KB (3.0%) jujud: -3.5MB (4.2%) For #6853. Change-Id: Icf096fd884a0a0cb9f280f46f7a26c70a9006c96 Reviewed-on: https://go-review.googlesource.com/21285 Reviewed-by: Ian Lance Taylor <iant@golang.org> Run-TryBot: David Crawshaw <crawshaw@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-03-28 10:32:27 -04:00
if ut == nil || uint(i) >= uint(ut.mcount) {
panic("reflect: internal error: invalid method index")
}
cmd/compile, etc: store method tables as offsets This CL introduces the typeOff type and a lookup method of the same name that can turn a typeOff offset into an *rtype. In a typical Go binary (built with buildmode=exe, pie, c-archive, or c-shared), there is one moduledata and all typeOff values are offsets relative to firstmoduledata.types. This makes computing the pointer cheap in typical programs. With buildmode=shared (and one day, buildmode=plugin) there are multiple modules whose relative offset is determined at runtime. We identify a type in the general case by the pair of the original *rtype that references it and its typeOff value. We determine the module from the original pointer, and then use the typeOff from there to compute the final *rtype. To ensure there is only one *rtype representing each type, the runtime initializes a typemap for each module, using any identical type from an earlier module when resolving that offset. This means that types computed from an offset match the type mapped by the pointer dynamic relocations. A series of followup CLs will replace other *rtype values with typeOff (and name/*string with nameOff). For types created at runtime by reflect, type offsets are treated as global IDs and reference into a reflect offset map kept by the runtime. darwin/amd64: cmd/go: -57KB (0.6%) jujud: -557KB (0.8%) linux/amd64 PIE: cmd/go: -361KB (3.0%) jujud: -3.5MB (4.2%) For #6853. Change-Id: Icf096fd884a0a0cb9f280f46f7a26c70a9006c96 Reviewed-on: https://go-review.googlesource.com/21285 Reviewed-by: Ian Lance Taylor <iant@golang.org> Run-TryBot: David Crawshaw <crawshaw@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-03-28 10:32:27 -04:00
m := ut.methods()[i]
if !m.name.isExported() {
panic("reflect: " + op + " of unexported method")
}
cmd/compile, etc: store method tables as offsets This CL introduces the typeOff type and a lookup method of the same name that can turn a typeOff offset into an *rtype. In a typical Go binary (built with buildmode=exe, pie, c-archive, or c-shared), there is one moduledata and all typeOff values are offsets relative to firstmoduledata.types. This makes computing the pointer cheap in typical programs. With buildmode=shared (and one day, buildmode=plugin) there are multiple modules whose relative offset is determined at runtime. We identify a type in the general case by the pair of the original *rtype that references it and its typeOff value. We determine the module from the original pointer, and then use the typeOff from there to compute the final *rtype. To ensure there is only one *rtype representing each type, the runtime initializes a typemap for each module, using any identical type from an earlier module when resolving that offset. This means that types computed from an offset match the type mapped by the pointer dynamic relocations. A series of followup CLs will replace other *rtype values with typeOff (and name/*string with nameOff). For types created at runtime by reflect, type offsets are treated as global IDs and reference into a reflect offset map kept by the runtime. darwin/amd64: cmd/go: -57KB (0.6%) jujud: -557KB (0.8%) linux/amd64 PIE: cmd/go: -361KB (3.0%) jujud: -3.5MB (4.2%) For #6853. Change-Id: Icf096fd884a0a0cb9f280f46f7a26c70a9006c96 Reviewed-on: https://go-review.googlesource.com/21285 Reviewed-by: Ian Lance Taylor <iant@golang.org> Run-TryBot: David Crawshaw <crawshaw@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-03-28 10:32:27 -04:00
ifn := v.typ.textOff(m.ifn)
fn = unsafe.Pointer(&ifn)
t = v.typ.typeOff(m.mtyp)
}
return
}
// v is a method receiver. Store at p the word which is used to
// encode that receiver at the start of the argument list.
// Reflect uses the "interface" calling convention for
// methods, which always uses one word to record the receiver.
func storeRcvr(v Value, p unsafe.Pointer) {
t := v.typ
if t.Kind() == Interface {
// the interface data word becomes the receiver word
iface := (*nonEmptyInterface)(v.ptr)
*(*unsafe.Pointer)(p) = iface.word
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
} else if v.flag&flagIndir != 0 && !ifaceIndir(t) {
*(*unsafe.Pointer)(p) = *(*unsafe.Pointer)(v.ptr)
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*unsafe.Pointer)(p) = v.ptr
}
}
// align returns the result of rounding x up to a multiple of n.
// n must be a power of two.
func align(x, n uintptr) uintptr {
return (x + n - 1) &^ (n - 1)
}
// callMethod is the call implementation used by a function returned
// by makeMethodValue (used by v.Method(i).Interface()).
// It is a streamlined version of the usual reflect call: the caller has
// already laid out the argument frame for us, so we don't have
// to deal with individual Values for each argument.
// It is in this file so that it can be next to the two similar functions above.
// The remainder of the makeMethodValue implementation is in makefunc.go.
runtime, cmd/gc, cmd/ld: ignore method wrappers in recover Bug #1: Issue 5406 identified an interesting case: defer iface.M() may end up calling a wrapper that copies an indirect receiver from the iface value and then calls the real M method. That's two calls down, not just one, and so recover() == nil always in the real M method, even during a panic. [For the purposes of this entire discussion, a wrapper's implementation is a function containing an ordinary call, not the optimized tail call form that is somtimes possible. The tail call does not create a second frame, so it is already handled correctly.] Fix this bug by introducing g->panicwrap, which counts the number of bytes on current stack segment that are due to wrapper calls that should not count against the recover check. All wrapper functions must now adjust g->panicwrap up on entry and back down on exit. This adds slightly to their expense; on the x86 it is a single instruction at entry and exit; on the ARM it is three. However, the alternative is to make a call to recover depend on being able to walk the stack, which I very much want to avoid. We have enough problems walking the stack for garbage collection and profiling. Also, if performance is critical in a specific case, it is already faster to use a pointer receiver and avoid this kind of wrapper entirely. Bug #2: The old code, which did not consider the possibility of two calls, already contained a check to see if the call had split its stack and so the panic-created segment was one behind the current segment. In the wrapper case, both of the two calls might split their stacks, so the panic-created segment can be two behind the current segment. Fix this by propagating the Stktop.panic flag forward during stack splits instead of looking backward during recover. Fixes #5406. R=golang-dev, iant CC=golang-dev https://golang.org/cl/13367052
2013-09-12 14:00:16 -04:00
//
// NOTE: This function must be marked as a "wrapper" in the generated code,
// so that the linker can make it work correctly for panic and recover.
// The gc compilers know to do that for the name "reflect.callMethod".
func callMethod(ctxt *methodValue, frame unsafe.Pointer) {
rcvr := ctxt.rcvr
rcvrtype, t, fn := methodReceiver("call", rcvr, ctxt.method)
frametype, argSize, retOffset, _, framePool := funcLayout(t, rcvrtype)
// Make a new frame that is one word bigger so we can store the receiver.
args := framePool.Get().(unsafe.Pointer)
// Copy in receiver and rest of args.
storeRcvr(rcvr, args)
typedmemmovepartial(frametype, unsafe.Pointer(uintptr(args)+ptrSize), frame, ptrSize, argSize-ptrSize)
// Call.
call(frametype, fn, args, uint32(frametype.size), uint32(retOffset))
// Copy return values. On amd64p32, the beginning of return values
// is 64-bit aligned, so the caller's frame layout (which doesn't have
// a receiver) is different from the layout of the fn call, which has
// a receiver.
// Ignore any changes to args and just copy return values.
callerRetOffset := retOffset - ptrSize
if runtime.GOARCH == "amd64p32" {
callerRetOffset = align(argSize-ptrSize, 8)
}
typedmemmovepartial(frametype,
unsafe.Pointer(uintptr(frame)+callerRetOffset),
unsafe.Pointer(uintptr(args)+retOffset),
retOffset,
frametype.size-retOffset)
memclr(args, frametype.size)
framePool.Put(args)
}
// funcName returns the name of f, for use in error messages.
func funcName(f func([]Value) []Value) string {
pc := *(*uintptr)(unsafe.Pointer(&f))
rf := runtime.FuncForPC(pc)
if rf != nil {
return rf.Name()
}
return "closure"
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// Cap returns v's capacity.
// It panics if v's Kind is not Array, Chan, or Slice.
func (v Value) Cap() int {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Array:
return v.typ.Len()
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Chan:
return int(chancap(v.pointer()))
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Slice:
// Slice is always bigger than a word; assume flagIndir.
return (*sliceHeader)(v.ptr).Cap
}
panic(&ValueError{"reflect.Value.Cap", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Close closes the channel v.
// It panics if v's Kind is not Chan.
func (v Value) Close() {
v.mustBe(Chan)
v.mustBeExported()
chanclose(v.pointer())
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Complex returns v's underlying value, as a complex128.
// It panics if v's Kind is not Complex64 or Complex128
func (v Value) Complex() complex128 {
k := v.kind()
switch k {
case Complex64:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return complex128(*(*complex64)(v.ptr))
case Complex128:
return *(*complex128)(v.ptr)
}
panic(&ValueError{"reflect.Value.Complex", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Elem returns the value that the interface v contains
// or that the pointer v points to.
// It panics if v's Kind is not Interface or Ptr.
// It returns the zero Value if v is nil.
func (v Value) Elem() Value {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Interface:
var eface interface{}
if v.typ.NumMethod() == 0 {
eface = *(*interface{})(v.ptr)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
} else {
eface = (interface{})(*(*interface {
M()
})(v.ptr))
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
x := unpackEface(eface)
if x.flag != 0 {
x.flag |= v.flag & flagRO
}
return x
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Ptr:
ptr := v.ptr
if v.flag&flagIndir != 0 {
ptr = *(*unsafe.Pointer)(ptr)
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// The returned value's address is v's value.
if ptr == nil {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
return Value{}
}
tt := (*ptrType)(unsafe.Pointer(v.typ))
typ := tt.elem
fl := v.flag&flagRO | flagIndir | flagAddr
fl |= flag(typ.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ, ptr, fl}
}
panic(&ValueError{"reflect.Value.Elem", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Field returns the i'th field of the struct v.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Struct or i is out of range.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Field(i int) Value {
if v.kind() != Struct {
panic(&ValueError{"reflect.Value.Field", v.kind()})
}
tt := (*structType)(unsafe.Pointer(v.typ))
if uint(i) >= uint(len(tt.fields)) {
panic("reflect: Field index out of range")
}
field := &tt.fields[i]
typ := field.typ
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// Inherit permission bits from v, but clear flagEmbedRO.
fl := v.flag&(flagStickyRO|flagIndir|flagAddr) | flag(typ.Kind())
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// Using an unexported field forces flagRO.
if !field.name.isExported() {
if field.name.name() == "" {
fl |= flagEmbedRO
} else {
fl |= flagStickyRO
}
}
// Either flagIndir is set and v.ptr points at struct,
// or flagIndir is not set and v.ptr is the actual struct data.
// In the former case, we want v.ptr + offset.
// In the latter case, we must have field.offset = 0,
// so v.ptr + field.offset is still okay.
ptr := unsafe.Pointer(uintptr(v.ptr) + field.offset)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ, ptr, fl}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// FieldByIndex returns the nested field corresponding to index.
// It panics if v's Kind is not struct.
func (v Value) FieldByIndex(index []int) Value {
if len(index) == 1 {
return v.Field(index[0])
}
v.mustBe(Struct)
for i, x := range index {
if i > 0 {
if v.Kind() == Ptr && v.typ.Elem().Kind() == Struct {
if v.IsNil() {
panic("reflect: indirection through nil pointer to embedded struct")
}
v = v.Elem()
}
}
v = v.Field(x)
}
return v
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// FieldByName returns the struct field with the given name.
// It returns the zero Value if no field was found.
// It panics if v's Kind is not struct.
func (v Value) FieldByName(name string) Value {
v.mustBe(Struct)
if f, ok := v.typ.FieldByName(name); ok {
return v.FieldByIndex(f.Index)
}
return Value{}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// FieldByNameFunc returns the struct field with a name
// that satisfies the match function.
// It panics if v's Kind is not struct.
// It returns the zero Value if no field was found.
func (v Value) FieldByNameFunc(match func(string) bool) Value {
if f, ok := v.typ.FieldByNameFunc(match); ok {
return v.FieldByIndex(f.Index)
}
return Value{}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Float returns v's underlying value, as a float64.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// It panics if v's Kind is not Float32 or Float64
func (v Value) Float() float64 {
k := v.kind()
switch k {
case Float32:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return float64(*(*float32)(v.ptr))
case Float64:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return *(*float64)(v.ptr)
}
panic(&ValueError{"reflect.Value.Float", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
var uint8Type = TypeOf(uint8(0)).(*rtype)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// Index returns v's i'th element.
// It panics if v's Kind is not Array, Slice, or String or i is out of range.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Index(i int) Value {
switch v.kind() {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Array:
tt := (*arrayType)(unsafe.Pointer(v.typ))
if uint(i) >= uint(tt.len) {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect: array index out of range")
}
typ := tt.elem
offset := uintptr(i) * typ.size
// Either flagIndir is set and v.ptr points at array,
// or flagIndir is not set and v.ptr is the actual array data.
// In the former case, we want v.ptr + offset.
// In the latter case, we must be doing Index(0), so offset = 0,
// so v.ptr + offset is still okay.
val := unsafe.Pointer(uintptr(v.ptr) + offset)
fl := v.flag&(flagRO|flagIndir|flagAddr) | flag(typ.Kind()) // bits same as overall array
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ, val, fl}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Slice:
// Element flag same as Elem of Ptr.
// Addressable, indirect, possibly read-only.
s := (*sliceHeader)(v.ptr)
if uint(i) >= uint(s.Len) {
panic("reflect: slice index out of range")
}
tt := (*sliceType)(unsafe.Pointer(v.typ))
typ := tt.elem
val := arrayAt(s.Data, i, typ.size)
fl := flagAddr | flagIndir | v.flag&flagRO | flag(typ.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ, val, fl}
case String:
s := (*stringHeader)(v.ptr)
if uint(i) >= uint(s.Len) {
panic("reflect: string index out of range")
}
p := arrayAt(s.Data, i, 1)
fl := v.flag&flagRO | flag(Uint8) | flagIndir
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{uint8Type, p, fl}
}
panic(&ValueError{"reflect.Value.Index", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Int returns v's underlying value, as an int64.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Int, Int8, Int16, Int32, or Int64.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Int() int64 {
k := v.kind()
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
p := v.ptr
switch k {
case Int:
return int64(*(*int)(p))
case Int8:
return int64(*(*int8)(p))
case Int16:
return int64(*(*int16)(p))
case Int32:
return int64(*(*int32)(p))
case Int64:
return int64(*(*int64)(p))
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
panic(&ValueError{"reflect.Value.Int", v.kind()})
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
// CanInterface reports whether Interface can be used without panicking.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
func (v Value) CanInterface() bool {
if v.flag == 0 {
panic(&ValueError{"reflect.Value.CanInterface", Invalid})
}
return v.flag&flagRO == 0
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Interface returns v's current value as an interface{}.
// It is equivalent to:
// var i interface{} = (v's underlying value)
// It panics if the Value was obtained by accessing
// unexported struct fields.
func (v Value) Interface() (i interface{}) {
return valueInterface(v, true)
}
func valueInterface(v Value, safe bool) interface{} {
if v.flag == 0 {
panic(&ValueError{"reflect.Value.Interface", 0})
}
if safe && v.flag&flagRO != 0 {
// Do not allow access to unexported values via Interface,
// because they might be pointers that should not be
// writable or methods or function that should not be callable.
panic("reflect.Value.Interface: cannot return value obtained from unexported field or method")
}
if v.flag&flagMethod != 0 {
v = makeMethodValue("Interface", v)
}
if v.kind() == Interface {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// Special case: return the element inside the interface.
// Empty interface has one layout, all interfaces with
// methods have a second layout.
if v.NumMethod() == 0 {
return *(*interface{})(v.ptr)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
return *(*interface {
M()
})(v.ptr)
}
// TODO: pass safe to packEface so we don't need to copy if safe==true?
return packEface(v)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// InterfaceData returns the interface v's value as a uintptr pair.
// It panics if v's Kind is not Interface.
func (v Value) InterfaceData() [2]uintptr {
// TODO: deprecate this
v.mustBe(Interface)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// We treat this as a read operation, so we allow
// it even for unexported data, because the caller
// has to import "unsafe" to turn it into something
// that can be abused.
// Interface value is always bigger than a word; assume flagIndir.
return *(*[2]uintptr)(v.ptr)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// IsNil reports whether its argument v is nil. The argument must be
// a chan, func, interface, map, pointer, or slice value; if it is
// not, IsNil panics. Note that IsNil is not always equivalent to a
// regular comparison with nil in Go. For example, if v was created
// by calling ValueOf with an uninitialized interface variable i,
// i==nil will be true but v.IsNil will panic as v will be the zero
// Value.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) IsNil() bool {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Chan, Func, Map, Ptr:
if v.flag&flagMethod != 0 {
return false
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
ptr := v.ptr
if v.flag&flagIndir != 0 {
ptr = *(*unsafe.Pointer)(ptr)
}
return ptr == nil
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Interface, Slice:
// Both interface and slice are nil if first word is 0.
// Both are always bigger than a word; assume flagIndir.
return *(*unsafe.Pointer)(v.ptr) == nil
}
panic(&ValueError{"reflect.Value.IsNil", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// IsValid reports whether v represents a value.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// It returns false if v is the zero Value.
// If IsValid returns false, all other methods except String panic.
// Most functions and methods never return an invalid value.
// If one does, its documentation states the conditions explicitly.
func (v Value) IsValid() bool {
return v.flag != 0
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Kind returns v's Kind.
// If v is the zero Value (IsValid returns false), Kind returns Invalid.
func (v Value) Kind() Kind {
return v.kind()
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Len returns v's length.
// It panics if v's Kind is not Array, Chan, Map, Slice, or String.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Len() int {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Array:
tt := (*arrayType)(unsafe.Pointer(v.typ))
return int(tt.len)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Chan:
return chanlen(v.pointer())
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Map:
return maplen(v.pointer())
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Slice:
// Slice is bigger than a word; assume flagIndir.
return (*sliceHeader)(v.ptr).Len
case String:
// String is bigger than a word; assume flagIndir.
return (*stringHeader)(v.ptr).Len
}
panic(&ValueError{"reflect.Value.Len", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// MapIndex returns the value associated with key in the map v.
// It panics if v's Kind is not Map.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It returns the zero Value if key is not found in the map or if v represents a nil map.
// As in Go, the key's value must be assignable to the map's key type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) MapIndex(key Value) Value {
v.mustBe(Map)
tt := (*mapType)(unsafe.Pointer(v.typ))
// Do not require key to be exported, so that DeepEqual
// and other programs can use all the keys returned by
// MapKeys as arguments to MapIndex. If either the map
// or the key is unexported, though, the result will be
// considered unexported. This is consistent with the
// behavior for structs, which allow read but not write
// of unexported fields.
key = key.assignTo("reflect.Value.MapIndex", tt.key, nil)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
var k unsafe.Pointer
if key.flag&flagIndir != 0 {
k = key.ptr
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
k = unsafe.Pointer(&key.ptr)
}
e := mapaccess(v.typ, v.pointer(), k)
if e == nil {
return Value{}
}
typ := tt.elem
fl := (v.flag | key.flag) & flagRO
fl |= flag(typ.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(typ) {
// Copy result so future changes to the map
// won't change the underlying value.
c := unsafe_New(typ)
typedmemmove(typ, c, e)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ, c, fl | flagIndir}
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ, *(*unsafe.Pointer)(e), fl}
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// MapKeys returns a slice containing all the keys present in the map,
// in unspecified order.
// It panics if v's Kind is not Map.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It returns an empty slice if v represents a nil map.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) MapKeys() []Value {
v.mustBe(Map)
tt := (*mapType)(unsafe.Pointer(v.typ))
keyType := tt.key
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
fl := v.flag&flagRO | flag(keyType.Kind())
m := v.pointer()
mlen := int(0)
if m != nil {
mlen = maplen(m)
}
it := mapiterinit(v.typ, m)
a := make([]Value, mlen)
var i int
for i = 0; i < len(a); i++ {
key := mapiterkey(it)
if key == nil {
// Someone deleted an entry from the map since we
// called maplen above. It's a data race, but nothing
// we can do about it.
break
}
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(keyType) {
// Copy result so future changes to the map
// won't change the underlying value.
c := unsafe_New(keyType)
typedmemmove(keyType, c, key)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
a[i] = Value{keyType, c, fl | flagIndir}
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
a[i] = Value{keyType, *(*unsafe.Pointer)(key), fl}
}
mapiternext(it)
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
return a[:i]
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Method returns a function value corresponding to v's i'th method.
// The arguments to a Call on the returned function should not include
// a receiver; the returned function will always use v as the receiver.
// Method panics if i is out of range or if v is a nil interface value.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Method(i int) Value {
if v.typ == nil {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic(&ValueError{"reflect.Value.Method", Invalid})
}
if v.flag&flagMethod != 0 || uint(i) >= uint(v.typ.NumMethod()) {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect: Method index out of range")
}
if v.typ.Kind() == Interface && v.IsNil() {
panic("reflect: Method on nil interface value")
}
fl := v.flag & (flagStickyRO | flagIndir) // Clear flagEmbedRO
fl |= flag(Func)
fl |= flag(i)<<flagMethodShift | flagMethod
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{v.typ, v.ptr, fl}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// NumMethod returns the number of methods in the value's method set.
func (v Value) NumMethod() int {
if v.typ == nil {
panic(&ValueError{"reflect.Value.NumMethod", Invalid})
}
if v.flag&flagMethod != 0 {
return 0
}
return v.typ.NumMethod()
}
// MethodByName returns a function value corresponding to the method
// of v with the given name.
// The arguments to a Call on the returned function should not include
// a receiver; the returned function will always use v as the receiver.
// It returns the zero Value if no method was found.
func (v Value) MethodByName(name string) Value {
if v.typ == nil {
panic(&ValueError{"reflect.Value.MethodByName", Invalid})
}
if v.flag&flagMethod != 0 {
return Value{}
}
m, ok := v.typ.MethodByName(name)
if !ok {
return Value{}
}
return v.Method(m.Index)
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// NumField returns the number of fields in the struct v.
// It panics if v's Kind is not Struct.
func (v Value) NumField() int {
v.mustBe(Struct)
tt := (*structType)(unsafe.Pointer(v.typ))
return len(tt.fields)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// OverflowComplex reports whether the complex128 x cannot be represented by v's type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// It panics if v's Kind is not Complex64 or Complex128.
func (v Value) OverflowComplex(x complex128) bool {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Complex64:
return overflowFloat32(real(x)) || overflowFloat32(imag(x))
case Complex128:
return false
}
panic(&ValueError{"reflect.Value.OverflowComplex", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// OverflowFloat reports whether the float64 x cannot be represented by v's type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// It panics if v's Kind is not Float32 or Float64.
func (v Value) OverflowFloat(x float64) bool {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Float32:
return overflowFloat32(x)
case Float64:
return false
}
panic(&ValueError{"reflect.Value.OverflowFloat", v.kind()})
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
func overflowFloat32(x float64) bool {
if x < 0 {
x = -x
}
return math.MaxFloat32 < x && x <= math.MaxFloat64
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// OverflowInt reports whether the int64 x cannot be represented by v's type.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Int, Int8, int16, Int32, or Int64.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) OverflowInt(x int64) bool {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Int, Int8, Int16, Int32, Int64:
bitSize := v.typ.size * 8
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
trunc := (x << (64 - bitSize)) >> (64 - bitSize)
return x != trunc
}
panic(&ValueError{"reflect.Value.OverflowInt", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// OverflowUint reports whether the uint64 x cannot be represented by v's type.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Uint, Uintptr, Uint8, Uint16, Uint32, or Uint64.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) OverflowUint(x uint64) bool {
k := v.kind()
switch k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Uint, Uintptr, Uint8, Uint16, Uint32, Uint64:
bitSize := v.typ.size * 8
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
trunc := (x << (64 - bitSize)) >> (64 - bitSize)
return x != trunc
}
panic(&ValueError{"reflect.Value.OverflowUint", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Pointer returns v's value as a uintptr.
// It returns uintptr instead of unsafe.Pointer so that
// code using reflect cannot obtain unsafe.Pointers
// without importing the unsafe package explicitly.
// It panics if v's Kind is not Chan, Func, Map, Ptr, Slice, or UnsafePointer.
//
// If v's Kind is Func, the returned pointer is an underlying
// code pointer, but not necessarily enough to identify a
// single function uniquely. The only guarantee is that the
// result is zero if and only if v is a nil func Value.
//
// If v's Kind is Slice, the returned pointer is to the first
// element of the slice. If the slice is nil the returned value
// is 0. If the slice is empty but non-nil the return value is non-zero.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Pointer() uintptr {
// TODO: deprecate
k := v.kind()
switch k {
case Chan, Map, Ptr, UnsafePointer:
return uintptr(v.pointer())
case Func:
if v.flag&flagMethod != 0 {
// As the doc comment says, the returned pointer is an
// underlying code pointer but not necessarily enough to
// identify a single function uniquely. All method expressions
// created via reflect have the same underlying code pointer,
// so their Pointers are equal. The function used here must
// match the one used in makeMethodValue.
f := methodValueCall
return **(**uintptr)(unsafe.Pointer(&f))
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
p := v.pointer()
// Non-nil func value points at data block.
// First word of data block is actual code.
if p != nil {
p = *(*unsafe.Pointer)(p)
}
return uintptr(p)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Slice:
return (*SliceHeader)(v.ptr).Data
}
panic(&ValueError{"reflect.Value.Pointer", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Recv receives and returns a value from the channel v.
// It panics if v's Kind is not Chan.
// The receive blocks until a value is ready.
// The boolean value ok is true if the value x corresponds to a send
// on the channel, false if it is a zero value received because the channel is closed.
func (v Value) Recv() (x Value, ok bool) {
v.mustBe(Chan)
v.mustBeExported()
return v.recv(false)
}
// internal recv, possibly non-blocking (nb).
// v is known to be a channel.
func (v Value) recv(nb bool) (val Value, ok bool) {
tt := (*chanType)(unsafe.Pointer(v.typ))
if ChanDir(tt.dir)&RecvDir == 0 {
panic("reflect: recv on send-only channel")
}
t := tt.elem
val = Value{t, nil, flag(t.Kind())}
var p unsafe.Pointer
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(t) {
p = unsafe_New(t)
val.ptr = p
val.flag |= flagIndir
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
p = unsafe.Pointer(&val.ptr)
}
selected, ok := chanrecv(v.typ, v.pointer(), nb, p)
if !selected {
val = Value{}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
return
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Send sends x on the channel v.
// It panics if v's kind is not Chan or if x's type is not the same type as v's element type.
// As in Go, x's value must be assignable to the channel's element type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Send(x Value) {
v.mustBe(Chan)
v.mustBeExported()
v.send(x, false)
}
// internal send, possibly non-blocking.
// v is known to be a channel.
func (v Value) send(x Value, nb bool) (selected bool) {
tt := (*chanType)(unsafe.Pointer(v.typ))
if ChanDir(tt.dir)&SendDir == 0 {
panic("reflect: send on recv-only channel")
}
x.mustBeExported()
x = x.assignTo("reflect.Value.Send", tt.elem, nil)
var p unsafe.Pointer
if x.flag&flagIndir != 0 {
p = x.ptr
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
p = unsafe.Pointer(&x.ptr)
}
return chansend(v.typ, v.pointer(), p, nb)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Set assigns x to the value v.
// It panics if CanSet returns false.
// As in Go, x's value must be assignable to v's type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Set(x Value) {
v.mustBeAssignable()
x.mustBeExported() // do not let unexported x leak
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
var target unsafe.Pointer
if v.kind() == Interface {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
target = v.ptr
}
x = x.assignTo("reflect.Set", v.typ, target)
if x.flag&flagIndir != 0 {
typedmemmove(v.typ, v.ptr, x.ptr)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*unsafe.Pointer)(v.ptr) = x.ptr
}
}
// SetBool sets v's underlying value.
// It panics if v's Kind is not Bool or if CanSet() is false.
func (v Value) SetBool(x bool) {
v.mustBeAssignable()
v.mustBe(Bool)
*(*bool)(v.ptr) = x
}
// SetBytes sets v's underlying value.
// It panics if v's underlying value is not a slice of bytes.
func (v Value) SetBytes(x []byte) {
v.mustBeAssignable()
v.mustBe(Slice)
if v.typ.Elem().Kind() != Uint8 {
panic("reflect.Value.SetBytes of non-byte slice")
}
*(*[]byte)(v.ptr) = x
}
// setRunes sets v's underlying value.
// It panics if v's underlying value is not a slice of runes (int32s).
func (v Value) setRunes(x []rune) {
v.mustBeAssignable()
v.mustBe(Slice)
if v.typ.Elem().Kind() != Int32 {
panic("reflect.Value.setRunes of non-rune slice")
}
*(*[]rune)(v.ptr) = x
}
// SetComplex sets v's underlying value to x.
// It panics if v's Kind is not Complex64 or Complex128, or if CanSet() is false.
func (v Value) SetComplex(x complex128) {
v.mustBeAssignable()
switch k := v.kind(); k {
default:
panic(&ValueError{"reflect.Value.SetComplex", v.kind()})
case Complex64:
*(*complex64)(v.ptr) = complex64(x)
case Complex128:
*(*complex128)(v.ptr) = x
}
}
// SetFloat sets v's underlying value to x.
// It panics if v's Kind is not Float32 or Float64, or if CanSet() is false.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) SetFloat(x float64) {
v.mustBeAssignable()
switch k := v.kind(); k {
default:
panic(&ValueError{"reflect.Value.SetFloat", v.kind()})
case Float32:
*(*float32)(v.ptr) = float32(x)
case Float64:
*(*float64)(v.ptr) = x
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// SetInt sets v's underlying value to x.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Int, Int8, Int16, Int32, or Int64, or if CanSet() is false.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) SetInt(x int64) {
v.mustBeAssignable()
switch k := v.kind(); k {
default:
panic(&ValueError{"reflect.Value.SetInt", v.kind()})
case Int:
*(*int)(v.ptr) = int(x)
case Int8:
*(*int8)(v.ptr) = int8(x)
case Int16:
*(*int16)(v.ptr) = int16(x)
case Int32:
*(*int32)(v.ptr) = int32(x)
case Int64:
*(*int64)(v.ptr) = x
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// SetLen sets v's length to n.
// It panics if v's Kind is not Slice or if n is negative or
// greater than the capacity of the slice.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) SetLen(n int) {
v.mustBeAssignable()
v.mustBe(Slice)
s := (*sliceHeader)(v.ptr)
if uint(n) > uint(s.Cap) {
panic("reflect: slice length out of range in SetLen")
}
s.Len = n
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// SetCap sets v's capacity to n.
// It panics if v's Kind is not Slice or if n is smaller than the length or
// greater than the capacity of the slice.
func (v Value) SetCap(n int) {
v.mustBeAssignable()
v.mustBe(Slice)
s := (*sliceHeader)(v.ptr)
if n < int(s.Len) || n > int(s.Cap) {
panic("reflect: slice capacity out of range in SetCap")
}
s.Cap = n
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// SetMapIndex sets the value associated with key in the map v to val.
// It panics if v's Kind is not Map.
// If val is the zero Value, SetMapIndex deletes the key from the map.
// Otherwise if v holds a nil map, SetMapIndex will panic.
// As in Go, key's value must be assignable to the map's key type,
// and val's value must be assignable to the map's value type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) SetMapIndex(key, val Value) {
v.mustBe(Map)
v.mustBeExported()
key.mustBeExported()
tt := (*mapType)(unsafe.Pointer(v.typ))
key = key.assignTo("reflect.Value.SetMapIndex", tt.key, nil)
var k unsafe.Pointer
if key.flag&flagIndir != 0 {
k = key.ptr
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
k = unsafe.Pointer(&key.ptr)
}
if val.typ == nil {
mapdelete(v.typ, v.pointer(), k)
return
}
val.mustBeExported()
val = val.assignTo("reflect.Value.SetMapIndex", tt.elem, nil)
var e unsafe.Pointer
if val.flag&flagIndir != 0 {
e = val.ptr
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
e = unsafe.Pointer(&val.ptr)
}
mapassign(v.typ, v.pointer(), k, e)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// SetUint sets v's underlying value to x.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Uint, Uintptr, Uint8, Uint16, Uint32, or Uint64, or if CanSet() is false.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) SetUint(x uint64) {
v.mustBeAssignable()
switch k := v.kind(); k {
default:
panic(&ValueError{"reflect.Value.SetUint", v.kind()})
case Uint:
*(*uint)(v.ptr) = uint(x)
case Uint8:
*(*uint8)(v.ptr) = uint8(x)
case Uint16:
*(*uint16)(v.ptr) = uint16(x)
case Uint32:
*(*uint32)(v.ptr) = uint32(x)
case Uint64:
*(*uint64)(v.ptr) = x
case Uintptr:
*(*uintptr)(v.ptr) = uintptr(x)
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// SetPointer sets the unsafe.Pointer value v to x.
// It panics if v's Kind is not UnsafePointer.
func (v Value) SetPointer(x unsafe.Pointer) {
v.mustBeAssignable()
v.mustBe(UnsafePointer)
*(*unsafe.Pointer)(v.ptr) = x
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// SetString sets v's underlying value to x.
// It panics if v's Kind is not String or if CanSet() is false.
func (v Value) SetString(x string) {
v.mustBeAssignable()
v.mustBe(String)
*(*string)(v.ptr) = x
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Slice returns v[i:j].
// It panics if v's Kind is not Array, Slice or String, or if v is an unaddressable array,
// or if the indexes are out of bounds.
func (v Value) Slice(i, j int) Value {
var (
cap int
typ *sliceType
base unsafe.Pointer
)
switch kind := v.kind(); kind {
default:
panic(&ValueError{"reflect.Value.Slice", v.kind()})
case Array:
if v.flag&flagAddr == 0 {
panic("reflect.Value.Slice: slice of unaddressable array")
}
tt := (*arrayType)(unsafe.Pointer(v.typ))
cap = int(tt.len)
typ = (*sliceType)(unsafe.Pointer(tt.slice))
base = v.ptr
case Slice:
typ = (*sliceType)(unsafe.Pointer(v.typ))
s := (*sliceHeader)(v.ptr)
base = unsafe.Pointer(s.Data)
cap = s.Cap
case String:
s := (*stringHeader)(v.ptr)
if i < 0 || j < i || j > s.Len {
panic("reflect.Value.Slice: string slice index out of bounds")
}
t := stringHeader{arrayAt(s.Data, i, 1), j - i}
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{v.typ, unsafe.Pointer(&t), v.flag}
}
if i < 0 || j < i || j > cap {
panic("reflect.Value.Slice: slice index out of bounds")
}
// Declare slice so that gc can see the base pointer in it.
var x []unsafe.Pointer
// Reinterpret as *sliceHeader to edit.
s := (*sliceHeader)(unsafe.Pointer(&x))
s.Len = j - i
s.Cap = cap - i
if cap-i > 0 {
s.Data = arrayAt(base, i, typ.elem.Size())
} else {
// do not advance pointer, to avoid pointing beyond end of slice
s.Data = base
}
fl := v.flag&flagRO | flagIndir | flag(Slice)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ.common(), unsafe.Pointer(&x), fl}
}
// Slice3 is the 3-index form of the slice operation: it returns v[i:j:k].
// It panics if v's Kind is not Array or Slice, or if v is an unaddressable array,
// or if the indexes are out of bounds.
func (v Value) Slice3(i, j, k int) Value {
var (
cap int
typ *sliceType
base unsafe.Pointer
)
switch kind := v.kind(); kind {
default:
panic(&ValueError{"reflect.Value.Slice3", v.kind()})
case Array:
if v.flag&flagAddr == 0 {
panic("reflect.Value.Slice3: slice of unaddressable array")
}
tt := (*arrayType)(unsafe.Pointer(v.typ))
cap = int(tt.len)
typ = (*sliceType)(unsafe.Pointer(tt.slice))
base = v.ptr
case Slice:
typ = (*sliceType)(unsafe.Pointer(v.typ))
s := (*sliceHeader)(v.ptr)
base = s.Data
cap = s.Cap
}
if i < 0 || j < i || k < j || k > cap {
panic("reflect.Value.Slice3: slice index out of bounds")
}
// Declare slice so that the garbage collector
// can see the base pointer in it.
var x []unsafe.Pointer
// Reinterpret as *sliceHeader to edit.
s := (*sliceHeader)(unsafe.Pointer(&x))
s.Len = j - i
s.Cap = k - i
if k-i > 0 {
s.Data = arrayAt(base, i, typ.elem.Size())
} else {
// do not advance pointer, to avoid pointing beyond end of slice
s.Data = base
}
fl := v.flag&flagRO | flagIndir | flag(Slice)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ.common(), unsafe.Pointer(&x), fl}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// String returns the string v's underlying value, as a string.
// String is a special case because of Go's String method convention.
// Unlike the other getters, it does not panic if v's Kind is not String.
// Instead, it returns a string of the form "<T value>" where T is v's type.
// The fmt package treats Values specially. It does not call their String
// method implicitly but instead prints the concrete values they hold.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) String() string {
switch k := v.kind(); k {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Invalid:
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
return "<invalid Value>"
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case String:
return *(*string)(v.ptr)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// If you call String on a reflect.Value of other type, it's better to
// print something than to panic. Useful in debugging.
return "<" + v.Type().String() + " Value>"
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// TryRecv attempts to receive a value from the channel v but will not block.
// It panics if v's Kind is not Chan.
// If the receive delivers a value, x is the transferred value and ok is true.
// If the receive cannot finish without blocking, x is the zero Value and ok is false.
// If the channel is closed, x is the zero value for the channel's element type and ok is false.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) TryRecv() (x Value, ok bool) {
v.mustBe(Chan)
v.mustBeExported()
return v.recv(true)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// TrySend attempts to send x on the channel v but will not block.
// It panics if v's Kind is not Chan.
// It reports whether the value was sent.
// As in Go, x's value must be assignable to the channel's element type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) TrySend(x Value) bool {
v.mustBe(Chan)
v.mustBeExported()
return v.send(x, true)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Type returns v's type.
func (v Value) Type() Type {
f := v.flag
if f == 0 {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic(&ValueError{"reflect.Value.Type", Invalid})
}
if f&flagMethod == 0 {
// Easy case
return v.typ
}
// Method value.
// v.typ describes the receiver, not the method type.
i := int(v.flag) >> flagMethodShift
if v.typ.Kind() == Interface {
// Method on interface.
tt := (*interfaceType)(unsafe.Pointer(v.typ))
if uint(i) >= uint(len(tt.methods)) {
panic("reflect: internal error: invalid method index")
}
m := &tt.methods[i]
return m.typ
}
// Method on concrete type.
ut := v.typ.uncommon()
cmd/compile, etc: store method tables as offsets This CL introduces the typeOff type and a lookup method of the same name that can turn a typeOff offset into an *rtype. In a typical Go binary (built with buildmode=exe, pie, c-archive, or c-shared), there is one moduledata and all typeOff values are offsets relative to firstmoduledata.types. This makes computing the pointer cheap in typical programs. With buildmode=shared (and one day, buildmode=plugin) there are multiple modules whose relative offset is determined at runtime. We identify a type in the general case by the pair of the original *rtype that references it and its typeOff value. We determine the module from the original pointer, and then use the typeOff from there to compute the final *rtype. To ensure there is only one *rtype representing each type, the runtime initializes a typemap for each module, using any identical type from an earlier module when resolving that offset. This means that types computed from an offset match the type mapped by the pointer dynamic relocations. A series of followup CLs will replace other *rtype values with typeOff (and name/*string with nameOff). For types created at runtime by reflect, type offsets are treated as global IDs and reference into a reflect offset map kept by the runtime. darwin/amd64: cmd/go: -57KB (0.6%) jujud: -557KB (0.8%) linux/amd64 PIE: cmd/go: -361KB (3.0%) jujud: -3.5MB (4.2%) For #6853. Change-Id: Icf096fd884a0a0cb9f280f46f7a26c70a9006c96 Reviewed-on: https://go-review.googlesource.com/21285 Reviewed-by: Ian Lance Taylor <iant@golang.org> Run-TryBot: David Crawshaw <crawshaw@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-03-28 10:32:27 -04:00
if ut == nil || uint(i) >= uint(ut.mcount) {
panic("reflect: internal error: invalid method index")
}
cmd/compile, etc: store method tables as offsets This CL introduces the typeOff type and a lookup method of the same name that can turn a typeOff offset into an *rtype. In a typical Go binary (built with buildmode=exe, pie, c-archive, or c-shared), there is one moduledata and all typeOff values are offsets relative to firstmoduledata.types. This makes computing the pointer cheap in typical programs. With buildmode=shared (and one day, buildmode=plugin) there are multiple modules whose relative offset is determined at runtime. We identify a type in the general case by the pair of the original *rtype that references it and its typeOff value. We determine the module from the original pointer, and then use the typeOff from there to compute the final *rtype. To ensure there is only one *rtype representing each type, the runtime initializes a typemap for each module, using any identical type from an earlier module when resolving that offset. This means that types computed from an offset match the type mapped by the pointer dynamic relocations. A series of followup CLs will replace other *rtype values with typeOff (and name/*string with nameOff). For types created at runtime by reflect, type offsets are treated as global IDs and reference into a reflect offset map kept by the runtime. darwin/amd64: cmd/go: -57KB (0.6%) jujud: -557KB (0.8%) linux/amd64 PIE: cmd/go: -361KB (3.0%) jujud: -3.5MB (4.2%) For #6853. Change-Id: Icf096fd884a0a0cb9f280f46f7a26c70a9006c96 Reviewed-on: https://go-review.googlesource.com/21285 Reviewed-by: Ian Lance Taylor <iant@golang.org> Run-TryBot: David Crawshaw <crawshaw@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-03-28 10:32:27 -04:00
m := ut.methods()[i]
return v.typ.typeOff(m.mtyp)
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// Uint returns v's underlying value, as a uint64.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v's Kind is not Uint, Uintptr, Uint8, Uint16, Uint32, or Uint64.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) Uint() uint64 {
k := v.kind()
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
p := v.ptr
switch k {
case Uint:
return uint64(*(*uint)(p))
case Uint8:
return uint64(*(*uint8)(p))
case Uint16:
return uint64(*(*uint16)(p))
case Uint32:
return uint64(*(*uint32)(p))
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
case Uint64:
return uint64(*(*uint64)(p))
case Uintptr:
return uint64(*(*uintptr)(p))
}
panic(&ValueError{"reflect.Value.Uint", v.kind()})
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
// UnsafeAddr returns a pointer to v's data.
// It is for advanced clients that also import the "unsafe" package.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// It panics if v is not addressable.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func (v Value) UnsafeAddr() uintptr {
// TODO: deprecate
if v.typ == nil {
panic(&ValueError{"reflect.Value.UnsafeAddr", Invalid})
}
if v.flag&flagAddr == 0 {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect.Value.UnsafeAddr of unaddressable value")
}
return uintptr(v.ptr)
}
// StringHeader is the runtime representation of a string.
// It cannot be used safely or portably and its representation may
// change in a later release.
// Moreover, the Data field is not sufficient to guarantee the data
// it references will not be garbage collected, so programs must keep
// a separate, correctly typed pointer to the underlying data.
type StringHeader struct {
Data uintptr
Len int
}
// stringHeader is a safe version of StringHeader used within this package.
type stringHeader struct {
Data unsafe.Pointer
Len int
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// SliceHeader is the runtime representation of a slice.
// It cannot be used safely or portably and its representation may
// change in a later release.
// Moreover, the Data field is not sufficient to guarantee the data
// it references will not be garbage collected, so programs must keep
// a separate, correctly typed pointer to the underlying data.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
type SliceHeader struct {
Data uintptr
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
Len int
Cap int
}
// sliceHeader is a safe version of SliceHeader used within this package.
type sliceHeader struct {
Data unsafe.Pointer
Len int
Cap int
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
func typesMustMatch(what string, t1, t2 Type) {
if t1 != t2 {
panic(what + ": " + t1.String() + " != " + t2.String())
}
}
// arrayAt returns the i-th element of p, a C-array whose elements are
// eltSize wide (in bytes).
func arrayAt(p unsafe.Pointer, i int, eltSize uintptr) unsafe.Pointer {
return unsafe.Pointer(uintptr(p) + uintptr(i)*eltSize)
}
// grow grows the slice s so that it can hold extra more values, allocating
// more capacity if needed. It also returns the old and new slice lengths.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func grow(s Value, extra int) (Value, int, int) {
i0 := s.Len()
i1 := i0 + extra
if i1 < i0 {
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
panic("reflect.Append: slice overflow")
}
m := s.Cap()
if i1 <= m {
return s.Slice(0, i1), i0, i1
}
if m == 0 {
m = extra
} else {
for m < i1 {
if i0 < 1024 {
m += m
} else {
m += m / 4
}
}
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
t := MakeSlice(s.Type(), i1, m)
Copy(t, s)
return t, i0, i1
}
// Append appends the values x to a slice s and returns the resulting slice.
// As in Go, each x's value must be assignable to the slice's element type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func Append(s Value, x ...Value) Value {
s.mustBe(Slice)
s, i0, i1 := grow(s, len(x))
for i, j := i0, 0; i < i1; i, j = i+1, j+1 {
s.Index(i).Set(x[j])
}
return s
}
// AppendSlice appends a slice t to a slice s and returns the resulting slice.
// The slices s and t must have the same element type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func AppendSlice(s, t Value) Value {
s.mustBe(Slice)
t.mustBe(Slice)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
typesMustMatch("reflect.AppendSlice", s.Type().Elem(), t.Type().Elem())
s, i0, i1 := grow(s, t.Len())
Copy(s.Slice(i0, i1), t)
return s
}
// Copy copies the contents of src into dst until either
// dst has been filled or src has been exhausted.
// It returns the number of elements copied.
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// Dst and src each must have kind Slice or Array, and
// dst and src must have the same element type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func Copy(dst, src Value) int {
dk := dst.kind()
if dk != Array && dk != Slice {
panic(&ValueError{"reflect.Copy", dk})
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
if dk == Array {
dst.mustBeAssignable()
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
dst.mustBeExported()
sk := src.kind()
if sk != Array && sk != Slice {
panic(&ValueError{"reflect.Copy", sk})
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
src.mustBeExported()
de := dst.typ.Elem()
se := src.typ.Elem()
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
typesMustMatch("reflect.Copy", de, se)
var ds, ss sliceHeader
if dk == Array {
ds.Data = dst.ptr
ds.Len = dst.Len()
ds.Cap = ds.Len
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
} else {
ds = *(*sliceHeader)(dst.ptr)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
if sk == Array {
ss.Data = src.ptr
ss.Len = src.Len()
ss.Cap = ss.Len
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
} else {
ss = *(*sliceHeader)(src.ptr)
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
return typedslicecopy(de.common(), ds, ss)
}
// A runtimeSelect is a single case passed to rselect.
// This must match ../runtime/select.go:/runtimeSelect
type runtimeSelect struct {
dir SelectDir // SelectSend, SelectRecv or SelectDefault
typ *rtype // channel type
ch unsafe.Pointer // channel
val unsafe.Pointer // ptr to data (SendDir) or ptr to receive buffer (RecvDir)
}
// rselect runs a select. It returns the index of the chosen case.
// If the case was a receive, val is filled in with the received value.
// The conventional OK bool indicates whether the receive corresponds
// to a sent value.
//go:noescape
func rselect([]runtimeSelect) (chosen int, recvOK bool)
// A SelectDir describes the communication direction of a select case.
type SelectDir int
// NOTE: These values must match ../runtime/select.go:/selectDir.
const (
_ SelectDir = iota
SelectSend // case Chan <- Send
SelectRecv // case <-Chan:
SelectDefault // default
)
// A SelectCase describes a single case in a select operation.
// The kind of case depends on Dir, the communication direction.
//
// If Dir is SelectDefault, the case represents a default case.
// Chan and Send must be zero Values.
//
// If Dir is SelectSend, the case represents a send operation.
// Normally Chan's underlying value must be a channel, and Send's underlying value must be
// assignable to the channel's element type. As a special case, if Chan is a zero Value,
// then the case is ignored, and the field Send will also be ignored and may be either zero
// or non-zero.
//
// If Dir is SelectRecv, the case represents a receive operation.
// Normally Chan's underlying value must be a channel and Send must be a zero Value.
// If Chan is a zero Value, then the case is ignored, but Send must still be a zero Value.
// When a receive operation is selected, the received Value is returned by Select.
//
type SelectCase struct {
Dir SelectDir // direction of case
Chan Value // channel to use (for send or receive)
Send Value // value to send (for send)
}
// Select executes a select operation described by the list of cases.
// Like the Go select statement, it blocks until at least one of the cases
// can proceed, makes a uniform pseudo-random choice,
// and then executes that case. It returns the index of the chosen case
// and, if that case was a receive operation, the value received and a
// boolean indicating whether the value corresponds to a send on the channel
// (as opposed to a zero value received because the channel is closed).
func Select(cases []SelectCase) (chosen int, recv Value, recvOK bool) {
// NOTE: Do not trust that caller is not modifying cases data underfoot.
// The range is safe because the caller cannot modify our copy of the len
// and each iteration makes its own copy of the value c.
runcases := make([]runtimeSelect, len(cases))
haveDefault := false
for i, c := range cases {
rc := &runcases[i]
rc.dir = c.Dir
switch c.Dir {
default:
panic("reflect.Select: invalid Dir")
case SelectDefault: // default
if haveDefault {
panic("reflect.Select: multiple default cases")
}
haveDefault = true
if c.Chan.IsValid() {
panic("reflect.Select: default case has Chan value")
}
if c.Send.IsValid() {
panic("reflect.Select: default case has Send value")
}
case SelectSend:
ch := c.Chan
if !ch.IsValid() {
break
}
ch.mustBe(Chan)
ch.mustBeExported()
tt := (*chanType)(unsafe.Pointer(ch.typ))
if ChanDir(tt.dir)&SendDir == 0 {
panic("reflect.Select: SendDir case using recv-only channel")
}
rc.ch = ch.pointer()
rc.typ = &tt.rtype
v := c.Send
if !v.IsValid() {
panic("reflect.Select: SendDir case missing Send value")
}
v.mustBeExported()
v = v.assignTo("reflect.Select", tt.elem, nil)
if v.flag&flagIndir != 0 {
rc.val = v.ptr
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
rc.val = unsafe.Pointer(&v.ptr)
}
case SelectRecv:
if c.Send.IsValid() {
panic("reflect.Select: RecvDir case has Send value")
}
ch := c.Chan
if !ch.IsValid() {
break
}
ch.mustBe(Chan)
ch.mustBeExported()
tt := (*chanType)(unsafe.Pointer(ch.typ))
if ChanDir(tt.dir)&RecvDir == 0 {
panic("reflect.Select: RecvDir case using send-only channel")
}
rc.ch = ch.pointer()
rc.typ = &tt.rtype
rc.val = unsafe_New(tt.elem)
}
}
chosen, recvOK = rselect(runcases)
if runcases[chosen].dir == SelectRecv {
tt := (*chanType)(unsafe.Pointer(runcases[chosen].typ))
t := tt.elem
p := runcases[chosen].val
fl := flag(t.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(t) {
recv = Value{t, p, fl | flagIndir}
} else {
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
recv = Value{t, *(*unsafe.Pointer)(p), fl}
}
}
return chosen, recv, recvOK
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
/*
* constructors
*/
// implemented in package runtime
func unsafe_New(*rtype) unsafe.Pointer
func unsafe_NewArray(*rtype, int) unsafe.Pointer
// MakeSlice creates a new zero-initialized slice value
// for the specified slice type, length, and capacity.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func MakeSlice(typ Type, len, cap int) Value {
if typ.Kind() != Slice {
panic("reflect.MakeSlice of non-slice type")
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
if len < 0 {
panic("reflect.MakeSlice: negative len")
}
if cap < 0 {
panic("reflect.MakeSlice: negative cap")
}
if len > cap {
panic("reflect.MakeSlice: len > cap")
}
s := sliceHeader{unsafe_NewArray(typ.Elem().(*rtype), cap), len, cap}
return Value{typ.common(), unsafe.Pointer(&s), flagIndir | flag(Slice)}
}
// MakeChan creates a new channel with the specified type and buffer size.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func MakeChan(typ Type, buffer int) Value {
if typ.Kind() != Chan {
panic("reflect.MakeChan of non-chan type")
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
if buffer < 0 {
panic("reflect.MakeChan: negative buffer size")
}
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
if typ.ChanDir() != BothDir {
panic("reflect.MakeChan: unidirectional channel type")
}
ch := makechan(typ.(*rtype), uint64(buffer))
return Value{typ.common(), ch, flag(Chan)}
}
// MakeMap creates a new map of the specified type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func MakeMap(typ Type) Value {
if typ.Kind() != Map {
panic("reflect.MakeMap of non-map type")
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
}
m := makemap(typ.(*rtype))
return Value{typ.common(), m, flag(Map)}
}
// Indirect returns the value that v points to.
// If v is a nil pointer, Indirect returns a zero Value.
// If v is not a pointer, Indirect returns v.
func Indirect(v Value) Value {
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
if v.Kind() != Ptr {
return v
}
return v.Elem()
}
// ValueOf returns a new Value initialized to the concrete value
// stored in the interface i. ValueOf(nil) returns the zero Value.
func ValueOf(i interface{}) Value {
if i == nil {
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
return Value{}
}
// TODO: Maybe allow contents of a Value to live on the stack.
// For now we make the contents always escape to the heap. It
// makes life easier in a few places (see chanrecv/mapassign
// comment below).
escapes(i)
return unpackEface(i)
}
// Zero returns a Value representing the zero value for the specified type.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
// The result is different from the zero value of the Value struct,
// which represents no value at all.
// For example, Zero(TypeOf(42)) returns a Value with Kind Int and value 0.
// The returned value is neither addressable nor settable.
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
func Zero(typ Type) Value {
if typ == nil {
reflect: new Type and Value definitions Type is now an interface that implements all the possible type methods. Instead of a type switch on a reflect.Type t, switch on t.Kind(). If a method is invoked on the wrong kind of type (for example, calling t.Field(0) when t.Kind() != Struct), the call panics. There is one method renaming: t.(*ChanType).Dir() is now t.ChanDir(). Value is now a struct value that implements all the possible value methods. Instead of a type switch on a reflect.Value v, switch on v.Kind(). If a method is invoked on the wrong kind of value (for example, calling t.Recv() when t.Kind() != Chan), the call panics. Since Value is now a struct, not an interface, its zero value cannot be compared to nil. Instead of v != nil, use v.IsValid(). Instead of other uses of nil as a Value, use Value{}, the zero value. Many methods have been renamed, most due to signature conflicts: OLD NEW v.(*ArrayValue).Elem v.Index v.(*BoolValue).Get v.Bool v.(*BoolValue).Set v.SetBool v.(*ChanType).Dir v.ChanDir v.(*ChanValue).Get v.Pointer v.(*ComplexValue).Get v.Complex v.(*ComplexValue).Overflow v.OverflowComplex v.(*ComplexValue).Set v.SetComplex v.(*FloatValue).Get v.Float v.(*FloatValue).Overflow v.OverflowFloat v.(*FloatValue).Set v.SetFloat v.(*FuncValue).Get v.Pointer v.(*InterfaceValue).Get v.InterfaceData v.(*IntValue).Get v.Int v.(*IntValue).Overflow v.OverflowInt v.(*IntValue).Set v.SetInt v.(*MapValue).Elem v.MapIndex v.(*MapValue).Get v.Pointer v.(*MapValue).Keys v.MapKeys v.(*MapValue).SetElem v.SetMapIndex v.(*PtrValue).Get v.Pointer v.(*SliceValue).Elem v.Index v.(*SliceValue).Get v.Pointer v.(*StringValue).Get v.String v.(*StringValue).Set v.SetString v.(*UintValue).Get v.Uint v.(*UintValue).Overflow v.OverflowUint v.(*UintValue).Set v.SetUint v.(*UnsafePointerValue).Get v.Pointer v.(*UnsafePointerValue).Set v.SetPointer Part of the motivation for this change is to enable a more efficient implementation of Value, one that does not allocate memory during most operations. To reduce the size of the CL, this CL's implementation is a wrapper around the old API. Later CLs will make the implementation more efficient without changing the API. Other CLs to be submitted at the same time as this one add support for this change to gofix (4343047) and update the Go source tree (4353043). R=gri, iant, niemeyer, r, rog, gustavo, r2 CC=golang-dev https://golang.org/cl/4281055
2011-04-08 12:26:51 -04:00
panic("reflect: Zero(nil)")
}
t := typ.common()
fl := flag(t.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
if ifaceIndir(t) {
return Value{t, unsafe_New(typ.(*rtype)), fl | flagIndir}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{t, nil, fl}
}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
// New returns a Value representing a pointer to a new zero value
// for the specified type. That is, the returned Value's Type is PtrTo(typ).
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
func New(typ Type) Value {
if typ == nil {
panic("reflect: New(nil)")
}
ptr := unsafe_New(typ.(*rtype))
fl := flag(Ptr)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ.common().ptrTo(), ptr, fl}
reflect: more efficient; cannot Set result of NewValue anymore * Reduces malloc counts during gob encoder/decoder test from 6/6 to 3/5. The current reflect uses Set to mean two subtly different things. (1) If you have a reflect.Value v, it might just represent itself (as in v = reflect.NewValue(42)), in which case calling v.Set only changed v, not any other data in the program. (2) If you have a reflect Value v derived from a pointer or a slice (as in x := []int{42}; v = reflect.NewValue(x).Index(0)), v represents the value held there. Changing x[0] affects the value returned by v.Int(), and calling v.Set affects x[0]. This was not really by design; it just happened that way. The motivation for the new reflect implementation was to remove mallocs. The use case (1) has an implicit malloc inside it. If you can do: v := reflect.NewValue(0) v.Set(42) i := v.Int() // i = 42 then that implies that v is referring to some underlying chunk of memory in order to remember the 42; that is, NewValue must have allocated some memory. Almost all the time you are using reflect the goal is to inspect or to change other data, not to manipulate data stored solely inside a reflect.Value. This CL removes use case (1), so that an assignable reflect.Value must always refer to some other piece of data in the program. Put another way, removing this case would make v := reflect.NewValue(0) v.Set(42) as illegal as 0 = 42. It would also make this illegal: x := 0 v := reflect.NewValue(x) v.Set(42) for the same reason. (Note that right now, v.Set(42) "succeeds" but does not change the value of x.) If you really wanted to make v refer to x, you'd start with &x and dereference it: x := 0 v := reflect.NewValue(&x).Elem() // v = *&x v.Set(42) It's pretty rare, except in tests, to want to use NewValue and then call Set to change the Value itself instead of some other piece of data in the program. I haven't seen it happen once yet while making the tree build with this change. For the same reasons, reflect.Zero (formerly reflect.MakeZero) would also return an unassignable, unaddressable value. This invalidates the (awkward) idiom: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.PointTo(v) which, when the API changed, turned into: pv := ... some Ptr Value we have ... v := reflect.Zero(pv.Type().Elem()) pv.Set(v.Addr()) In both, it is far from clear what the code is trying to do. Now that it is possible, this CL adds reflect.New(Type) Value that does the obvious thing (same as Go's new), so this code would be replaced by: pv := ... some Ptr Value we have ... pv.Set(reflect.New(pv.Type().Elem())) The changes just described can be confusing to think about, but I believe it is because the old API was confusing - it was conflating two different kinds of Values - and that the new API by itself is pretty simple: you can only Set (or call Addr on) a Value if it actually addresses some real piece of data; that is, only if it is the result of dereferencing a Ptr or indexing a Slice. If you really want the old behavior, you'd get it by translating: v := reflect.NewValue(x) into v := reflect.New(reflect.Typeof(x)).Elem() v.Set(reflect.NewValue(x)) Gofix will not be able to help with this, because whether and how to change the code depends on whether the original code meant use (1) or use (2), so the developer has to read and think about the code. You can see the effect on packages in the tree in https://golang.org/cl/4423043/. R=r CC=golang-dev https://golang.org/cl/4435042
2011-04-18 14:35:33 -04:00
}
// NewAt returns a Value representing a pointer to a value of the
// specified type, using p as that pointer.
func NewAt(typ Type, p unsafe.Pointer) Value {
fl := flag(Ptr)
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{typ.common().ptrTo(), p, fl}
}
// assignTo returns a value v that can be assigned directly to typ.
// It panics if v is not assignable to typ.
// For a conversion to an interface type, target is a suggested scratch space to use.
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
func (v Value) assignTo(context string, dst *rtype, target unsafe.Pointer) Value {
if v.flag&flagMethod != 0 {
v = makeMethodValue(context, v)
}
switch {
case directlyAssignable(dst, v.typ):
// Overwrite type so that they match.
// Same memory layout, so no harm done.
v.typ = dst
fl := v.flag & (flagRO | flagAddr | flagIndir)
fl |= flag(dst.Kind())
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{dst, v.ptr, fl}
case implements(dst, v.typ):
if target == nil {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
target = unsafe_New(dst)
}
x := valueInterface(v, false)
if dst.NumMethod() == 0 {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
*(*interface{})(target) = x
} else {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
ifaceE2I(dst, x, target)
}
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
return Value{dst, target, flagIndir | flag(Interface)}
}
// Failed.
panic(context + ": value of type " + v.typ.String() + " is not assignable to type " + dst.String())
}
// Convert returns the value v converted to type t.
// If the usual Go conversion rules do not allow conversion
// of the value v to type t, Convert panics.
func (v Value) Convert(t Type) Value {
if v.flag&flagMethod != 0 {
v = makeMethodValue("Convert", v)
}
op := convertOp(t.common(), v.typ)
if op == nil {
panic("reflect.Value.Convert: value of type " + v.typ.String() + " cannot be converted to type " + t.String())
}
return op(v, t)
}
// convertOp returns the function to convert a value of type src
// to a value of type dst. If the conversion is illegal, convertOp returns nil.
func convertOp(dst, src *rtype) func(Value, Type) Value {
switch src.Kind() {
case Int, Int8, Int16, Int32, Int64:
switch dst.Kind() {
case Int, Int8, Int16, Int32, Int64, Uint, Uint8, Uint16, Uint32, Uint64, Uintptr:
return cvtInt
case Float32, Float64:
return cvtIntFloat
case String:
return cvtIntString
}
case Uint, Uint8, Uint16, Uint32, Uint64, Uintptr:
switch dst.Kind() {
case Int, Int8, Int16, Int32, Int64, Uint, Uint8, Uint16, Uint32, Uint64, Uintptr:
return cvtUint
case Float32, Float64:
return cvtUintFloat
case String:
return cvtUintString
}
case Float32, Float64:
switch dst.Kind() {
case Int, Int8, Int16, Int32, Int64:
return cvtFloatInt
case Uint, Uint8, Uint16, Uint32, Uint64, Uintptr:
return cvtFloatUint
case Float32, Float64:
return cvtFloat
}
case Complex64, Complex128:
switch dst.Kind() {
case Complex64, Complex128:
return cvtComplex
}
case String:
if dst.Kind() == Slice && dst.Elem().PkgPath() == "" {
switch dst.Elem().Kind() {
case Uint8:
return cvtStringBytes
case Int32:
return cvtStringRunes
}
}
case Slice:
if dst.Kind() == String && src.Elem().PkgPath() == "" {
switch src.Elem().Kind() {
case Uint8:
return cvtBytesString
case Int32:
return cvtRunesString
}
}
}
// dst and src have same underlying type.
if haveIdenticalUnderlyingType(dst, src) {
return cvtDirect
}
// dst and src are unnamed pointer types with same underlying base type.
if dst.Kind() == Ptr && dst.Name() == "" &&
src.Kind() == Ptr && src.Name() == "" &&
haveIdenticalUnderlyingType(dst.Elem().common(), src.Elem().common()) {
return cvtDirect
}
if implements(dst, src) {
if src.Kind() == Interface {
return cvtI2I
}
return cvtT2I
}
return nil
}
// makeInt returns a Value of type t equal to bits (possibly truncated),
// where t is a signed or unsigned int type.
func makeInt(f flag, bits uint64, t Type) Value {
typ := t.common()
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
ptr := unsafe_New(typ)
switch typ.size {
case 1:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*uint8)(unsafe.Pointer(ptr)) = uint8(bits)
case 2:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*uint16)(unsafe.Pointer(ptr)) = uint16(bits)
case 4:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*uint32)(unsafe.Pointer(ptr)) = uint32(bits)
case 8:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*uint64)(unsafe.Pointer(ptr)) = bits
}
return Value{typ, ptr, f | flagIndir | flag(typ.Kind())}
}
// makeFloat returns a Value of type t equal to v (possibly truncated to float32),
// where t is a float32 or float64 type.
func makeFloat(f flag, v float64, t Type) Value {
typ := t.common()
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
ptr := unsafe_New(typ)
switch typ.size {
case 4:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*float32)(unsafe.Pointer(ptr)) = float32(v)
case 8:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*float64)(unsafe.Pointer(ptr)) = v
}
return Value{typ, ptr, f | flagIndir | flag(typ.Kind())}
}
// makeComplex returns a Value of type t equal to v (possibly truncated to complex64),
// where t is a complex64 or complex128 type.
func makeComplex(f flag, v complex128, t Type) Value {
typ := t.common()
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
ptr := unsafe_New(typ)
switch typ.size {
case 8:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*complex64)(unsafe.Pointer(ptr)) = complex64(v)
case 16:
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
*(*complex128)(unsafe.Pointer(ptr)) = v
}
return Value{typ, ptr, f | flagIndir | flag(typ.Kind())}
}
func makeString(f flag, v string, t Type) Value {
ret := New(t).Elem()
ret.SetString(v)
ret.flag = ret.flag&^flagAddr | f
return ret
}
func makeBytes(f flag, v []byte, t Type) Value {
ret := New(t).Elem()
ret.SetBytes(v)
ret.flag = ret.flag&^flagAddr | f
return ret
}
func makeRunes(f flag, v []rune, t Type) Value {
ret := New(t).Elem()
ret.setRunes(v)
ret.flag = ret.flag&^flagAddr | f
return ret
}
// These conversion functions are returned by convertOp
// for classes of conversions. For example, the first function, cvtInt,
// takes any value v of signed int type and returns the value converted
// to type t, where t is any signed or unsigned int type.
// convertOp: intXX -> [u]intXX
func cvtInt(v Value, t Type) Value {
return makeInt(v.flag&flagRO, uint64(v.Int()), t)
}
// convertOp: uintXX -> [u]intXX
func cvtUint(v Value, t Type) Value {
return makeInt(v.flag&flagRO, v.Uint(), t)
}
// convertOp: floatXX -> intXX
func cvtFloatInt(v Value, t Type) Value {
return makeInt(v.flag&flagRO, uint64(int64(v.Float())), t)
}
// convertOp: floatXX -> uintXX
func cvtFloatUint(v Value, t Type) Value {
return makeInt(v.flag&flagRO, uint64(v.Float()), t)
}
// convertOp: intXX -> floatXX
func cvtIntFloat(v Value, t Type) Value {
return makeFloat(v.flag&flagRO, float64(v.Int()), t)
}
// convertOp: uintXX -> floatXX
func cvtUintFloat(v Value, t Type) Value {
return makeFloat(v.flag&flagRO, float64(v.Uint()), t)
}
// convertOp: floatXX -> floatXX
func cvtFloat(v Value, t Type) Value {
return makeFloat(v.flag&flagRO, v.Float(), t)
}
// convertOp: complexXX -> complexXX
func cvtComplex(v Value, t Type) Value {
return makeComplex(v.flag&flagRO, v.Complex(), t)
}
// convertOp: intXX -> string
func cvtIntString(v Value, t Type) Value {
return makeString(v.flag&flagRO, string(v.Int()), t)
}
// convertOp: uintXX -> string
func cvtUintString(v Value, t Type) Value {
return makeString(v.flag&flagRO, string(v.Uint()), t)
}
// convertOp: []byte -> string
func cvtBytesString(v Value, t Type) Value {
return makeString(v.flag&flagRO, string(v.Bytes()), t)
}
// convertOp: string -> []byte
func cvtStringBytes(v Value, t Type) Value {
return makeBytes(v.flag&flagRO, []byte(v.String()), t)
}
// convertOp: []rune -> string
func cvtRunesString(v Value, t Type) Value {
return makeString(v.flag&flagRO, string(v.runes()), t)
}
// convertOp: string -> []rune
func cvtStringRunes(v Value, t Type) Value {
return makeRunes(v.flag&flagRO, []rune(v.String()), t)
}
// convertOp: direct copy
func cvtDirect(v Value, typ Type) Value {
f := v.flag
t := typ.common()
ptr := v.ptr
if f&flagAddr != 0 {
// indirect, mutable word - make a copy
c := unsafe_New(t)
typedmemmove(t, c, ptr)
ptr = c
f &^= flagAddr
}
reflect: shorten value to 3 words scalar is no longer needed, now that interfaces always hold pointers. Comparing best of 5 with TurboBoost turned off, on a 2012 Retina MacBook Pro Core i5. Still not completely confident in these numbers, but the gob and template improvements seem real. benchmark old ns/op new ns/op delta BenchmarkBinaryTree17 3819892491 3803008185 -0.44% BenchmarkFannkuch11 3623876405 3611776426 -0.33% BenchmarkFmtFprintfEmpty 119 118 -0.84% BenchmarkFmtFprintfString 294 292 -0.68% BenchmarkFmtFprintfInt 310 304 -1.94% BenchmarkFmtFprintfIntInt 513 507 -1.17% BenchmarkFmtFprintfPrefixedInt 427 426 -0.23% BenchmarkFmtFprintfFloat 562 554 -1.42% BenchmarkFmtManyArgs 1873 1832 -2.19% BenchmarkGobDecode 15824504 14746565 -6.81% BenchmarkGobEncode 14347378 14208743 -0.97% BenchmarkGzip 537229271 537973492 +0.14% BenchmarkGunzip 134996775 135406149 +0.30% BenchmarkHTTPClientServer 119065 116937 -1.79% BenchmarkJSONEncode 29134359 28928099 -0.71% BenchmarkJSONDecode 106867289 105770161 -1.03% BenchmarkMandelbrot200 5798475 5791433 -0.12% BenchmarkGoParse 5299169 5379201 +1.51% BenchmarkRegexpMatchEasy0_32 195 195 +0.00% BenchmarkRegexpMatchEasy0_1K 477 477 +0.00% BenchmarkRegexpMatchEasy1_32 170 170 +0.00% BenchmarkRegexpMatchEasy1_1K 1412 1397 -1.06% BenchmarkRegexpMatchMedium_32 336 337 +0.30% BenchmarkRegexpMatchMedium_1K 109025 108977 -0.04% BenchmarkRegexpMatchHard_32 5854 5856 +0.03% BenchmarkRegexpMatchHard_1K 184914 184748 -0.09% BenchmarkRevcomp 829233526 836598734 +0.89% BenchmarkTemplate 142055312 137016166 -3.55% BenchmarkTimeParse 598 597 -0.17% BenchmarkTimeFormat 564 568 +0.71% Fixes #7425. LGTM=r R=golang-codereviews, r CC=golang-codereviews, iant, khr https://golang.org/cl/158890043
2014-10-15 14:24:18 -04:00
return Value{t, ptr, v.flag&flagRO | f} // v.flag&flagRO|f == f?
}
// convertOp: concrete -> interface
func cvtT2I(v Value, typ Type) Value {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
target := unsafe_New(typ.common())
x := valueInterface(v, false)
if typ.NumMethod() == 0 {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
*(*interface{})(target) = x
} else {
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
ifaceE2I(typ.(*rtype), x, target)
}
reflect: allocate correct type in assignTo and cvtT2I I came across this while debugging a GC problem in gccgo. There is code in assignTo and cvtT2I that handles assignment to all interface values. It allocates an empty interface even if the real type is a non-empty interface. The fields are then set for a non-empty interface, but the memory is recorded as holding an empty interface. This means that the GC has incorrect information. This is extremely unlikely to fail, because the code in the GC that handles empty interfaces looks like this: obj = nil; typ = eface->type; if(typ != nil) { if(!(typ->kind&KindDirectIface) || !(typ->kind&KindNoPointers)) obj = eface->data; In the current runtime the condition is always true--if KindDirectIface is set, then KindNoPointers is clear--and we always want to set obj = eface->data. So the question is what happens when we incorrectly store a non-empty interface value in memory marked as an empty interface. In that case eface->type will not be a *rtype as we expect, but will instead be a pointer to an Itab. We are going to use this pointer to look at a *rtype kind field. The *rtype struct starts out like this: type rtype struct { size uintptr hash uint32 // hash of type; avoids computation in hash tables _ uint8 // unused/padding align uint8 // alignment of variable with this type fieldAlign uint8 // alignment of struct field with this type kind uint8 // enumeration for C An Itab always has at least two pointers, so on a little-endian 64-bit system the kind field will be the high byte of the second pointer. This will normally be zero, so the test of typ->kind will succeed, which is what we want. On a 32-bit system it might be possible to construct a failing case by somehow getting the Itab for an interface with one method to be immediately followed by a word that is all ones. The effect would be that the test would sometimes fail and the GC would not mark obj, leading to an invalid dangling pointer. I have not tried to construct this test. I noticed this in gccgo, where this error is much more likely to cause trouble for a rather random reason: gccgo uses a different layout of rtype, and in gccgo the kind field happens to be the low byte of a pointer, not the high byte. LGTM=rsc R=rsc CC=golang-codereviews https://golang.org/cl/155450044
2014-10-20 10:43:43 -07:00
return Value{typ.common(), target, v.flag&flagRO | flagIndir | flag(Interface)}
}
// convertOp: interface -> interface
func cvtI2I(v Value, typ Type) Value {
if v.IsNil() {
ret := Zero(typ)
ret.flag |= v.flag & flagRO
return ret
}
return cvtT2I(v.Elem(), typ)
}
// implemented in ../runtime
func chancap(ch unsafe.Pointer) int
func chanclose(ch unsafe.Pointer)
func chanlen(ch unsafe.Pointer) int
// Note: some of the noescape annotations below are technically a lie,
// but safe in the context of this package. Functions like chansend
// and mapassign don't escape the referent, but may escape anything
// the referent points to (they do shallow copies of the referent).
// It is safe in this package because the referent may only point
// to something a Value may point to, and that is always in the heap
// (due to the escapes() call in ValueOf).
//go:noescape
func chanrecv(t *rtype, ch unsafe.Pointer, nb bool, val unsafe.Pointer) (selected, received bool)
//go:noescape
func chansend(t *rtype, ch unsafe.Pointer, val unsafe.Pointer, nb bool) bool
func makechan(typ *rtype, size uint64) (ch unsafe.Pointer)
func makemap(t *rtype) (m unsafe.Pointer)
//go:noescape
func mapaccess(t *rtype, m unsafe.Pointer, key unsafe.Pointer) (val unsafe.Pointer)
//go:noescape
func mapassign(t *rtype, m unsafe.Pointer, key, val unsafe.Pointer)
//go:noescape
func mapdelete(t *rtype, m unsafe.Pointer, key unsafe.Pointer)
// m escapes into the return value, but the caller of mapiterinit
// doesn't let the return value escape.
//go:noescape
func mapiterinit(t *rtype, m unsafe.Pointer) unsafe.Pointer
//go:noescape
func mapiterkey(it unsafe.Pointer) (key unsafe.Pointer)
//go:noescape
func mapiternext(it unsafe.Pointer)
//go:noescape
func maplen(m unsafe.Pointer) int
// call calls fn with a copy of the n argument bytes pointed at by arg.
// After fn returns, reflectcall copies n-retoffset result bytes
// back into arg+retoffset before returning. If copying result bytes back,
// the caller must pass the argument frame type as argtype, so that
// call can execute appropriate write barriers during the copy.
func call(argtype *rtype, fn, arg unsafe.Pointer, n uint32, retoffset uint32)
func ifaceE2I(t *rtype, src interface{}, dst unsafe.Pointer)
// typedmemmove copies a value of type t to dst from src.
//go:noescape
func typedmemmove(t *rtype, dst, src unsafe.Pointer)
// typedmemmovepartial is like typedmemmove but assumes that
// dst and src point off bytes into the value and only copies size bytes.
//go:noescape
func typedmemmovepartial(t *rtype, dst, src unsafe.Pointer, off, size uintptr)
// typedslicecopy copies a slice of elemType values from src to dst,
// returning the number of elements copied.
//go:noescape
func typedslicecopy(elemType *rtype, dst, src sliceHeader) int
//go:noescape
func memclr(ptr unsafe.Pointer, n uintptr)
// Dummy annotation marking that the value x escapes,
// for use in cases where the reflect code is so clever that
// the compiler cannot follow.
func escapes(x interface{}) {
if dummy.b {
dummy.x = x
}
}
var dummy struct {
b bool
x interface{}
}