[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
// Copyright 2021 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 noder
|
|
|
|
|
|
|
|
|
|
import (
|
2021-05-27 02:50:17 -07:00
|
|
|
"fmt"
|
|
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
"cmd/compile/internal/base"
|
|
|
|
|
"cmd/compile/internal/ir"
|
|
|
|
|
"cmd/compile/internal/syntax"
|
|
|
|
|
"cmd/compile/internal/typecheck"
|
|
|
|
|
"cmd/compile/internal/types"
|
|
|
|
|
"cmd/compile/internal/types2"
|
2021-01-19 13:54:33 -08:00
|
|
|
"cmd/internal/src"
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
)
|
|
|
|
|
|
|
|
|
|
func (g *irgen) expr(expr syntax.Expr) ir.Node {
|
2021-05-27 02:50:17 -07:00
|
|
|
expr = unparen(expr) // skip parens; unneeded after parse+typecheck
|
|
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
if expr == nil {
|
|
|
|
|
return nil
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if expr, ok := expr.(*syntax.Name); ok && expr.Value == "_" {
|
|
|
|
|
return ir.BlankNode
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
tv, ok := g.info.Types[expr]
|
|
|
|
|
if !ok {
|
|
|
|
|
base.FatalfAt(g.pos(expr), "missing type for %v (%T)", expr, expr)
|
|
|
|
|
}
|
|
|
|
|
switch {
|
|
|
|
|
case tv.IsBuiltin():
|
2021-04-21 02:11:15 -07:00
|
|
|
// Qualified builtins, such as unsafe.Add and unsafe.Slice.
|
|
|
|
|
if expr, ok := expr.(*syntax.SelectorExpr); ok {
|
|
|
|
|
if name, ok := expr.X.(*syntax.Name); ok {
|
|
|
|
|
if _, ok := g.info.Uses[name].(*types2.PkgName); ok {
|
|
|
|
|
return g.use(expr.Sel)
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
return g.use(expr.(*syntax.Name))
|
|
|
|
|
case tv.IsType():
|
|
|
|
|
return ir.TypeNode(g.typ(tv.Type))
|
|
|
|
|
case tv.IsValue(), tv.IsVoid():
|
|
|
|
|
// ok
|
|
|
|
|
default:
|
|
|
|
|
base.FatalfAt(g.pos(expr), "unrecognized type-checker result")
|
|
|
|
|
}
|
|
|
|
|
|
2021-01-20 12:54:23 -08:00
|
|
|
// The gc backend expects all expressions to have a concrete type, and
|
|
|
|
|
// types2 mostly satisfies this expectation already. But there are a few
|
|
|
|
|
// cases where the Go spec doesn't require converting to concrete type,
|
|
|
|
|
// and so types2 leaves them untyped. So we need to fix those up here.
|
|
|
|
|
typ := tv.Type
|
|
|
|
|
if basic, ok := typ.(*types2.Basic); ok && basic.Info()&types2.IsUntyped != 0 {
|
|
|
|
|
switch basic.Kind() {
|
|
|
|
|
case types2.UntypedNil:
|
|
|
|
|
// ok; can appear in type switch case clauses
|
|
|
|
|
// TODO(mdempsky): Handle as part of type switches instead?
|
|
|
|
|
case types2.UntypedBool:
|
|
|
|
|
typ = types2.Typ[types2.Bool] // expression in "if" or "for" condition
|
|
|
|
|
case types2.UntypedString:
|
|
|
|
|
typ = types2.Typ[types2.String] // argument to "append" or "copy" calls
|
|
|
|
|
default:
|
|
|
|
|
base.FatalfAt(g.pos(expr), "unexpected untyped type: %v", basic)
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
// Constant expression.
|
|
|
|
|
if tv.Value != nil {
|
2021-05-27 02:50:17 -07:00
|
|
|
typ := g.typ(typ)
|
|
|
|
|
value := FixValue(typ, tv.Value)
|
|
|
|
|
return OrigConst(g.pos(expr), typ, value, constExprOp(expr), syntax.String(expr))
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
|
2021-01-20 12:54:23 -08:00
|
|
|
n := g.expr0(typ, expr)
|
cmd/compile: replace calls to typecheck with transform functions
For additions, compares, and slices, create transform functions that do
just the transformations for those nodes by the typecheck package (given
that the code has been fully typechecked by types2). For nodes that have
no args with typeparams, we call these transform functions directly in
noder2. But for nodes that have args with typeparams, we have to delay
and call the tranform functions during stenciling, since we don't know
the specific types involved.
We indicate that a node still needs transformation by setting Typecheck
to a new value 3. This value means the current type of the node has been
set (via types2), but the node may still need transformation.
Had to export typcheck.IsCmp and typecheck.Assignop from the typecheck
package.
Added new tests list2.go (required delaying compare typecheck/transform
because of != compare in checkList) and adder.go (requires delaying add
typecheck/transform, since it can do addition for numbers or strings).
There are several more transformation functions needed for expressions
(indexing, calls, etc.) and several more complicated ones needed for
statements (mainly various kinds of assignments).
Change-Id: I7d89d13a4108308ea0304a4b815ab60b40c59b0a
Reviewed-on: https://go-review.googlesource.com/c/go/+/303091
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
2021-03-18 14:36:39 -07:00
|
|
|
if n.Typecheck() != 1 && n.Typecheck() != 3 {
|
2021-01-20 12:54:23 -08:00
|
|
|
base.FatalfAt(g.pos(expr), "missed typecheck: %+v", n)
|
|
|
|
|
}
|
|
|
|
|
if !g.match(n.Type(), typ, tv.HasOk()) {
|
|
|
|
|
base.FatalfAt(g.pos(expr), "expected %L to have type %v", n, typ)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
return n
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func (g *irgen) expr0(typ types2.Type, expr syntax.Expr) ir.Node {
|
|
|
|
|
pos := g.pos(expr)
|
|
|
|
|
|
|
|
|
|
switch expr := expr.(type) {
|
|
|
|
|
case *syntax.Name:
|
|
|
|
|
if _, isNil := g.info.Uses[expr].(*types2.Nil); isNil {
|
|
|
|
|
return Nil(pos, g.typ(typ))
|
|
|
|
|
}
|
2021-03-17 17:54:41 -07:00
|
|
|
return g.use(expr)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
|
|
|
|
|
case *syntax.CompositeLit:
|
|
|
|
|
return g.compLit(typ, expr)
|
2021-02-08 14:33:51 -08:00
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
case *syntax.FuncLit:
|
|
|
|
|
return g.funcLit(typ, expr)
|
|
|
|
|
|
|
|
|
|
case *syntax.AssertExpr:
|
|
|
|
|
return Assert(pos, g.expr(expr.X), g.typeExpr(expr.Type))
|
2021-02-08 14:33:51 -08:00
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
case *syntax.CallExpr:
|
2021-02-02 13:04:16 -08:00
|
|
|
fun := g.expr(expr.Fun)
|
2021-03-09 18:24:51 -08:00
|
|
|
|
2021-03-29 16:07:18 -07:00
|
|
|
// The key for the Inferred map is the CallExpr (if inferring
|
|
|
|
|
// types required the function arguments) or the IndexExpr below
|
|
|
|
|
// (if types could be inferred without the function arguments).
|
2021-06-03 21:58:32 -07:00
|
|
|
if inferred, ok := g.info.Inferred[expr]; ok && len(inferred.TArgs) > 0 {
|
2021-03-29 16:07:18 -07:00
|
|
|
// This is the case where inferring types required the
|
|
|
|
|
// types of the function arguments.
|
2021-06-03 21:58:32 -07:00
|
|
|
targs := make([]ir.Node, len(inferred.TArgs))
|
|
|
|
|
for i, targ := range inferred.TArgs {
|
2021-02-02 13:04:16 -08:00
|
|
|
targs[i] = ir.TypeNode(g.typ(targ))
|
|
|
|
|
}
|
|
|
|
|
if fun.Op() == ir.OFUNCINST {
|
|
|
|
|
// Replace explicit type args with the full list that
|
|
|
|
|
// includes the additional inferred type args
|
|
|
|
|
fun.(*ir.InstExpr).Targs = targs
|
|
|
|
|
} else {
|
|
|
|
|
// Create a function instantiation here, given
|
|
|
|
|
// there are only inferred type args (e.g.
|
|
|
|
|
// min(5,6), where min is a generic function)
|
|
|
|
|
inst := ir.NewInstExpr(pos, ir.OFUNCINST, fun, targs)
|
|
|
|
|
typed(fun.Type(), inst)
|
|
|
|
|
fun = inst
|
|
|
|
|
}
|
|
|
|
|
|
2021-01-30 08:43:58 -08:00
|
|
|
}
|
2021-02-02 13:04:16 -08:00
|
|
|
return Call(pos, g.typ(typ), fun, g.exprs(expr.ArgList), expr.HasDots)
|
2021-02-08 14:33:51 -08:00
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
case *syntax.IndexExpr:
|
2021-02-02 12:17:57 -08:00
|
|
|
var targs []ir.Node
|
2021-03-29 16:07:18 -07:00
|
|
|
|
2021-06-03 21:58:32 -07:00
|
|
|
if inferred, ok := g.info.Inferred[expr]; ok && len(inferred.TArgs) > 0 {
|
2021-03-29 16:07:18 -07:00
|
|
|
// This is the partial type inference case where the types
|
|
|
|
|
// can be inferred from other type arguments without using
|
|
|
|
|
// the types of the function arguments.
|
2021-06-03 21:58:32 -07:00
|
|
|
targs = make([]ir.Node, len(inferred.TArgs))
|
|
|
|
|
for i, targ := range inferred.TArgs {
|
2021-03-29 16:07:18 -07:00
|
|
|
targs[i] = ir.TypeNode(g.typ(targ))
|
|
|
|
|
}
|
|
|
|
|
} else if _, ok := expr.Index.(*syntax.ListExpr); ok {
|
2021-02-02 12:17:57 -08:00
|
|
|
targs = g.exprList(expr.Index)
|
2021-01-30 21:15:40 -08:00
|
|
|
} else {
|
2021-02-02 12:17:57 -08:00
|
|
|
index := g.expr(expr.Index)
|
|
|
|
|
if index.Op() != ir.OTYPE {
|
|
|
|
|
// This is just a normal index expression
|
cmd/compile: getting more built-ins to work with generics
For Builtin ops, we currently stay with using the old
typechecker to transform the call to a more specific expression
and possibly use more specific ops. However, for a bunch of the
ops, we delay calling the old typechecker if any of the args have
type params, for a variety of reasons.
In the near future, we will start creating separate functions that do
the same transformations as the old typechecker for calls, builtins,
indexing, comparisons, etc. These functions can then be called at noder
time for nodes with no type params, and at stenciling time for nodes
with type params.
Remove unnecessary calls to types1 typechecker for most kinds of
statements (still need it for SendStmt, AssignStmt, ReturnStmt, and
SelectStmt). In particular, we don't need it for RangeStmt, and this
avoids some complaints by the types1 typechecker on generic code.
Other small changes:
- Fix check on whether to delay calling types1-typechecker on type
conversions. Should check if HasTParam is true, rather than if the
type is directly a TYPEPARAM.
- Don't call types1-typechecker on an indexing operation if the left
operand has a typeparam in its type and is not obviously a TMAP,
TSLICE, or TARRAY. As above, we will eventually have to create a new
function that can do the required transformations (for complicated
cases) at noder time or stenciling time.
- Copy n.BuiltinOp in subster.node()
- The complex arithmetic example in absdiff.go now works.
- Added new tests double.go and append.go
- Added new example with a new() call in settable.go
Change-Id: I8f377afb6126cab1826bd3c2732aa8cdf1f7e0b4
Reviewed-on: https://go-review.googlesource.com/c/go/+/301951
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
2021-03-12 11:36:02 -08:00
|
|
|
return Index(pos, g.typ(typ), g.expr(expr.X), index)
|
2021-01-30 21:15:40 -08:00
|
|
|
}
|
2021-02-02 12:17:57 -08:00
|
|
|
// This is generic function instantiation with a single type
|
|
|
|
|
targs = []ir.Node{index}
|
|
|
|
|
}
|
2021-03-29 16:07:18 -07:00
|
|
|
// This is a generic function instantiation (e.g. min[int]).
|
|
|
|
|
// Generic type instantiation is handled in the type
|
|
|
|
|
// section of expr() above (using g.typ).
|
2021-02-02 12:17:57 -08:00
|
|
|
x := g.expr(expr.X)
|
|
|
|
|
if x.Op() != ir.ONAME || x.Type().Kind() != types.TFUNC {
|
|
|
|
|
panic("Incorrect argument for generic func instantiation")
|
2021-01-30 08:43:58 -08:00
|
|
|
}
|
2021-02-02 12:17:57 -08:00
|
|
|
n := ir.NewInstExpr(pos, ir.OFUNCINST, x, targs)
|
|
|
|
|
typed(g.typ(typ), n)
|
|
|
|
|
return n
|
|
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
case *syntax.SelectorExpr:
|
2021-01-20 13:54:53 -08:00
|
|
|
// Qualified identifier.
|
|
|
|
|
if name, ok := expr.X.(*syntax.Name); ok {
|
|
|
|
|
if _, ok := g.info.Uses[name].(*types2.PkgName); ok {
|
2021-03-17 17:54:41 -07:00
|
|
|
return g.use(expr.Sel)
|
2021-01-20 13:54:53 -08:00
|
|
|
}
|
|
|
|
|
}
|
2021-02-08 14:33:51 -08:00
|
|
|
return g.selectorExpr(pos, typ, expr)
|
2021-01-20 13:54:53 -08:00
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
case *syntax.SliceExpr:
|
cmd/compile: replace calls to typecheck with transform functions
For additions, compares, and slices, create transform functions that do
just the transformations for those nodes by the typecheck package (given
that the code has been fully typechecked by types2). For nodes that have
no args with typeparams, we call these transform functions directly in
noder2. But for nodes that have args with typeparams, we have to delay
and call the tranform functions during stenciling, since we don't know
the specific types involved.
We indicate that a node still needs transformation by setting Typecheck
to a new value 3. This value means the current type of the node has been
set (via types2), but the node may still need transformation.
Had to export typcheck.IsCmp and typecheck.Assignop from the typecheck
package.
Added new tests list2.go (required delaying compare typecheck/transform
because of != compare in checkList) and adder.go (requires delaying add
typecheck/transform, since it can do addition for numbers or strings).
There are several more transformation functions needed for expressions
(indexing, calls, etc.) and several more complicated ones needed for
statements (mainly various kinds of assignments).
Change-Id: I7d89d13a4108308ea0304a4b815ab60b40c59b0a
Reviewed-on: https://go-review.googlesource.com/c/go/+/303091
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
2021-03-18 14:36:39 -07:00
|
|
|
return Slice(pos, g.typ(typ), g.expr(expr.X), g.expr(expr.Index[0]), g.expr(expr.Index[1]), g.expr(expr.Index[2]))
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
|
|
|
|
|
case *syntax.Operation:
|
|
|
|
|
if expr.Y == nil {
|
2021-03-29 08:28:01 -07:00
|
|
|
return Unary(pos, g.typ(typ), g.op(expr.Op, unOps[:]), g.expr(expr.X))
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
switch op := g.op(expr.Op, binOps[:]); op {
|
|
|
|
|
case ir.OEQ, ir.ONE, ir.OLT, ir.OLE, ir.OGT, ir.OGE:
|
|
|
|
|
return Compare(pos, g.typ(typ), op, g.expr(expr.X), g.expr(expr.Y))
|
|
|
|
|
default:
|
cmd/compile: replace calls to typecheck with transform functions
For additions, compares, and slices, create transform functions that do
just the transformations for those nodes by the typecheck package (given
that the code has been fully typechecked by types2). For nodes that have
no args with typeparams, we call these transform functions directly in
noder2. But for nodes that have args with typeparams, we have to delay
and call the tranform functions during stenciling, since we don't know
the specific types involved.
We indicate that a node still needs transformation by setting Typecheck
to a new value 3. This value means the current type of the node has been
set (via types2), but the node may still need transformation.
Had to export typcheck.IsCmp and typecheck.Assignop from the typecheck
package.
Added new tests list2.go (required delaying compare typecheck/transform
because of != compare in checkList) and adder.go (requires delaying add
typecheck/transform, since it can do addition for numbers or strings).
There are several more transformation functions needed for expressions
(indexing, calls, etc.) and several more complicated ones needed for
statements (mainly various kinds of assignments).
Change-Id: I7d89d13a4108308ea0304a4b815ab60b40c59b0a
Reviewed-on: https://go-review.googlesource.com/c/go/+/303091
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
2021-03-18 14:36:39 -07:00
|
|
|
return Binary(pos, op, g.typ(typ), g.expr(expr.X), g.expr(expr.Y))
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
default:
|
|
|
|
|
g.unhandled("expression", expr)
|
|
|
|
|
panic("unreachable")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2021-01-19 13:54:33 -08:00
|
|
|
// selectorExpr resolves the choice of ODOT, ODOTPTR, OCALLPART (eventually
|
|
|
|
|
// ODOTMETH & ODOTINTER), and OMETHEXPR and deals with embedded fields here rather
|
|
|
|
|
// than in typecheck.go.
|
2021-02-08 14:33:51 -08:00
|
|
|
func (g *irgen) selectorExpr(pos src.XPos, typ types2.Type, expr *syntax.SelectorExpr) ir.Node {
|
|
|
|
|
x := g.expr(expr.X)
|
2021-03-10 17:27:30 -08:00
|
|
|
if x.Type().HasTParam() {
|
2021-02-08 14:33:51 -08:00
|
|
|
// Leave a method call on a type param as an OXDOT, since it can
|
|
|
|
|
// only be fully transformed once it has an instantiated type.
|
|
|
|
|
n := ir.NewSelectorExpr(pos, ir.OXDOT, x, typecheck.Lookup(expr.Sel.Value))
|
|
|
|
|
typed(g.typ(typ), n)
|
|
|
|
|
return n
|
|
|
|
|
}
|
2021-01-22 13:29:59 -08:00
|
|
|
|
2021-02-08 14:33:51 -08:00
|
|
|
selinfo := g.info.Selections[expr]
|
2021-01-22 13:29:59 -08:00
|
|
|
// Everything up to the last selection is an implicit embedded field access,
|
|
|
|
|
// and the last selection is determined by selinfo.Kind().
|
|
|
|
|
index := selinfo.Index()
|
|
|
|
|
embeds, last := index[:len(index)-1], index[len(index)-1]
|
|
|
|
|
|
2021-01-24 09:59:20 -08:00
|
|
|
origx := x
|
2021-01-22 13:29:59 -08:00
|
|
|
for _, ix := range embeds {
|
|
|
|
|
x = Implicit(DotField(pos, x, ix))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
kind := selinfo.Kind()
|
|
|
|
|
if kind == types2.FieldVal {
|
|
|
|
|
return DotField(pos, x, last)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// TODO(danscales,mdempsky): Interface method sets are not sorted the
|
|
|
|
|
// same between types and types2. In particular, using "last" here
|
|
|
|
|
// without conversion will likely fail if an interface contains
|
|
|
|
|
// unexported methods from two different packages (due to cross-package
|
|
|
|
|
// interface embedding).
|
|
|
|
|
|
2021-01-24 09:59:20 -08:00
|
|
|
var n ir.Node
|
2021-02-11 10:50:20 -08:00
|
|
|
method2 := selinfo.Obj().(*types2.Func)
|
2021-01-22 13:29:59 -08:00
|
|
|
|
2021-01-24 09:59:20 -08:00
|
|
|
if kind == types2.MethodExpr {
|
|
|
|
|
// OMETHEXPR is unusual in using directly the node and type of the
|
|
|
|
|
// original OTYPE node (origx) before passing through embedded
|
|
|
|
|
// fields, even though the method is selected from the type
|
|
|
|
|
// (x.Type()) reached after following the embedded fields. We will
|
|
|
|
|
// actually drop any ODOT nodes we created due to the embedded
|
|
|
|
|
// fields.
|
|
|
|
|
n = MethodExpr(pos, origx, x.Type(), last)
|
|
|
|
|
} else {
|
|
|
|
|
// Add implicit addr/deref for method values, if needed.
|
2021-02-11 10:50:20 -08:00
|
|
|
if x.Type().IsInterface() {
|
|
|
|
|
n = DotMethod(pos, x, last)
|
|
|
|
|
} else {
|
|
|
|
|
recvType2 := method2.Type().(*types2.Signature).Recv().Type()
|
|
|
|
|
_, wantPtr := recvType2.(*types2.Pointer)
|
2021-01-24 09:59:20 -08:00
|
|
|
havePtr := x.Type().IsPtr()
|
|
|
|
|
|
|
|
|
|
if havePtr != wantPtr {
|
|
|
|
|
if havePtr {
|
2021-03-29 08:28:01 -07:00
|
|
|
x = Implicit(Deref(pos, x.Type().Elem(), x))
|
2021-01-24 09:59:20 -08:00
|
|
|
} else {
|
|
|
|
|
x = Implicit(Addr(pos, x))
|
|
|
|
|
}
|
|
|
|
|
}
|
2021-02-11 10:50:20 -08:00
|
|
|
recvType2Base := recvType2
|
|
|
|
|
if wantPtr {
|
2021-02-18 15:09:38 -08:00
|
|
|
recvType2Base = types2.AsPointer(recvType2).Elem()
|
2021-02-11 10:50:20 -08:00
|
|
|
}
|
2021-02-18 15:09:38 -08:00
|
|
|
if len(types2.AsNamed(recvType2Base).TParams()) > 0 {
|
2021-02-11 10:50:20 -08:00
|
|
|
// recvType2 is the original generic type that is
|
|
|
|
|
// instantiated for this method call.
|
|
|
|
|
// selinfo.Recv() is the instantiated type
|
|
|
|
|
recvType2 = recvType2Base
|
[dev.typeparams] cmd/compile: get export/import of generic types & functions working
The general idea is that we now export/import typeparams, typeparam
lists for generic types and functions, and instantiated types
(instantiations of generic types with either new typeparams or concrete
types).
This changes the export format -- the next CL in the stack adds the
export versions and checks for it in the appropriate places.
We always export/import generic function bodies, using the same code
that we use for exporting/importing the bodies of inlineable functions.
To avoid complicated scoping, we consider all type params as unique and
give them unique names for types1. We therefore include the types2 ids
(subscripts) in the export format and re-create on import. We always
access the same unique types1 typeParam type for the same typeparam
name.
We create fully-instantiated generic types and functions in the original
source package. We do an extra NeedRuntimeType() call to make sure that
the correct DWARF information is written out. We call SetDupOK(true) for
the functions/methods to have the linker automatically drop duplicate
instantiations.
Other miscellaneous details:
- Export/import of typeparam bounds works for methods (but not
typelists) for now, but will change with the typeset changes.
- Added a new types.Instantiate function roughly analogous to the
types2.Instantiate function recently added.
- Always access methods info from the original/base generic type, since
the methods of an instantiated type are not filled in (in types2 or
types1).
- New field OrigSym in types.Type to keep track of base generic type
that instantiated type was based on. We use the generic type's symbol
(OrigSym) as the link, rather than a Type pointer, since we haven't
always created the base type yet when we want to set the link (during
types2 to types1 conversion).
- Added types2.AsTypeParam(), (*types2.TypeParam).SetId()
- New test minimp.dir, which tests use of generic function Min across
packages. Another test stringimp.dir, which also exports a generic
function Stringify across packages, where the type param has a bound
(Stringer) as well. New test pairimp.dir, which tests use of generic
type Pair (with no methods) across packages.
- New test valimp.dir, which tests use of generic type (with methods
and related functions) across packages.
- Modified several other tests (adder.go, settable.go, smallest.go,
stringable.go, struct.go, sum.go) to export their generic
functions/types to show that generic functions/types can be exported
successfully (but this doesn't test import).
Change-Id: Ie61ce9d54a46d368ddc7a76c41399378963bb57f
Reviewed-on: https://go-review.googlesource.com/c/go/+/319930
Trust: Dan Scales <danscales@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
2021-04-13 15:37:36 -07:00
|
|
|
recvTypeSym := g.pkg(method2.Pkg()).Lookup(recvType2.(*types2.Named).Obj().Name())
|
|
|
|
|
recvType := recvTypeSym.Def.(*ir.Name).Type()
|
|
|
|
|
// method is the generic method associated with
|
|
|
|
|
// the base generic type. The instantiated type may not
|
|
|
|
|
// have method bodies filled in, if it was imported.
|
|
|
|
|
method := recvType.Methods().Index(last).Nname.(*ir.Name)
|
2021-05-07 13:20:34 -07:00
|
|
|
n = ir.NewSelectorExpr(pos, ir.OCALLPART, x, typecheck.Lookup(expr.Sel.Value))
|
2021-02-11 10:50:20 -08:00
|
|
|
n.(*ir.SelectorExpr).Selection = types.NewField(pos, method.Sym(), method.Type())
|
|
|
|
|
n.(*ir.SelectorExpr).Selection.Nname = method
|
|
|
|
|
typed(method.Type(), n)
|
|
|
|
|
|
|
|
|
|
// selinfo.Targs() are the types used to
|
|
|
|
|
// instantiate the type of receiver
|
2021-03-04 12:09:04 -08:00
|
|
|
targs2 := getTargs(selinfo)
|
2021-02-11 10:50:20 -08:00
|
|
|
targs := make([]ir.Node, len(targs2))
|
|
|
|
|
for i, targ2 := range targs2 {
|
|
|
|
|
targs[i] = ir.TypeNode(g.typ(targ2))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Create function instantiation with the type
|
|
|
|
|
// args for the receiver type for the method call.
|
|
|
|
|
n = ir.NewInstExpr(pos, ir.OFUNCINST, n, targs)
|
|
|
|
|
typed(g.typ(typ), n)
|
|
|
|
|
return n
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if !g.match(x.Type(), recvType2, false) {
|
|
|
|
|
base.FatalfAt(pos, "expected %L to have type %v", x, recvType2)
|
|
|
|
|
} else {
|
|
|
|
|
n = DotMethod(pos, x, last)
|
2021-01-19 13:54:33 -08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
2021-02-11 10:50:20 -08:00
|
|
|
if have, want := n.Sym(), g.selector(method2); have != want {
|
2021-01-22 13:29:59 -08:00
|
|
|
base.FatalfAt(pos, "bad Sym: have %v, want %v", have, want)
|
|
|
|
|
}
|
|
|
|
|
return n
|
2021-01-19 13:54:33 -08:00
|
|
|
}
|
|
|
|
|
|
2021-03-04 12:09:04 -08:00
|
|
|
// getTargs gets the targs associated with the receiver of a selected method
|
|
|
|
|
func getTargs(selinfo *types2.Selection) []types2.Type {
|
|
|
|
|
r := selinfo.Recv()
|
|
|
|
|
if p := types2.AsPointer(r); p != nil {
|
|
|
|
|
r = p.Elem()
|
|
|
|
|
}
|
|
|
|
|
n := types2.AsNamed(r)
|
|
|
|
|
if n == nil {
|
|
|
|
|
base.Fatalf("Incorrect type for selinfo %v", selinfo)
|
|
|
|
|
}
|
|
|
|
|
return n.TArgs()
|
|
|
|
|
}
|
|
|
|
|
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
func (g *irgen) exprList(expr syntax.Expr) []ir.Node {
|
2021-05-27 02:50:17 -07:00
|
|
|
return g.exprs(unpackListExpr(expr))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func unpackListExpr(expr syntax.Expr) []syntax.Expr {
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
switch expr := expr.(type) {
|
|
|
|
|
case nil:
|
|
|
|
|
return nil
|
|
|
|
|
case *syntax.ListExpr:
|
2021-05-27 02:50:17 -07:00
|
|
|
return expr.ElemList
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
default:
|
2021-05-27 02:50:17 -07:00
|
|
|
return []syntax.Expr{expr}
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func (g *irgen) exprs(exprs []syntax.Expr) []ir.Node {
|
|
|
|
|
nodes := make([]ir.Node, len(exprs))
|
|
|
|
|
for i, expr := range exprs {
|
|
|
|
|
nodes[i] = g.expr(expr)
|
|
|
|
|
}
|
|
|
|
|
return nodes
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func (g *irgen) compLit(typ types2.Type, lit *syntax.CompositeLit) ir.Node {
|
|
|
|
|
if ptr, ok := typ.Underlying().(*types2.Pointer); ok {
|
2021-01-20 12:54:23 -08:00
|
|
|
n := ir.NewAddrExpr(g.pos(lit), g.compLit(ptr.Elem(), lit))
|
|
|
|
|
n.SetOp(ir.OPTRLIT)
|
|
|
|
|
return typed(g.typ(typ), n)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
_, isStruct := typ.Underlying().(*types2.Struct)
|
|
|
|
|
|
|
|
|
|
exprs := make([]ir.Node, len(lit.ElemList))
|
|
|
|
|
for i, elem := range lit.ElemList {
|
|
|
|
|
switch elem := elem.(type) {
|
|
|
|
|
case *syntax.KeyValueExpr:
|
2021-06-04 00:01:22 -07:00
|
|
|
var key ir.Node
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
if isStruct {
|
2021-06-04 00:01:22 -07:00
|
|
|
key = ir.NewIdent(g.pos(elem.Key), g.name(elem.Key.(*syntax.Name)))
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
} else {
|
2021-06-04 00:01:22 -07:00
|
|
|
key = g.expr(elem.Key)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
2021-06-04 00:01:22 -07:00
|
|
|
exprs[i] = ir.NewKeyExpr(g.pos(elem), key, g.expr(elem.Value))
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
default:
|
|
|
|
|
exprs[i] = g.expr(elem)
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2021-04-07 07:58:10 -07:00
|
|
|
n := ir.NewCompLitExpr(g.pos(lit), ir.OCOMPLIT, nil, exprs)
|
|
|
|
|
typed(g.typ(typ), n)
|
|
|
|
|
return transformCompLit(n)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
|
2021-02-21 10:54:38 -08:00
|
|
|
func (g *irgen) funcLit(typ2 types2.Type, expr *syntax.FuncLit) ir.Node {
|
[dev.typeparams] cmd/compile: refactor closure construction
typecheck.tcClosure is complicated with many code flows because all of
its callers setup the closure funcs in slightly different ways. E.g.,
it's non-obvious who's responsible for setting the underlying func's
Sym or adding it to target.Decls, or how to write new code that
constructs a closure without interfering with existing code.
This CL refactors everything to use three common functions in package
ir: NewClosureFunc (which handle creating the Func, Name, and
ClosureExpr and wiring them together), NameClosure (which generates
and assigns its unique Sym), and UseClosure (which handles adding the
Func to target.Decls).
Most IR builders can actually name the closure right away, but the
legacy noder+typecheck path may not yet know the name of the enclosing
function. In particular, for methods declared with aliased receiver
parameters, we need to wait until after typechecking top-level
declarations to know the method's true name. So they're left anonymous
until typecheck.
UseClosure does relatively little work today, but it serves as a
useful spot to check that the code setting up closures got it right.
It may also eventually serve as an optimization point for early
lifting of trivial closures, which may or may not ultimately be
beneficial.
Change-Id: I7da1e93c70d268f575b12d6aaeb2336eb910a6f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/327051
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
2021-06-11 03:09:26 -07:00
|
|
|
fn := ir.NewClosureFunc(g.pos(expr), ir.CurFunc)
|
|
|
|
|
ir.NameClosure(fn.OClosure, ir.CurFunc)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
|
2021-02-21 10:54:38 -08:00
|
|
|
typ := g.typ(typ2)
|
|
|
|
|
typed(typ, fn.Nname)
|
|
|
|
|
typed(typ, fn.OClosure)
|
[dev.typeparams] cmd/compile: refactor closure construction
typecheck.tcClosure is complicated with many code flows because all of
its callers setup the closure funcs in slightly different ways. E.g.,
it's non-obvious who's responsible for setting the underlying func's
Sym or adding it to target.Decls, or how to write new code that
constructs a closure without interfering with existing code.
This CL refactors everything to use three common functions in package
ir: NewClosureFunc (which handle creating the Func, Name, and
ClosureExpr and wiring them together), NameClosure (which generates
and assigns its unique Sym), and UseClosure (which handles adding the
Func to target.Decls).
Most IR builders can actually name the closure right away, but the
legacy noder+typecheck path may not yet know the name of the enclosing
function. In particular, for methods declared with aliased receiver
parameters, we need to wait until after typechecking top-level
declarations to know the method's true name. So they're left anonymous
until typecheck.
UseClosure does relatively little work today, but it serves as a
useful spot to check that the code setting up closures got it right.
It may also eventually serve as an optimization point for early
lifting of trivial closures, which may or may not ultimately be
beneficial.
Change-Id: I7da1e93c70d268f575b12d6aaeb2336eb910a6f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/327051
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
2021-06-11 03:09:26 -07:00
|
|
|
fn.SetTypecheck(1)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
|
|
|
|
|
g.funcBody(fn, nil, expr.Type, expr.Body)
|
|
|
|
|
|
|
|
|
|
ir.FinishCaptureNames(fn.Pos(), ir.CurFunc, fn)
|
|
|
|
|
|
|
|
|
|
// TODO(mdempsky): ir.CaptureName should probably handle
|
|
|
|
|
// copying these fields from the canonical variable.
|
|
|
|
|
for _, cv := range fn.ClosureVars {
|
|
|
|
|
cv.SetType(cv.Canonical().Type())
|
|
|
|
|
cv.SetTypecheck(1)
|
|
|
|
|
cv.SetWalkdef(1)
|
|
|
|
|
}
|
|
|
|
|
|
[dev.typeparams] cmd/compile: refactor closure construction
typecheck.tcClosure is complicated with many code flows because all of
its callers setup the closure funcs in slightly different ways. E.g.,
it's non-obvious who's responsible for setting the underlying func's
Sym or adding it to target.Decls, or how to write new code that
constructs a closure without interfering with existing code.
This CL refactors everything to use three common functions in package
ir: NewClosureFunc (which handle creating the Func, Name, and
ClosureExpr and wiring them together), NameClosure (which generates
and assigns its unique Sym), and UseClosure (which handles adding the
Func to target.Decls).
Most IR builders can actually name the closure right away, but the
legacy noder+typecheck path may not yet know the name of the enclosing
function. In particular, for methods declared with aliased receiver
parameters, we need to wait until after typechecking top-level
declarations to know the method's true name. So they're left anonymous
until typecheck.
UseClosure does relatively little work today, but it serves as a
useful spot to check that the code setting up closures got it right.
It may also eventually serve as an optimization point for early
lifting of trivial closures, which may or may not ultimately be
beneficial.
Change-Id: I7da1e93c70d268f575b12d6aaeb2336eb910a6f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/327051
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
2021-06-11 03:09:26 -07:00
|
|
|
return ir.UseClosure(fn.OClosure, g.target)
|
[dev.typeparams] cmd/compile: add types2-based noder
This CL adds "irgen", a new noding implementation that utilizes types2
to guide IR construction. Notably, it completely skips dealing with
constant and type expressions (aside from using ir.TypeNode to
interoperate with the types1 typechecker), because types2 already
handled those. It also omits any syntax checking, trusting that types2
already rejected any errors.
It currently still utilizes the types1 typechecker for the desugaring
operations it handles (e.g., turning OAS2 into OAS2FUNC/etc, inserting
implicit conversions, rewriting f(g()) functions, and so on). However,
the IR is constructed in a fully incremental fashion, so it should be
easy to now piecemeal replace those dependencies as needed.
Nearly all of "go test std cmd" passes with -G=3 enabled by
default. The main remaining blocker is the number of test/run.go
failures. There also appear to be cases where types2 does not provide
us with position information. These will be iterated upon.
Portions and ideas from Dan Scales's CL 276653.
Change-Id: Ic99e8f2d0267b0312d30c10d5d043f5817a59c9d
Reviewed-on: https://go-review.googlesource.com/c/go/+/281932
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
2021-01-09 00:57:55 -08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func (g *irgen) typeExpr(typ syntax.Expr) *types.Type {
|
|
|
|
|
n := g.expr(typ)
|
|
|
|
|
if n.Op() != ir.OTYPE {
|
|
|
|
|
base.FatalfAt(g.pos(typ), "expected type: %L", n)
|
|
|
|
|
}
|
|
|
|
|
return n.Type()
|
|
|
|
|
}
|
2021-05-27 02:50:17 -07:00
|
|
|
|
|
|
|
|
// constExprOp returns an ir.Op that represents the outermost
|
|
|
|
|
// operation of the given constant expression. It's intended for use
|
|
|
|
|
// with ir.RawOrigExpr.
|
|
|
|
|
func constExprOp(expr syntax.Expr) ir.Op {
|
|
|
|
|
switch expr := expr.(type) {
|
|
|
|
|
default:
|
|
|
|
|
panic(fmt.Sprintf("%s: unexpected expression: %T", expr.Pos(), expr))
|
|
|
|
|
|
|
|
|
|
case *syntax.BasicLit:
|
|
|
|
|
return ir.OLITERAL
|
|
|
|
|
case *syntax.Name, *syntax.SelectorExpr:
|
|
|
|
|
return ir.ONAME
|
|
|
|
|
case *syntax.CallExpr:
|
|
|
|
|
return ir.OCALL
|
|
|
|
|
case *syntax.Operation:
|
|
|
|
|
if expr.Y == nil {
|
|
|
|
|
return unOps[expr.Op]
|
|
|
|
|
}
|
|
|
|
|
return binOps[expr.Op]
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
func unparen(expr syntax.Expr) syntax.Expr {
|
|
|
|
|
for {
|
|
|
|
|
paren, ok := expr.(*syntax.ParenExpr)
|
|
|
|
|
if !ok {
|
|
|
|
|
return expr
|
|
|
|
|
}
|
|
|
|
|
expr = paren.X
|
|
|
|
|
}
|
|
|
|
|
}
|