go/src/cmd/compile/internal/importer/iimport.go

775 lines
18 KiB
Go
Raw Normal View History

// Copyright 2018 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.
// Indexed package import.
// See cmd/compile/internal/typecheck/iexport.go for the export data format.
package importer
import (
"cmd/compile/internal/syntax"
"cmd/compile/internal/types2"
"encoding/binary"
"fmt"
"go/constant"
"go/token"
"io"
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
"math/big"
"sort"
"strings"
)
type intReader struct {
*strings.Reader
path string
}
func (r *intReader) int64() int64 {
i, err := binary.ReadVarint(r.Reader)
if err != nil {
errorf("import %q: read varint error: %v", r.path, err)
}
return i
}
func (r *intReader) uint64() uint64 {
i, err := binary.ReadUvarint(r.Reader)
if err != nil {
errorf("import %q: read varint error: %v", r.path, err)
}
return i
}
// Keep this in sync with constants in iexport.go.
const (
iexportVersionGo1_11 = 0
iexportVersionPosCol = 1
iexportVersionGenerics = 2
// Start of the unstable series of versions, remove "+ n" before release.
iexportVersionCurrent = iexportVersionGenerics + 1
)
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
type ident struct {
pkg string
name string
}
const predeclReserved = 32
type itag uint64
const (
// Types
definedType itag = iota
pointerType
sliceType
arrayType
chanType
mapType
signatureType
structType
interfaceType
[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
typeParamType
instType
unionType
)
const io_SeekCurrent = 1 // io.SeekCurrent (not defined in Go 1.4)
// iImportData imports a package from the serialized package data
// and returns the number of bytes consumed and a reference to the package.
// If the export data version is not recognized or the format is otherwise
// compromised, an error is returned.
func ImportData(imports map[string]*types2.Package, data, path string) (pkg *types2.Package, err error) {
const currentVersion = iexportVersionCurrent
version := int64(-1)
defer func() {
if e := recover(); e != nil {
if version > currentVersion {
err = fmt.Errorf("cannot import %q (%v), export data is newer version - update tool", path, e)
} else {
err = fmt.Errorf("cannot import %q (%v), possibly version skew - reinstall package", path, e)
}
}
}()
r := &intReader{strings.NewReader(data), path}
version = int64(r.uint64())
switch version {
case currentVersion, iexportVersionPosCol, iexportVersionGo1_11:
default:
if version > iexportVersionGenerics {
errorf("unstable iexport format version %d, just rebuild compiler and std library", version)
} else {
errorf("unknown iexport format version %d", version)
}
}
sLen := int64(r.uint64())
dLen := int64(r.uint64())
whence, _ := r.Seek(0, io_SeekCurrent)
stringData := data[whence : whence+sLen]
declData := data[whence+sLen : whence+sLen+dLen]
r.Seek(sLen+dLen, io_SeekCurrent)
p := iimporter{
exportVersion: version,
ipath: path,
version: int(version),
stringData: stringData,
pkgCache: make(map[uint64]*types2.Package),
posBaseCache: make(map[uint64]*syntax.PosBase),
declData: declData,
pkgIndex: make(map[*types2.Package]map[string]uint64),
typCache: make(map[uint64]types2.Type),
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
// Separate map for typeparams, keyed by their package and unique
// name (name with subscript).
tparamIndex: make(map[ident]types2.Type),
}
for i, pt := range predeclared {
p.typCache[uint64(i)] = pt
}
pkgList := make([]*types2.Package, r.uint64())
for i := range pkgList {
pkgPathOff := r.uint64()
pkgPath := p.stringAt(pkgPathOff)
pkgName := p.stringAt(r.uint64())
pkgHeight := int(r.uint64())
if pkgPath == "" {
pkgPath = path
}
pkg := imports[pkgPath]
if pkg == nil {
pkg = types2.NewPackageHeight(pkgPath, pkgName, pkgHeight)
imports[pkgPath] = pkg
} else {
if pkg.Name() != pkgName {
errorf("conflicting names %s and %s for package %q", pkg.Name(), pkgName, path)
}
if pkg.Height() != pkgHeight {
errorf("conflicting heights %v and %v for package %q", pkg.Height(), pkgHeight, path)
}
}
p.pkgCache[pkgPathOff] = pkg
nameIndex := make(map[string]uint64)
for nSyms := r.uint64(); nSyms > 0; nSyms-- {
name := p.stringAt(r.uint64())
nameIndex[name] = r.uint64()
}
p.pkgIndex[pkg] = nameIndex
pkgList[i] = pkg
}
localpkg := pkgList[0]
names := make([]string, 0, len(p.pkgIndex[localpkg]))
for name := range p.pkgIndex[localpkg] {
names = append(names, name)
}
sort.Strings(names)
for _, name := range names {
p.doDecl(localpkg, name)
}
// record all referenced packages as imports
list := append(([]*types2.Package)(nil), pkgList[1:]...)
sort.Sort(byPath(list))
localpkg.SetImports(list)
// package was imported completely and without errors
localpkg.MarkComplete()
return localpkg, nil
}
type iimporter struct {
exportVersion int64
ipath string
version int
stringData string
pkgCache map[uint64]*types2.Package
posBaseCache map[uint64]*syntax.PosBase
declData string
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
pkgIndex map[*types2.Package]map[string]uint64
typCache map[uint64]types2.Type
tparamIndex map[ident]types2.Type
interfaceList []*types2.Interface
}
func (p *iimporter) doDecl(pkg *types2.Package, name string) {
// See if we've already imported this declaration.
if obj := pkg.Scope().Lookup(name); obj != nil {
return
}
off, ok := p.pkgIndex[pkg][name]
if !ok {
errorf("%v.%v not in index", pkg, name)
}
r := &importReader{p: p, currPkg: pkg}
// Reader.Reset is not available in Go 1.4.
// Use bytes.NewReader for now.
// r.declReader.Reset(p.declData[off:])
r.declReader = *strings.NewReader(p.declData[off:])
r.obj(name)
}
func (p *iimporter) stringAt(off uint64) string {
var x [binary.MaxVarintLen64]byte
n := copy(x[:], p.stringData[off:])
slen, n := binary.Uvarint(x[:n])
if n <= 0 {
errorf("varint failed")
}
spos := off + uint64(n)
return p.stringData[spos : spos+slen]
}
func (p *iimporter) pkgAt(off uint64) *types2.Package {
if pkg, ok := p.pkgCache[off]; ok {
return pkg
}
path := p.stringAt(off)
errorf("missing package %q in %q", path, p.ipath)
return nil
}
func (p *iimporter) posBaseAt(off uint64) *syntax.PosBase {
if posBase, ok := p.posBaseCache[off]; ok {
return posBase
}
filename := p.stringAt(off)
posBase := syntax.NewFileBase(filename)
p.posBaseCache[off] = posBase
return posBase
}
func (p *iimporter) typAt(off uint64, base *types2.Named) types2.Type {
if t, ok := p.typCache[off]; ok && (base == nil || !isInterface(t)) {
return t
}
if off < predeclReserved {
errorf("predeclared type missing from cache: %v", off)
}
r := &importReader{p: p}
// Reader.Reset is not available in Go 1.4.
// Use bytes.NewReader for now.
// r.declReader.Reset(p.declData[off-predeclReserved:])
r.declReader = *strings.NewReader(p.declData[off-predeclReserved:])
t := r.doType(base)
if base == nil || !isInterface(t) {
p.typCache[off] = t
}
return t
}
type importReader struct {
p *iimporter
declReader strings.Reader
currPkg *types2.Package
prevPosBase *syntax.PosBase
prevLine int64
prevColumn int64
}
func (r *importReader) obj(name string) {
tag := r.byte()
pos := r.pos()
switch tag {
case 'A':
typ := r.typ()
r.declare(types2.NewTypeName(pos, r.currPkg, name, typ))
case 'C':
typ, val := r.value()
r.declare(types2.NewConst(pos, r.currPkg, name, typ, val))
case 'F':
var tparams []*types2.TypeName
if r.p.exportVersion >= iexportVersionGenerics {
tparams = r.tparamList()
}
sig := r.signature(nil)
[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
sig.SetTParams(tparams)
r.declare(types2.NewFunc(pos, r.currPkg, name, sig))
case 'T':
var tparams []*types2.TypeName
if r.p.exportVersion >= iexportVersionGenerics {
tparams = r.tparamList()
}
[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
// Types can be recursive. We need to setup a stub
// declaration before recursing.
obj := types2.NewTypeName(pos, r.currPkg, name, nil)
named := types2.NewNamed(obj, nil, nil)
[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
named.SetTParams(tparams)
r.declare(obj)
underlying := r.p.typAt(r.uint64(), named).Underlying()
named.SetUnderlying(underlying)
if !isInterface(underlying) {
for n := r.uint64(); n > 0; n-- {
mpos := r.pos()
mname := r.ident()
recv := r.param()
msig := r.signature(recv)
[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
// If the receiver has any targs, set those as the
// rparams of the method (since those are the
// typeparams being used in the method sig/body).
targs := baseType(msig.Recv().Type()).TArgs()
if len(targs) > 0 {
rparams := make([]*types2.TypeName, len(targs))
for i, targ := range targs {
rparams[i] = types2.AsTypeParam(targ).Obj()
}
msig.SetRParams(rparams)
}
named.AddMethod(types2.NewFunc(mpos, r.currPkg, mname, msig))
}
}
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
case 'P':
// We need to "declare" a typeparam in order to have a name that
// can be referenced recursively (if needed) in the type param's
// bound.
if r.p.exportVersion < iexportVersionGenerics {
errorf("unexpected type param type")
}
name0, sub := parseSubscript(name)
tn := types2.NewTypeName(pos, r.currPkg, name0, nil)
t := (*types2.Checker)(nil).NewTypeParam(tn, nil)
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
if sub == 0 {
errorf("missing subscript")
}
t.SetId(sub)
// To handle recursive references to the typeparam within its
// bound, save the partial type in tparamIndex before reading the bounds.
id := ident{r.currPkg.Name(), name}
r.p.tparamIndex[id] = t
t.SetConstraint(r.typ())
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
case 'V':
typ := r.typ()
r.declare(types2.NewVar(pos, r.currPkg, name, typ))
default:
errorf("unexpected tag: %v", tag)
}
}
func (r *importReader) declare(obj types2.Object) {
obj.Pkg().Scope().Insert(obj)
}
func (r *importReader) value() (typ types2.Type, val constant.Value) {
typ = r.typ()
switch b := typ.Underlying().(*types2.Basic); b.Info() & types2.IsConstType {
case types2.IsBoolean:
val = constant.MakeBool(r.bool())
case types2.IsString:
val = constant.MakeString(r.string())
case types2.IsInteger:
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
var x big.Int
r.mpint(&x, b)
val = constant.Make(&x)
case types2.IsFloat:
val = r.mpfloat(b)
case types2.IsComplex:
re := r.mpfloat(b)
im := r.mpfloat(b)
val = constant.BinaryOp(re, token.ADD, constant.MakeImag(im))
default:
errorf("unexpected type %v", typ) // panics
panic("unreachable")
}
return
}
func intSize(b *types2.Basic) (signed bool, maxBytes uint) {
if (b.Info() & types2.IsUntyped) != 0 {
return true, 64
}
switch b.Kind() {
case types2.Float32, types2.Complex64:
return true, 3
case types2.Float64, types2.Complex128:
return true, 7
}
signed = (b.Info() & types2.IsUnsigned) == 0
switch b.Kind() {
case types2.Int8, types2.Uint8:
maxBytes = 1
case types2.Int16, types2.Uint16:
maxBytes = 2
case types2.Int32, types2.Uint32:
maxBytes = 4
default:
maxBytes = 8
}
return
}
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
func (r *importReader) mpint(x *big.Int, typ *types2.Basic) {
signed, maxBytes := intSize(typ)
maxSmall := 256 - maxBytes
if signed {
maxSmall = 256 - 2*maxBytes
}
if maxBytes == 1 {
maxSmall = 256
}
n, _ := r.declReader.ReadByte()
if uint(n) < maxSmall {
v := int64(n)
if signed {
v >>= 1
if n&1 != 0 {
v = ^v
}
}
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
x.SetInt64(v)
return
}
v := -n
if signed {
v = -(n &^ 1) >> 1
}
if v < 1 || uint(v) > maxBytes {
errorf("weird decoding: %v, %v => %v", n, signed, v)
}
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
b := make([]byte, v)
io.ReadFull(&r.declReader, b)
x.SetBytes(b)
if signed && n&1 != 0 {
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
x.Neg(x)
}
}
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
func (r *importReader) mpfloat(typ *types2.Basic) constant.Value {
var mant big.Int
r.mpint(&mant, typ)
var f big.Float
f.SetInt(&mant)
if f.Sign() != 0 {
f.SetMantExp(&f, int(r.int64()))
}
[dev.typeparams] cmd/compile/internal/importer: adjust importer to match compiler importer The compiler chooses the literal value export format by type not by constant.Kind. That is, a floating-point constant is always exported as a (big) float value, not a (big) rational value, even though the internal representation may be that of a rational number. (This is a possibility now that the compiler also uses the go/constant package.) Naturally, during import, a floating-point value is read as a float and represented as a (big) float in go/constant. The types2 importer (based on the go/types importer) read the floating-point number elements (mantissa, exponent) but then constructed the float go/constant value through a series of elementary operations, typically leading to a rational, but sometimes even an integer number (e.g. for math.MaxFloat64). There is no problem with that (the value is the same) but if we want to impose bitsize limits on overlarge integer values we quickly run into trouble with large floats represented as integers. This change matches the code importing float literals with the code used by the compiler. Note: At some point we may want to relax the import/export code for constant values and export them by representation rather than by type. As is, we lose accuracy since all floating-point point values, even the ones internally represented as rational numbers end up being exported as floating-point numbers. Change-Id: Ic751b2046a0fd047f751da3d35cbef0a1b5fea3e Reviewed-on: https://go-review.googlesource.com/c/go/+/288632 Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2021-02-01 15:39:42 -08:00
return constant.Make(&f)
}
func (r *importReader) ident() string {
return r.string()
}
func (r *importReader) qualifiedIdent() (*types2.Package, string) {
name := r.string()
pkg := r.pkg()
return pkg, name
}
func (r *importReader) pos() syntax.Pos {
if r.p.version >= 1 {
r.posv1()
} else {
r.posv0()
}
if (r.prevPosBase == nil || r.prevPosBase.Filename() == "") && r.prevLine == 0 && r.prevColumn == 0 {
return syntax.Pos{}
}
return syntax.MakePos(r.prevPosBase, uint(r.prevLine), uint(r.prevColumn))
}
func (r *importReader) posv0() {
delta := r.int64()
if delta != deltaNewFile {
r.prevLine += delta
} else if l := r.int64(); l == -1 {
r.prevLine += deltaNewFile
} else {
r.prevPosBase = r.posBase()
r.prevLine = l
}
}
func (r *importReader) posv1() {
delta := r.int64()
r.prevColumn += delta >> 1
if delta&1 != 0 {
delta = r.int64()
r.prevLine += delta >> 1
if delta&1 != 0 {
r.prevPosBase = r.posBase()
}
}
}
func (r *importReader) typ() types2.Type {
return r.p.typAt(r.uint64(), nil)
}
func isInterface(t types2.Type) bool {
_, ok := t.(*types2.Interface)
return ok
}
func (r *importReader) pkg() *types2.Package { return r.p.pkgAt(r.uint64()) }
func (r *importReader) string() string { return r.p.stringAt(r.uint64()) }
func (r *importReader) posBase() *syntax.PosBase { return r.p.posBaseAt(r.uint64()) }
func (r *importReader) doType(base *types2.Named) types2.Type {
switch k := r.kind(); k {
default:
errorf("unexpected kind tag in %q: %v", r.p.ipath, k)
return nil
case definedType:
pkg, name := r.qualifiedIdent()
r.p.doDecl(pkg, name)
return pkg.Scope().Lookup(name).(*types2.TypeName).Type()
case pointerType:
return types2.NewPointer(r.typ())
case sliceType:
return types2.NewSlice(r.typ())
case arrayType:
n := r.uint64()
return types2.NewArray(r.typ(), int64(n))
case chanType:
dir := chanDir(int(r.uint64()))
return types2.NewChan(dir, r.typ())
case mapType:
return types2.NewMap(r.typ(), r.typ())
case signatureType:
r.currPkg = r.pkg()
return r.signature(nil)
case structType:
r.currPkg = r.pkg()
fields := make([]*types2.Var, r.uint64())
tags := make([]string, len(fields))
for i := range fields {
fpos := r.pos()
fname := r.ident()
ftyp := r.typ()
emb := r.bool()
tag := r.string()
fields[i] = types2.NewField(fpos, r.currPkg, fname, ftyp, emb)
tags[i] = tag
}
return types2.NewStruct(fields, tags)
case interfaceType:
r.currPkg = r.pkg()
embeddeds := make([]types2.Type, r.uint64())
for i := range embeddeds {
_ = r.pos()
embeddeds[i] = r.typ()
}
methods := make([]*types2.Func, r.uint64())
for i := range methods {
mpos := r.pos()
mname := r.ident()
// TODO(mdempsky): Matches bimport.go, but I
// don't agree with this.
var recv *types2.Var
if base != nil {
recv = types2.NewVar(syntax.Pos{}, r.currPkg, "", base)
}
msig := r.signature(recv)
methods[i] = types2.NewFunc(mpos, r.currPkg, mname, msig)
}
typ := types2.NewInterfaceType(methods, embeddeds)
r.p.interfaceList = append(r.p.interfaceList, typ)
return typ
[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
case typeParamType:
if r.p.exportVersion < iexportVersionGenerics {
errorf("unexpected type param type")
}
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
pkg, name := r.qualifiedIdent()
id := ident{pkg.Name(), name}
if t, ok := r.p.tparamIndex[id]; ok {
// We're already in the process of importing this typeparam.
return t
[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
}
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
// Otherwise, import the definition of the typeparam now.
r.p.doDecl(pkg, name)
return r.p.tparamIndex[id]
[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
case instType:
if r.p.exportVersion < iexportVersionGenerics {
errorf("unexpected instantiation type")
}
[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
pos := r.pos()
len := r.uint64()
targs := make([]types2.Type, len)
for i := range targs {
targs[i] = r.typ()
}
baseType := r.typ()
// The imported instantiated type doesn't include any methods, so
// we must always use the methods of the base (orig) type.
var check *types2.Checker // TODO provide a non-nil *Checker
t := check.Instantiate(pos, baseType, targs, nil, false)
[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
return t
case unionType:
if r.p.exportVersion < iexportVersionGenerics {
errorf("unexpected instantiation type")
}
terms := make([]*types2.Term, r.uint64())
for i := range terms {
terms[i] = types2.NewTerm(r.bool(), r.typ())
}
return types2.NewUnion(terms)
}
}
func (r *importReader) kind() itag {
return itag(r.uint64())
}
func (r *importReader) signature(recv *types2.Var) *types2.Signature {
params := r.paramList()
results := r.paramList()
variadic := params.Len() > 0 && r.bool()
return types2.NewSignature(recv, params, results, variadic)
}
[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
func (r *importReader) tparamList() []*types2.TypeName {
n := r.uint64()
if n == 0 {
return nil
}
xs := make([]*types2.TypeName, n)
for i := range xs {
typ := r.typ()
xs[i] = types2.AsTypeParam(typ).Obj()
}
return xs
}
func (r *importReader) paramList() *types2.Tuple {
xs := make([]*types2.Var, r.uint64())
for i := range xs {
xs[i] = r.param()
}
return types2.NewTuple(xs...)
}
func (r *importReader) param() *types2.Var {
pos := r.pos()
name := r.ident()
typ := r.typ()
return types2.NewParam(pos, r.currPkg, name, typ)
}
func (r *importReader) bool() bool {
return r.uint64() != 0
}
func (r *importReader) int64() int64 {
n, err := binary.ReadVarint(&r.declReader)
if err != nil {
errorf("readVarint: %v", err)
}
return n
}
func (r *importReader) uint64() uint64 {
n, err := binary.ReadUvarint(&r.declReader)
if err != nil {
errorf("readUvarint: %v", err)
}
return n
}
func (r *importReader) byte() byte {
x, err := r.declReader.ReadByte()
if err != nil {
errorf("declReader.ReadByte: %v", err)
}
return x
}
[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
func baseType(typ types2.Type) *types2.Named {
// pointer receivers are never types2.Named types
if p, _ := typ.(*types2.Pointer); p != nil {
typ = p.Elem()
}
// receiver base types are always (possibly generic) types2.Named types
n, _ := typ.(*types2.Named)
return n
}
[dev.typeparams] cmd/compile: export/import of recursive generic types. Deal with export/import of recursive generic types. This includes typeparams which have bounds that reference the typeparam. There are three main changes: - Change export/import of typeparams to have an implicit "declaration" (doDecl). We need to do a declaration of typeparams (via the typeparam's package and unique name), because it may be referenced within its bound during its own definition. - We delay most of the processing of the Instantiate call until we finish the creation of the top-most type (similar to the way we delay CheckSize). This is because we can't do the full instantiation properly until the base type is fully defined (with methods). The functions delayDoInst() and resumeDoInst() delay and resume the processing of the instantiations. - To do the full needed type substitutions for type instantiations during import, I had to separate out the type subster in stencil.go and move it to subr.go in the typecheck package. The subster in stencil.go now does node substitution and makes use of the type subster to do type substitutions. Notable other changes: - In types/builtins.go, put the newly defined typeparam for a union type (related to use of real/imag, etc.) in the current package, rather than the builtin package, so exports/imports work properly. - In types2, allowed NewTypeParam() to be called with a nil bound, and allow setting the bound later. (Needed to import a typeparam whose bound refers to the typeparam itself.) - During import of typeparams in types2 (importer/import.go), we need to keep an index of the typeparams by their package and unique name (with id). Use a new map typParamIndex[] for that. Again, this is needed to deal with typeparams whose bounds refer to the typeparam itself. - Added several new tests absdiffimp.go and orderedmapsimp.go. Some of the orderemapsimp tests are commented out for now, because there are some issues with closures inside instantiations (relating to unexported names of closure structs). - Renamed some typeparams in test value.go to make them all T (to make typeparam uniqueness is working fine). Change-Id: Ib47ed9471c19ee8e9fbb34e8506907dad3021e5a Reviewed-on: https://go-review.googlesource.com/c/go/+/323029 Trust: Dan Scales <danscales@google.com> Trust: Robert Griesemer <gri@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
2021-05-17 15:00:39 -07:00
func parseSubscript(name string) (string, uint64) {
// Extract the subscript value from the type param name. We export
// and import the subscript value, so that all type params have
// unique names.
sub := uint64(0)
startsub := -1
for i, r := range name {
if '₀' <= r && r < '₀'+10 {
if startsub == -1 {
startsub = i
}
sub = sub*10 + uint64(r-'₀')
}
}
if startsub >= 0 {
name = name[:startsub]
}
return name, sub
}