maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/lib/Sema/SemaTemplateInstantiate.cpp

2824 lines
116 KiB
C++
Raw Normal View History

Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
//===------- SemaTemplateInstantiate.cpp - C++ Template Instantiation ------===/
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//===----------------------------------------------------------------------===/
//
// This file implements C++ template instantiation.
//
//===----------------------------------------------------------------------===/
Split out a header to hold APIs meant for the Sema implementation from Sema.h. Clients of Sema don't need to know (for example) the list of diagnostics we support. llvm-svn: 112093
2010-08-26 06:03:47 +08:00
#include "clang/Sema/SemaInternal.h"
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
#include "TreeTransform.h"
Sort all of Clang's files under 'lib', and fix up the broken headers uncovered. This required manually correcting all of the incorrect main-module headers I could find, and running the new llvm/utils/sort_includes.py script over the files. I also manually added quite a few missing headers that were uncovered by shuffling the order or moving headers up to be main-module-headers. llvm-svn: 169237
2012-12-04 17:13:33 +08:00
#include "clang/AST/ASTConsumer.h"
#include "clang/AST/ASTContext.h"
#include "clang/AST/DeclTemplate.h"
#include "clang/AST/Expr.h"
#include "clang/Basic/LangOptions.h"
Another step in the process of making the parser depend on Sema: - move DeclSpec &c into the Sema library - move ParseAST into the Parse library Reflect this change in a thousand different includes. Reflect this change in the link orders. llvm-svn: 111667
2010-08-21 02:27:03 +08:00
#include "clang/Sema/DeclSpec.h"
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
#include "clang/Sema/Initialization.h"
Move Sema's headers into include/clang/Sema, renaming a few along the way. llvm-svn: 110945
2010-08-13 04:07:10 +08:00
#include "clang/Sema/Lookup.h"
Struggle mightily against header inclusion in Sema.h. llvm-svn: 111904
2010-08-24 15:21:54 +08:00
#include "clang/Sema/Template.h"
Move more stuff out of Sema.h. llvm-svn: 112026
2010-08-25 13:32:35 +08:00
#include "clang/Sema/TemplateDeduction.h"
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
using namespace clang;
Move more stuff out of Sema.h. llvm-svn: 112026
2010-08-25 13:32:35 +08:00
using namespace sema;
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
Add a notion of "post-diagnostic hooks", which are callbacks attached to a diagnostic that will be invoked after the diagnostic (if it is not suppressed). The hooks are allowed to produce additional diagnostics (typically notes) that provide more information. We should be able to use this to help diagnostic clients link notes back to the diagnostic they clarify. Comments welcome; I'll write up documentation and convert other clients (e.g., overload resolution failures) if there are no screams of protest. As the first client of post-diagnostic hooks, we now produce a template instantiation backtrace when a failure occurs during template instantiation. There's still more work to do to make this output pretty, if that's even possible. llvm-svn: 66557
2009-03-11 02:03:33 +08:00
//===----------------------------------------------------------------------===/
// Template Instantiation Support
//===----------------------------------------------------------------------===/
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
/// \brief Retrieve the template argument list(s) that should be used to
/// instantiate the definition of the given declaration.
Improve instantiation of default template arguments for nested templates. The instantiation of these default arguments must be (and now, is) delayed until the template argument is actually used, at which point we substitute all levels of template arguments concurrently. llvm-svn: 86578
2009-11-10 03:17:50 +08:00
///
/// \param D the declaration for which we are computing template instantiation
/// arguments.
///
/// \param Innermost if non-NULL, the innermost template argument list.
Default function arguments for function template specializations always come from the primary template, so gather the instantiation template arguments from the primary template. llvm-svn: 95380
2010-02-05 15:33:43 +08:00
///
/// \param RelativeToPrimary true if we should get the template
/// arguments relative to the primary template, even when we're
/// dealing with a specialization. This is only relevant for function
/// template specializations.
When computing the template arguments for the instantiation of a friend function template, be sure to adjust the computed template argument lists based on the location of the definition of the function template: it's possible that the definition we're instantiating with and the template declaration that we found when creating the specialization are in different contexts, which meant that we would end up using the wrong template arguments for instantiation. Fixes PR7013; all Boost.DynamicBitset tests now pass. llvm-svn: 102974
2010-05-04 07:29:10 +08:00
///
/// \param Pattern If non-NULL, indicates the pattern from which we will be
/// instantiating the definition of the given declaration, \p D. This is
/// used to determine the proper set of template instantiation arguments for
/// friend function template specializations.
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
MultiLevelTemplateArgumentList
Improve instantiation of default template arguments for nested templates. The instantiation of these default arguments must be (and now, is) delayed until the template argument is actually used, at which point we substitute all levels of template arguments concurrently. llvm-svn: 86578
2009-11-10 03:17:50 +08:00
Sema::getTemplateInstantiationArgs(NamedDecl *D,
Default function arguments for function template specializations always come from the primary template, so gather the instantiation template arguments from the primary template. llvm-svn: 95380
2010-02-05 15:33:43 +08:00
const TemplateArgumentList *Innermost,
When computing the template arguments for the instantiation of a friend function template, be sure to adjust the computed template argument lists based on the location of the definition of the function template: it's possible that the definition we're instantiating with and the template declaration that we found when creating the specialization are in different contexts, which meant that we would end up using the wrong template arguments for instantiation. Fixes PR7013; all Boost.DynamicBitset tests now pass. llvm-svn: 102974
2010-05-04 07:29:10 +08:00
bool RelativeToPrimary,
const FunctionDecl *Pattern) {
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
// Accumulate the set of template argument lists in this structure.
MultiLevelTemplateArgumentList Result;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Improve instantiation of default template arguments for nested templates. The instantiation of these default arguments must be (and now, is) delayed until the template argument is actually used, at which point we substitute all levels of template arguments concurrently. llvm-svn: 86578
2009-11-10 03:17:50 +08:00
if (Innermost)
Result.addOuterTemplateArguments(Innermost);
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
DeclContext *Ctx = dyn_cast<DeclContext>(D);
Add an assertion to tell us when we're going off the deep end trying to determine outer template arguments lists for template parameters. This is actually the problem behind PR9643, which I have yet to figure out how to fix. llvm-svn: 131822
2011-05-22 08:21:10 +08:00
if (!Ctx) {
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
Ctx = D->getDeclContext();
Started implementing variable templates. Top level declarations should be fully supported, up to some limitations documented as FIXMEs or TODO. Static data member templates work very partially. Static data member templates of class templates need particular attention... llvm-svn: 187762
2013-08-06 09:03:05 +08:00
// Add template arguments from a variable template instantiation.
if (VarTemplateSpecializationDecl *Spec =
dyn_cast<VarTemplateSpecializationDecl>(D)) {
// We're done when we hit an explicit specialization.
if (Spec->getSpecializationKind() == TSK_ExplicitSpecialization &&
!isa<VarTemplatePartialSpecializationDecl>(Spec))
return Result;
Result.addOuterTemplateArguments(&Spec->getTemplateInstantiationArgs());
// If this variable template specialization was instantiated from a
// specialized member that is a variable template, we're done.
assert(Spec->getSpecializedTemplate() && "No variable template?");
if (Spec->getSpecializedTemplate()->isMemberSpecialization())
return Result;
}
When performing substitution of default template template parameters before the template parameters have acquired a proper context (e.g., because the enclosing context has yet to be built), provide empty parameter lists for all outer template parameter scopes to inhibit any substitution for those template parameters. Fixes PR9643 / <rdar://problem/9251019>. llvm-svn: 133055
2011-06-15 22:20:42 +08:00
// If we have a template template parameter with translation unit context,
// then we're performing substitution into a default template argument of
// this template template parameter before we've constructed the template
// that will own this template template parameter. In this case, we
// use empty template parameter lists for all of the outer templates
// to avoid performing any substitutions.
if (Ctx->isTranslationUnit()) {
if (TemplateTemplateParmDecl *TTP
= dyn_cast<TemplateTemplateParmDecl>(D)) {
for (unsigned I = 0, N = TTP->getDepth() + 1; I != N; ++I)
A little ArrayRef'ization. llvm-svn: 182074
2013-05-17 11:04:50 +08:00
Result.addOuterTemplateArguments(None);
When performing substitution of default template template parameters before the template parameters have acquired a proper context (e.g., because the enclosing context has yet to be built), provide empty parameter lists for all outer template parameter scopes to inhibit any substitution for those template parameters. Fixes PR9643 / <rdar://problem/9251019>. llvm-svn: 133055
2011-06-15 22:20:42 +08:00
return Result;
}
}
Add an assertion to tell us when we're going off the deep end trying to determine outer template arguments lists for template parameters. This is actually the problem behind PR9643, which I have yet to figure out how to fix. llvm-svn: 131822
2011-05-22 08:21:10 +08:00
}
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
while (!Ctx->isFileContext()) {
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
// Add template arguments from a class template instantiation.
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
if (ClassTemplateSpecializationDecl *Spec
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
= dyn_cast<ClassTemplateSpecializationDecl>(Ctx)) {
// We're done when we hit an explicit specialization.
When performing substitution of template arguments within the body of a template, be sure to include the template arguments from the injected-class-name. Fixes PR7587. llvm-svn: 107895
2010-07-09 02:37:38 +08:00
if (Spec->getSpecializationKind() == TSK_ExplicitSpecialization &&
!isa<ClassTemplatePartialSpecializationDecl>(Spec))
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
break;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
Result.addOuterTemplateArguments(&Spec->getTemplateInstantiationArgs());
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
// If this class template specialization was instantiated from a
// specialized member that is a class template, we're done.
assert(Spec->getSpecializedTemplate() && "No class template?");
if (Spec->getSpecializedTemplate()->isMemberSpecialization())
break;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
}
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
// Add template arguments from a function template specialization.
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
else if (FunctionDecl *Function = dyn_cast<FunctionDecl>(Ctx)) {
Default function arguments for function template specializations always come from the primary template, so gather the instantiation template arguments from the primary template. llvm-svn: 95380
2010-02-05 15:33:43 +08:00
if (!RelativeToPrimary &&
Implement function template specialization at class scope extension in Microsoft mode. A new AST node is introduced: ClassScopeFunctionSpecialization. This node holds a FunctionDecl that is not yet specialized; then during the class template instantiation the ClassScopeFunctionSpecialization will spawn the actual function specialization. Example: template <class T> class A { public: template <class U> void f(U p) { } template <> void f(int p) { } // <== class scope specialization }; This extension is necessary to parse MSVC standard C++ headers, MFC and ATL code. BTW, with this feature in, clang can parse (-fsyntax-only) all the MSVC 2010 standard header files without any error. llvm-svn: 137573
2011-08-14 11:52:19 +08:00
(Function->getTemplateSpecializationKind() ==
TSK_ExplicitSpecialization &&
!Function->getClassScopeSpecializationPattern()))
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
break;
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
if (const TemplateArgumentList *TemplateArgs
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
= Function->getTemplateSpecializationArgs()) {
// Add the template arguments for this specialization.
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
Result.addOuterTemplateArguments(TemplateArgs);
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
// If this function was instantiated from a specialized member that is
// a function template, we're done.
assert(Function->getPrimaryTemplate() && "No function template?");
if (Function->getPrimaryTemplate()->isMemberSpecialization())
break;
When determining template instantiation arguments within a function template (not a specialization!), use the "injected" function template arguments, which correspond to the template parameters of the function template. This is required when substituting into the default template parameters of template template parameters within a function template. Fixes PR9016. llvm-svn: 127092
2011-03-06 01:54:25 +08:00
} else if (FunctionTemplateDecl *FunTmpl
= Function->getDescribedFunctionTemplate()) {
// Add the "injected" template arguments.
A little ArrayRef'ization. llvm-svn: 182074
2013-05-17 11:04:50 +08:00
Result.addOuterTemplateArguments(FunTmpl->getInjectedTemplateArgs());
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
}
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
// If this is a friend declaration and it declares an entity at
// namespace scope, take arguments from its lexical parent
When computing the template arguments for the instantiation of a friend function template, be sure to adjust the computed template argument lists based on the location of the definition of the function template: it's possible that the definition we're instantiating with and the template declaration that we found when creating the specialization are in different contexts, which meant that we would end up using the wrong template arguments for instantiation. Fixes PR7013; all Boost.DynamicBitset tests now pass. llvm-svn: 102974
2010-05-04 07:29:10 +08:00
// instead of its semantic parent, unless of course the pattern we're
// instantiating actually comes from the file's context!
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
if (Function->getFriendObjectKind() &&
When computing the template arguments for the instantiation of a friend function template, be sure to adjust the computed template argument lists based on the location of the definition of the function template: it's possible that the definition we're instantiating with and the template declaration that we found when creating the specialization are in different contexts, which meant that we would end up using the wrong template arguments for instantiation. Fixes PR7013; all Boost.DynamicBitset tests now pass. llvm-svn: 102974
2010-05-04 07:29:10 +08:00
Function->getDeclContext()->isFileContext() &&
(!Pattern || !Pattern->getLexicalDeclContext()->isFileContext())) {
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
Ctx = Function->getLexicalDeclContext();
Default function arguments for function template specializations always come from the primary template, so gather the instantiation template arguments from the primary template. llvm-svn: 95380
2010-02-05 15:33:43 +08:00
RelativeToPrimary = false;
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
continue;
When performing substitution of template arguments within the body of a template, be sure to include the template arguments from the injected-class-name. Fixes PR7587. llvm-svn: 107895
2010-07-09 02:37:38 +08:00
}
} else if (CXXRecordDecl *Rec = dyn_cast<CXXRecordDecl>(Ctx)) {
if (ClassTemplateDecl *ClassTemplate = Rec->getDescribedClassTemplate()) {
QualType T = ClassTemplate->getInjectedClassNameSpecialization();
A little ArrayRef'ization. llvm-svn: 182074
2013-05-17 11:04:50 +08:00
const TemplateSpecializationType *TST =
cast<TemplateSpecializationType>(Context.getCanonicalType(T));
Result.addOuterTemplateArguments(
llvm::makeArrayRef(TST->getArgs(), TST->getNumArgs()));
When performing substitution of template arguments within the body of a template, be sure to include the template arguments from the injected-class-name. Fixes PR7587. llvm-svn: 107895
2010-07-09 02:37:38 +08:00
if (ClassTemplate->isMemberSpecialization())
break;
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
}
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
}
Ensure code generation for friend declarations in class templates. llvm-svn: 80418
2009-08-29 11:16:09 +08:00
Ctx = Ctx->getParent();
Default function arguments for function template specializations always come from the primary template, so gather the instantiation template arguments from the primary template. llvm-svn: 95380
2010-02-05 15:33:43 +08:00
RelativeToPrimary = false;
Introduce basic support for instantiating the definitions of member functions of class templates. Only compound statements and expression statements are currently implemented. llvm-svn: 71814
2009-05-15 07:26:13 +08:00
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Collect multiple levels of template arguments into a new type, MultiLevelTemplateArgumentList. This is a baby step toward instantiating member templates; no intended functionality change yet. llvm-svn: 80380
2009-08-29 01:37:35 +08:00
return Result;
Introduce basic support for instantiating the definitions of member functions of class templates. Only compound statements and expression statements are currently implemented. llvm-svn: 71814
2009-05-15 07:26:13 +08:00
}
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
bool Sema::ActiveTemplateInstantiation::isInstantiationRecord() const {
switch (Kind) {
case TemplateInstantiation:
Implement DR1330 in C++11 mode, to support libstdc++4.7 which uses it. We have a new flavor of exception specification, EST_Uninstantiated. A function type with this exception specification carries a pointer to a FunctionDecl, and the exception specification for that FunctionDecl is instantiated (if needed) and used in the place of the function type's exception specification. When a function template declaration with a non-trivial exception specification is instantiated, the specialization's exception specification is set to this new 'uninstantiated' kind rather than being instantiated immediately. Expr::CanThrow has migrated onto Sema, so it can instantiate exception specs on-demand. Also, any odr-use of a function triggers the instantiation of its exception specification (the exception specification could be needed by IRGen). In passing, fix two places where a DeclRefExpr was created but the corresponding function was not actually marked odr-used. We used to get away with this, but don't any more. Also fix a bug where instantiating an exception specification which refers to function parameters resulted in a crash. We still have the same bug in default arguments, which I'll be looking into next. This, plus a tiny patch to fix libstdc++'s common_type, is enough for clang to parse (and, in very limited testing, support) all of libstdc++4.7's standard headers. llvm-svn: 154886
2012-04-17 08:58:00 +08:00
case ExceptionSpecInstantiation:
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
case DefaultTemplateArgumentInstantiation:
case DefaultFunctionArgumentInstantiation:
case ExplicitTemplateArgumentSubstitution:
case DeducedTemplateArgumentSubstitution:
case PriorTemplateArgumentSubstitution:
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
return true;
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
case DefaultTemplateArgumentChecking:
return false;
}
Remove unreachable code in Clang. (replace with llvm_unreachable where appropriate or when GCC requires it) llvm-svn: 148292
2012-01-17 14:56:22 +08:00
llvm_unreachable("Invalid InstantiationKind!");
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
}
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
Decl *Entity,
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation,
InstantiationRange);
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind = ActiveTemplateInstantiation::TemplateInstantiation;
Inst.PointOfInstantiation = PointOfInstantiation;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Entity;
Use StmtVisitor to handle the decoding of expressions for instantiation. This is roughly the structure we want to expression instantiation. llvm-svn: 66816
2009-03-13 02:36:18 +08:00
Inst.TemplateArgs = 0;
Inst.NumTemplateArgs = 0;
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
Inst.InstantiationRange = InstantiationRange;
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
SemaRef.InNonInstantiationSFINAEContext = false;
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
}
Implement DR1330 in C++11 mode, to support libstdc++4.7 which uses it. We have a new flavor of exception specification, EST_Uninstantiated. A function type with this exception specification carries a pointer to a FunctionDecl, and the exception specification for that FunctionDecl is instantiated (if needed) and used in the place of the function type's exception specification. When a function template declaration with a non-trivial exception specification is instantiated, the specialization's exception specification is set to this new 'uninstantiated' kind rather than being instantiated immediately. Expr::CanThrow has migrated onto Sema, so it can instantiate exception specs on-demand. Also, any odr-use of a function triggers the instantiation of its exception specification (the exception specification could be needed by IRGen). In passing, fix two places where a DeclRefExpr was created but the corresponding function was not actually marked odr-used. We used to get away with this, but don't any more. Also fix a bug where instantiating an exception specification which refers to function parameters resulted in a crash. We still have the same bug in default arguments, which I'll be looking into next. This, plus a tiny patch to fix libstdc++'s common_type, is enough for clang to parse (and, in very limited testing, support) all of libstdc++4.7's standard headers. llvm-svn: 154886
2012-04-17 08:58:00 +08:00
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
FunctionDecl *Entity, ExceptionSpecification,
SourceRange InstantiationRange)
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
Invalid = CheckInstantiationDepth(PointOfInstantiation,
InstantiationRange);
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind = ActiveTemplateInstantiation::ExceptionSpecInstantiation;
Inst.PointOfInstantiation = PointOfInstantiation;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Entity;
Implement DR1330 in C++11 mode, to support libstdc++4.7 which uses it. We have a new flavor of exception specification, EST_Uninstantiated. A function type with this exception specification carries a pointer to a FunctionDecl, and the exception specification for that FunctionDecl is instantiated (if needed) and used in the place of the function type's exception specification. When a function template declaration with a non-trivial exception specification is instantiated, the specialization's exception specification is set to this new 'uninstantiated' kind rather than being instantiated immediately. Expr::CanThrow has migrated onto Sema, so it can instantiate exception specs on-demand. Also, any odr-use of a function triggers the instantiation of its exception specification (the exception specification could be needed by IRGen). In passing, fix two places where a DeclRefExpr was created but the corresponding function was not actually marked odr-used. We used to get away with this, but don't any more. Also fix a bug where instantiating an exception specification which refers to function parameters resulted in a crash. We still have the same bug in default arguments, which I'll be looking into next. This, plus a tiny patch to fix libstdc++'s common_type, is enough for clang to parse (and, in very limited testing, support) all of libstdc++4.7's standard headers. llvm-svn: 154886
2012-04-17 08:58:00 +08:00
Inst.TemplateArgs = 0;
Inst.NumTemplateArgs = 0;
Inst.InstantiationRange = InstantiationRange;
SemaRef.InNonInstantiationSFINAEContext = false;
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
}
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
TemplateDecl *Template,
ArrayRef<TemplateArgument> TemplateArgs,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation,
InstantiationRange);
if (!Invalid) {
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
ActiveTemplateInstantiation Inst;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Inst.Kind
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
= ActiveTemplateInstantiation::DefaultTemplateArgumentInstantiation;
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
Inst.PointOfInstantiation = PointOfInstantiation;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Template;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
Inst.InstantiationRange = InstantiationRange;
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
SemaRef.InNonInstantiationSFINAEContext = false;
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
}
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
FunctionTemplateDecl *FunctionTemplate,
ArrayRef<TemplateArgument> TemplateArgs,
ActiveTemplateInstantiation::InstantiationKind Kind,
sema::TemplateDeductionInfo &DeductionInfo,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation, InstantiationRange);
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind = Kind;
Inst.PointOfInstantiation = PointOfInstantiation;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = FunctionTemplate;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
Inst.DeductionInfo = &DeductionInfo;
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
Inst.InstantiationRange = InstantiationRange;
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
SemaRef.InNonInstantiationSFINAEContext = false;
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
if (!Inst.isInstantiationRecord())
++SemaRef.NonInstantiationEntries;
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
}
}
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
ClassTemplatePartialSpecializationDecl *PartialSpec,
ArrayRef<TemplateArgument> TemplateArgs,
sema::TemplateDeductionInfo &DeductionInfo,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation, InstantiationRange);
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind = ActiveTemplateInstantiation::DeducedTemplateArgumentSubstitution;
Started implementing variable templates. Top level declarations should be fully supported, up to some limitations documented as FIXMEs or TODO. Static data member templates work very partially. Static data member templates of class templates need particular attention... llvm-svn: 187762
2013-08-06 09:03:05 +08:00
Inst.PointOfInstantiation = PointOfInstantiation;
Inst.Entity = PartialSpec;
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
Inst.DeductionInfo = &DeductionInfo;
Inst.InstantiationRange = InstantiationRange;
SemaRef.InNonInstantiationSFINAEContext = false;
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
}
Sema::InstantiatingTemplate::InstantiatingTemplate(
Sema &SemaRef, SourceLocation PointOfInstantiation,
VarTemplatePartialSpecializationDecl *PartialSpec,
ArrayRef<TemplateArgument> TemplateArgs,
sema::TemplateDeductionInfo &DeductionInfo, SourceRange InstantiationRange)
: SemaRef(SemaRef), SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext) {
Invalid = CheckInstantiationDepth(PointOfInstantiation, InstantiationRange);
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind =
ActiveTemplateInstantiation::DeducedTemplateArgumentSubstitution;
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Inst.PointOfInstantiation = PointOfInstantiation;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = PartialSpec;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Inst.DeductionInfo = &DeductionInfo;
Inst.InstantiationRange = InstantiationRange;
SemaRef.InNonInstantiationSFINAEContext = false;
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
Template argument deduction for member pointers. Also, introduced some of the framework for performing instantiation as part of template argument deduction. llvm-svn: 73175
2009-06-11 07:47:09 +08:00
}
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
ParmVarDecl *Param,
ArrayRef<TemplateArgument> TemplateArgs,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation, InstantiationRange);
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind
= ActiveTemplateInstantiation::DefaultFunctionArgumentInstantiation;
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
Inst.PointOfInstantiation = PointOfInstantiation;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Param;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
Inst.InstantiationRange = InstantiationRange;
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
SemaRef.InNonInstantiationSFINAEContext = false;
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
}
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
NamedDecl *Template, NonTypeTemplateParmDecl *Param,
ArrayRef<TemplateArgument> TemplateArgs,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation, InstantiationRange);
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind = ActiveTemplateInstantiation::PriorTemplateArgumentSubstitution;
Inst.PointOfInstantiation = PointOfInstantiation;
Inst.Template = Template;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Param;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Inst.InstantiationRange = InstantiationRange;
SemaRef.InNonInstantiationSFINAEContext = false;
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
}
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
NamedDecl *Template, TemplateTemplateParmDecl *Param,
ArrayRef<TemplateArgument> TemplateArgs,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Invalid = CheckInstantiationDepth(PointOfInstantiation, InstantiationRange);
if (!Invalid) {
ActiveTemplateInstantiation Inst;
Inst.Kind = ActiveTemplateInstantiation::PriorTemplateArgumentSubstitution;
Inst.PointOfInstantiation = PointOfInstantiation;
Inst.Template = Template;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Param;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
PR9793: Treat substitution as an instantiation step for the purpose of the -ftemplate-depth limit. There are various ways to get an infinite (or merely huge) stack of substitutions with no intervening instantiations. This is also consistent with gcc's behavior. llvm-svn: 159907
2012-07-08 10:38:24 +08:00
Inst.InstantiationRange = InstantiationRange;
SemaRef.InNonInstantiationSFINAEContext = false;
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
}
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
}
Sema::InstantiatingTemplate::
InstantiatingTemplate(Sema &SemaRef, SourceLocation PointOfInstantiation,
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
TemplateDecl *Template, NamedDecl *Param,
ArrayRef<TemplateArgument> TemplateArgs,
SourceRange InstantiationRange)
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
: SemaRef(SemaRef),
SavedInNonInstantiationSFINAEContext(
SemaRef.InNonInstantiationSFINAEContext)
{
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
Invalid = false;
ActiveTemplateInstantiation Inst;
Inst.Kind = ActiveTemplateInstantiation::DefaultTemplateArgumentChecking;
Inst.PointOfInstantiation = PointOfInstantiation;
Inst.Template = Template;
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Inst.Entity = Param;
PR13365: Fix code which was trying to treat an array of DeducedTemplateArgument as an array of its base class TemplateArgument. Switch the const TemplateArgument* parameters of InstantiatingTemplate's constructors to ArrayRef<TemplateArgument> to prevent this from happening again in the future. llvm-svn: 160245
2012-07-16 09:09:10 +08:00
Inst.TemplateArgs = TemplateArgs.data();
Inst.NumTemplateArgs = TemplateArgs.size();
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
Inst.InstantiationRange = InstantiationRange;
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
SemaRef.InNonInstantiationSFINAEContext = false;
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
SemaRef.ActiveTemplateInstantiations.push_back(Inst);
assert(!Inst.isInstantiationRecord());
++SemaRef.NonInstantiationEntries;
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
}
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
void Sema::InstantiatingTemplate::Clear() {
if (!Invalid) {
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
if (!SemaRef.ActiveTemplateInstantiations.back().isInstantiationRecord()) {
assert(SemaRef.NonInstantiationEntries > 0);
--SemaRef.NonInstantiationEntries;
}
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
SemaRef.InNonInstantiationSFINAEContext
= SavedInNonInstantiationSFINAEContext;
When we perform dependent name lookup during template instantiation, it's not sufficient to only consider names visible at the point of instantiation, because that may not include names that were visible when the template was defined. More generally, if the instantiation backtrace goes through a module M, then every declaration visible within M should be available to the instantiation. Any of those declarations might be part of the interface that M intended to export to a template that it instantiates. The fix here has two parts: 1) If we find a non-visible declaration during name lookup during template instantiation, check whether the declaration was visible from the defining module of all entities on the active template instantiation stack. The defining module is not the owning module in all cases: we look at the module in which a template was defined, not the module in which it was first instantiated. 2) Perform pending instantiations at the end of a module, not at the end of the translation unit. This is general goodness, since it significantly cuts down the amount of redundant work that is performed in every TU importing a module, and also implicitly adds the module containing the point of instantiation to the set of modules checked for declarations in a lookup within a template instantiation. There's a known issue here with template instantiations performed while building a module, if additional imports are added later on. I'll fix that in a subsequent commit. llvm-svn: 187167
2013-07-26 07:08:39 +08:00
// Name lookup no longer looks in this template's defining module.
assert(SemaRef.ActiveTemplateInstantiations.size() >=
SemaRef.ActiveTemplateInstantiationLookupModules.size() &&
"forgot to remove a lookup module for a template instantiation");
if (SemaRef.ActiveTemplateInstantiations.size() ==
SemaRef.ActiveTemplateInstantiationLookupModules.size()) {
if (Module *M = SemaRef.ActiveTemplateInstantiationLookupModules.back())
SemaRef.LookupModulesCache.erase(M);
SemaRef.ActiveTemplateInstantiationLookupModules.pop_back();
}
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
SemaRef.ActiveTemplateInstantiations.pop_back();
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
Invalid = true;
}
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
}
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
bool Sema::InstantiatingTemplate::CheckInstantiationDepth(
SourceLocation PointOfInstantiation,
SourceRange InstantiationRange) {
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
assert(SemaRef.NonInstantiationEntries <=
SemaRef.ActiveTemplateInstantiations.size());
if ((SemaRef.ActiveTemplateInstantiations.size() -
SemaRef.NonInstantiationEntries)
Unify naming of LangOptions variable/get function across the Clang stack (Lex to AST). The member variable is always "LangOpts" and the member function is always "getLangOpts". Reviewed by Chris Lattner llvm-svn: 152536
2012-03-11 15:00:24 +08:00
<= SemaRef.getLangOpts().InstantiationDepth)
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
return false;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
SemaRef.Diag(PointOfInstantiation,
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
diag::err_template_recursion_depth_exceeded)
Unify naming of LangOptions variable/get function across the Clang stack (Lex to AST). The member variable is always "LangOpts" and the member function is always "getLangOpts". Reviewed by Chris Lattner llvm-svn: 152536
2012-03-11 15:00:24 +08:00
<< SemaRef.getLangOpts().InstantiationDepth
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
<< InstantiationRange;
SemaRef.Diag(PointOfInstantiation, diag::note_template_recursion_depth)
Unify naming of LangOptions variable/get function across the Clang stack (Lex to AST). The member variable is always "LangOpts" and the member function is always "getLangOpts". Reviewed by Chris Lattner llvm-svn: 152536
2012-03-11 15:00:24 +08:00
<< SemaRef.getLangOpts().InstantiationDepth;
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
return true;
}
Add a notion of "post-diagnostic hooks", which are callbacks attached to a diagnostic that will be invoked after the diagnostic (if it is not suppressed). The hooks are allowed to produce additional diagnostics (typically notes) that provide more information. We should be able to use this to help diagnostic clients link notes back to the diagnostic they clarify. Comments welcome; I'll write up documentation and convert other clients (e.g., overload resolution failures) if there are no screams of protest. As the first client of post-diagnostic hooks, we now produce a template instantiation backtrace when a failure occurs during template instantiation. There's still more work to do to make this output pretty, if that's even possible. llvm-svn: 66557
2009-03-11 02:03:33 +08:00
/// \brief Prints the current instantiation stack through a series of
/// notes.
void Sema::PrintInstantiationStack() {
Introduce a limit on the depth of the template instantiation backtrace we will print with each error that occurs during template instantiation. When the backtrace is longer than that, we will print N/2 of the innermost backtrace entries and N/2 of the outermost backtrace entries, then skip the middle entries with a note such as: note: suppressed 2 template instantiation contexts; use -ftemplate-backtrace-limit=N to change the number of template instantiation entries shown This should eliminate some excessively long backtraces that aren't providing any value. llvm-svn: 101882
2010-04-20 15:18:24 +08:00
// Determine which template instantiations to skip, if any.
unsigned SkipStart = ActiveTemplateInstantiations.size(), SkipEnd = SkipStart;
unsigned Limit = Diags.getTemplateBacktraceLimit();
if (Limit && Limit < ActiveTemplateInstantiations.size()) {
SkipStart = Limit / 2 + Limit % 2;
SkipEnd = ActiveTemplateInstantiations.size() - Limit / 2;
}
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
// FIXME: In all of these cases, we need to show the template arguments
Introduce a limit on the depth of the template instantiation backtrace we will print with each error that occurs during template instantiation. When the backtrace is longer than that, we will print N/2 of the innermost backtrace entries and N/2 of the outermost backtrace entries, then skip the middle entries with a note such as: note: suppressed 2 template instantiation contexts; use -ftemplate-backtrace-limit=N to change the number of template instantiation entries shown This should eliminate some excessively long backtraces that aren't providing any value. llvm-svn: 101882
2010-04-20 15:18:24 +08:00
unsigned InstantiationIdx = 0;
Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily. llvm-svn: 185610
2013-07-04 11:08:24 +08:00
for (SmallVectorImpl<ActiveTemplateInstantiation>::reverse_iterator
Add a notion of "post-diagnostic hooks", which are callbacks attached to a diagnostic that will be invoked after the diagnostic (if it is not suppressed). The hooks are allowed to produce additional diagnostics (typically notes) that provide more information. We should be able to use this to help diagnostic clients link notes back to the diagnostic they clarify. Comments welcome; I'll write up documentation and convert other clients (e.g., overload resolution failures) if there are no screams of protest. As the first client of post-diagnostic hooks, we now produce a template instantiation backtrace when a failure occurs during template instantiation. There's still more work to do to make this output pretty, if that's even possible. llvm-svn: 66557
2009-03-11 02:03:33 +08:00
Active = ActiveTemplateInstantiations.rbegin(),
ActiveEnd = ActiveTemplateInstantiations.rend();
Active != ActiveEnd;
Introduce a limit on the depth of the template instantiation backtrace we will print with each error that occurs during template instantiation. When the backtrace is longer than that, we will print N/2 of the innermost backtrace entries and N/2 of the outermost backtrace entries, then skip the middle entries with a note such as: note: suppressed 2 template instantiation contexts; use -ftemplate-backtrace-limit=N to change the number of template instantiation entries shown This should eliminate some excessively long backtraces that aren't providing any value. llvm-svn: 101882
2010-04-20 15:18:24 +08:00
++Active, ++InstantiationIdx) {
// Skip this instantiation?
if (InstantiationIdx >= SkipStart && InstantiationIdx < SkipEnd) {
if (InstantiationIdx == SkipStart) {
// Note that we're skipping instantiations.
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Introduce a limit on the depth of the template instantiation backtrace we will print with each error that occurs during template instantiation. When the backtrace is longer than that, we will print N/2 of the innermost backtrace entries and N/2 of the outermost backtrace entries, then skip the middle entries with a note such as: note: suppressed 2 template instantiation contexts; use -ftemplate-backtrace-limit=N to change the number of template instantiation entries shown This should eliminate some excessively long backtraces that aren't providing any value. llvm-svn: 101882
2010-04-20 15:18:24 +08:00
diag::note_instantiation_contexts_suppressed)
<< unsigned(ActiveTemplateInstantiations.size() - Limit);
}
continue;
}
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
switch (Active->Kind) {
case ActiveTemplateInstantiation::TemplateInstantiation: {
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
Decl *D = Active->Entity;
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
if (CXXRecordDecl *Record = dyn_cast<CXXRecordDecl>(D)) {
unsigned DiagID = diag::note_template_member_class_here;
if (isa<ClassTemplateSpecializationDecl>(Record))
DiagID = diag::note_template_class_instantiation_here;
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation, DiagID)
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
<< Context.getTypeDeclType(Record)
<< Active->InstantiationRange;
Template instantiation for static data members that are defined out-of-line. Note that this also fixes a bug that affects non-template code, where we were not treating out-of-line static data members are "file-scope" variables, and therefore not checking their initializers. llvm-svn: 77002
2009-07-25 04:34:43 +08:00
} else if (FunctionDecl *Function = dyn_cast<FunctionDecl>(D)) {
Implicit instantiation for function template specializations. For a FunctionDecl that has been instantiated due to template argument deduction, we now store the primary template from which it was instantiated and the deduced template arguments. From this information, we can instantiate the body of the function template. llvm-svn: 74232
2009-06-26 08:10:03 +08:00
unsigned DiagID;
if (Function->getPrimaryTemplate())
DiagID = diag::note_function_template_spec_here;
else
DiagID = diag::note_template_member_function_here;
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation, DiagID)
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
<< Function
<< Active->InstantiationRange;
Implement support for C++0x alias templates. llvm-svn: 130953
2011-05-06 05:57:07 +08:00
} else if (VarDecl *VD = dyn_cast<VarDecl>(D)) {
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Bug fix: note diagnosis on expression narrowing should say "variable template" instead of "static data member" when appropriate llvm-svn: 188409
2013-08-15 04:15:02 +08:00
VD->isStaticDataMember()?
diag::note_template_static_data_member_def_here
: diag::note_template_variable_def_here)
Implement support for C++0x alias templates. llvm-svn: 130953
2011-05-06 05:57:07 +08:00
<< VD
<< Active->InstantiationRange;
Instantiating a class template should not instantiate the definition of any scoped enumeration members. Later uses of an enumeration temploid as a nested name specifier should cause its instantiation. Plus some groundwork for explicit specialization of member enumerations of class templates. llvm-svn: 152750
2012-03-15 07:13:10 +08:00
} else if (EnumDecl *ED = dyn_cast<EnumDecl>(D)) {
Diags.Report(Active->PointOfInstantiation,
diag::note_template_enum_def_here)
<< ED
<< Active->InstantiationRange;
Implement support for C++0x alias templates. llvm-svn: 130953
2011-05-06 05:57:07 +08:00
} else {
Diags.Report(Active->PointOfInstantiation,
diag::note_template_type_alias_instantiation_here)
<< cast<TypeAliasTemplateDecl>(D)
Template instantiation for static data members that are defined out-of-line. Note that this also fixes a bug that affects non-template code, where we were not treating out-of-line static data members are "file-scope" variables, and therefore not checking their initializers. llvm-svn: 77002
2009-07-25 04:34:43 +08:00
<< Active->InstantiationRange;
When instantiating the definition of a member function of a class template, introduce that member function into the template instantiation stack. Also, add diagnostics showing the member function within the instantiation stack and clean up the qualified-name printing so that we get something like: note: in instantiation of member function 'Switch1<int, 2, 2>::f' requested here in the template instantiation backtrace. llvm-svn: 72015
2009-05-19 01:01:57 +08:00
}
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
break;
}
case ActiveTemplateInstantiation::DefaultTemplateArgumentInstantiation: {
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
TemplateDecl *Template = cast<TemplateDecl>(Active->Entity);
Streamify getNameForDiagnostic and remove the string versions of PrintTemplateArgumentList. llvm-svn: 175894
2013-02-22 23:46:01 +08:00
SmallVector<char, 128> TemplateArgsStr;
llvm::raw_svector_ostream OS(TemplateArgsStr);
Template->printName(OS);
TemplateSpecializationType::PrintTemplateArgumentList(OS,
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Active->TemplateArgs,
Create a new PrintingPolicy class, which we pass down through the AST printing logic to help customize the output. For now, we use this rather than a special flag to suppress the "struct" when printing "struct X" and to print the Boolean type as "bool" in C++ but "_Bool" in C. llvm-svn: 72590
2009-05-30 04:38:28 +08:00
Active->NumTemplateArgs,
Only print _Bool as 'bool' when 'bool' is defined as an object-like macro whose only replacement token is '_Bool'. llvm-svn: 140656
2011-09-28 07:30:47 +08:00
getPrintingPolicy());
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
diag::note_default_arg_instantiation_here)
Streamify getNameForDiagnostic and remove the string versions of PrintTemplateArgumentList. llvm-svn: 175894
2013-02-22 23:46:01 +08:00
<< OS.str()
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
<< Active->InstantiationRange;
break;
}
Template argument deduction for member pointers. Also, introduced some of the framework for performing instantiation as part of template argument deduction. llvm-svn: 73175
2009-06-11 07:47:09 +08:00
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
case ActiveTemplateInstantiation::ExplicitTemplateArgumentSubstitution: {
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
FunctionTemplateDecl *FnTmpl = cast<FunctionTemplateDecl>(Active->Entity);
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
diag::note_explicit_template_arg_substitution_here)
Improve diagnostics when printing a template instantiation backtrace involving substitution of deduced template arguments into a class template partial specialization or function template, or when substituting explicitly-specific template arguments into a function template. We now print the actual deduced argument bindings so the user can see what got deduced. llvm-svn: 99923
2010-03-31 04:35:20 +08:00
<< FnTmpl
<< getTemplateArgumentBindingsText(FnTmpl->getTemplateParameters(),
Active->TemplateArgs,
Active->NumTemplateArgs)
<< Active->InstantiationRange;
Template argument deduction for member pointers. Also, introduced some of the framework for performing instantiation as part of template argument deduction. llvm-svn: 73175
2009-06-11 07:47:09 +08:00
break;
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
case ActiveTemplateInstantiation::DeducedTemplateArgumentSubstitution:
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
if (ClassTemplatePartialSpecializationDecl *PartialSpec =
dyn_cast<ClassTemplatePartialSpecializationDecl>(Active->Entity)) {
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
diag::note_partial_spec_deduct_instantiation_here)
<< Context.getTypeDeclType(PartialSpec)
Improve diagnostics when printing a template instantiation backtrace involving substitution of deduced template arguments into a class template partial specialization or function template, or when substituting explicitly-specific template arguments into a function template. We now print the actual deduced argument bindings so the user can see what got deduced. llvm-svn: 99923
2010-03-31 04:35:20 +08:00
<< getTemplateArgumentBindingsText(
PartialSpec->getTemplateParameters(),
Active->TemplateArgs,
Active->NumTemplateArgs)
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
<< Active->InstantiationRange;
} else {
FunctionTemplateDecl *FnTmpl
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
= cast<FunctionTemplateDecl>(Active->Entity);
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
diag::note_function_template_deduction_instantiation_here)
Improve diagnostics when printing a template instantiation backtrace involving substitution of deduced template arguments into a class template partial specialization or function template, or when substituting explicitly-specific template arguments into a function template. We now print the actual deduced argument bindings so the user can see what got deduced. llvm-svn: 99923
2010-03-31 04:35:20 +08:00
<< FnTmpl
<< getTemplateArgumentBindingsText(FnTmpl->getTemplateParameters(),
Active->TemplateArgs,
Active->NumTemplateArgs)
<< Active->InstantiationRange;
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
}
break;
Template argument deduction for member pointers. Also, introduced some of the framework for performing instantiation as part of template argument deduction. llvm-svn: 73175
2009-06-11 07:47:09 +08:00
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
case ActiveTemplateInstantiation::DefaultFunctionArgumentInstantiation: {
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
ParmVarDecl *Param = cast<ParmVarDecl>(Active->Entity);
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
FunctionDecl *FD = cast<FunctionDecl>(Param->getDeclContext());
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Streamify getNameForDiagnostic and remove the string versions of PrintTemplateArgumentList. llvm-svn: 175894
2013-02-22 23:46:01 +08:00
SmallVector<char, 128> TemplateArgsStr;
llvm::raw_svector_ostream OS(TemplateArgsStr);
FD->printName(OS);
TemplateSpecializationType::PrintTemplateArgumentList(OS,
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Active->TemplateArgs,
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
Active->NumTemplateArgs,
Only print _Bool as 'bool' when 'bool' is defined as an object-like macro whose only replacement token is '_Bool'. llvm-svn: 140656
2011-09-28 07:30:47 +08:00
getPrintingPolicy());
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
diag::note_default_function_arg_instantiation_here)
Streamify getNameForDiagnostic and remove the string versions of PrintTemplateArgumentList. llvm-svn: 175894
2013-02-22 23:46:01 +08:00
<< OS.str()
Use a separate diagnostic for default function argument expressions. llvm-svn: 81062
2009-09-05 13:14:19 +08:00
<< Active->InstantiationRange;
break;
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
case ActiveTemplateInstantiation::PriorTemplateArgumentSubstitution: {
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
NamedDecl *Parm = cast<NamedDecl>(Active->Entity);
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
std::string Name;
if (!Parm->getName().empty())
Name = std::string(" '") + Parm->getName().str() + "'";
Improve our handling of non-type template parameters in partial specializations. We weren't dealing with any of the cases where the type of the non-type template argument differs from the type of the corresponding template parameter in the primary template. We would think that the template parameter in the partial specialization was not deducible (and warn about it, incorrectly), then fail to convert a deduced parameter to the type of the template parameter in the partial specialization (which may involve truncation, among other things). Fixes PR8905. llvm-svn: 122851
2011-01-05 07:35:54 +08:00
TemplateParameterList *TemplateParams = 0;
if (TemplateDecl *Template = dyn_cast<TemplateDecl>(Active->Template))
TemplateParams = Template->getTemplateParameters();
else
TemplateParams =
cast<ClassTemplatePartialSpecializationDecl>(Active->Template)
->getTemplateParameters();
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
diag::note_prior_template_arg_substitution)
<< isa<TemplateTemplateParmDecl>(Parm)
<< Name
Improve our handling of non-type template parameters in partial specializations. We weren't dealing with any of the cases where the type of the non-type template argument differs from the type of the corresponding template parameter in the primary template. We would think that the template parameter in the partial specialization was not deducible (and warn about it, incorrectly), then fail to convert a deduced parameter to the type of the template parameter in the partial specialization (which may involve truncation, among other things). Fixes PR8905. llvm-svn: 122851
2011-01-05 07:35:54 +08:00
<< getTemplateArgumentBindingsText(TemplateParams,
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
Active->TemplateArgs,
Active->NumTemplateArgs)
<< Active->InstantiationRange;
break;
}
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
case ActiveTemplateInstantiation::DefaultTemplateArgumentChecking: {
Improve our handling of non-type template parameters in partial specializations. We weren't dealing with any of the cases where the type of the non-type template argument differs from the type of the corresponding template parameter in the primary template. We would think that the template parameter in the partial specialization was not deducible (and warn about it, incorrectly), then fail to convert a deduced parameter to the type of the template parameter in the partial specialization (which may involve truncation, among other things). Fixes PR8905. llvm-svn: 122851
2011-01-05 07:35:54 +08:00
TemplateParameterList *TemplateParams = 0;
if (TemplateDecl *Template = dyn_cast<TemplateDecl>(Active->Template))
TemplateParams = Template->getTemplateParameters();
else
TemplateParams =
cast<ClassTemplatePartialSpecializationDecl>(Active->Template)
->getTemplateParameters();
Refactoring of Diagnostic class. -Move the stuff of Diagnostic related to creating/querying diagnostic IDs into a new DiagnosticIDs class. -DiagnosticIDs can be shared among multiple Diagnostics for multiple translation units. -The rest of the state in Diagnostic object is considered related and tied to one translation unit. -Have Diagnostic point to the SourceManager that is related with. Diagnostic can now accept just a SourceLocation instead of a FullSourceLoc. -Reflect the changes to various interfaces. llvm-svn: 119730
2010-11-19 04:06:41 +08:00
Diags.Report(Active->PointOfInstantiation,
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
diag::note_template_default_arg_checking)
Improve our handling of non-type template parameters in partial specializations. We weren't dealing with any of the cases where the type of the non-type template argument differs from the type of the corresponding template parameter in the primary template. We would think that the template parameter in the partial specialization was not deducible (and warn about it, incorrectly), then fail to convert a deduced parameter to the type of the template parameter in the partial specialization (which may involve truncation, among other things). Fixes PR8905. llvm-svn: 122851
2011-01-05 07:35:54 +08:00
<< getTemplateArgumentBindingsText(TemplateParams,
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
Active->TemplateArgs,
Active->NumTemplateArgs)
<< Active->InstantiationRange;
break;
}
Implement DR1330 in C++11 mode, to support libstdc++4.7 which uses it. We have a new flavor of exception specification, EST_Uninstantiated. A function type with this exception specification carries a pointer to a FunctionDecl, and the exception specification for that FunctionDecl is instantiated (if needed) and used in the place of the function type's exception specification. When a function template declaration with a non-trivial exception specification is instantiated, the specialization's exception specification is set to this new 'uninstantiated' kind rather than being instantiated immediately. Expr::CanThrow has migrated onto Sema, so it can instantiate exception specs on-demand. Also, any odr-use of a function triggers the instantiation of its exception specification (the exception specification could be needed by IRGen). In passing, fix two places where a DeclRefExpr was created but the corresponding function was not actually marked odr-used. We used to get away with this, but don't any more. Also fix a bug where instantiating an exception specification which refers to function parameters resulted in a crash. We still have the same bug in default arguments, which I'll be looking into next. This, plus a tiny patch to fix libstdc++'s common_type, is enough for clang to parse (and, in very limited testing, support) all of libstdc++4.7's standard headers. llvm-svn: 154886
2012-04-17 08:58:00 +08:00
case ActiveTemplateInstantiation::ExceptionSpecInstantiation:
Diags.Report(Active->PointOfInstantiation,
diag::note_template_exception_spec_instantiation_here)
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
<< cast<FunctionDecl>(Active->Entity)
Implement DR1330 in C++11 mode, to support libstdc++4.7 which uses it. We have a new flavor of exception specification, EST_Uninstantiated. A function type with this exception specification carries a pointer to a FunctionDecl, and the exception specification for that FunctionDecl is instantiated (if needed) and used in the place of the function type's exception specification. When a function template declaration with a non-trivial exception specification is instantiated, the specialization's exception specification is set to this new 'uninstantiated' kind rather than being instantiated immediately. Expr::CanThrow has migrated onto Sema, so it can instantiate exception specs on-demand. Also, any odr-use of a function triggers the instantiation of its exception specification (the exception specification could be needed by IRGen). In passing, fix two places where a DeclRefExpr was created but the corresponding function was not actually marked odr-used. We used to get away with this, but don't any more. Also fix a bug where instantiating an exception specification which refers to function parameters resulted in a crash. We still have the same bug in default arguments, which I'll be looking into next. This, plus a tiny patch to fix libstdc++'s common_type, is enough for clang to parse (and, in very limited testing, support) all of libstdc++4.7's standard headers. llvm-svn: 154886
2012-04-17 08:58:00 +08:00
<< Active->InstantiationRange;
break;
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
}
Add a notion of "post-diagnostic hooks", which are callbacks attached to a diagnostic that will be invoked after the diagnostic (if it is not suppressed). The hooks are allowed to produce additional diagnostics (typically notes) that provide more information. We should be able to use this to help diagnostic clients link notes back to the diagnostic they clarify. Comments welcome; I'll write up documentation and convert other clients (e.g., overload resolution failures) if there are no screams of protest. As the first client of post-diagnostic hooks, we now produce a template instantiation backtrace when a failure occurs during template instantiation. There's still more work to do to make this output pretty, if that's even possible. llvm-svn: 66557
2009-03-11 02:03:33 +08:00
}
}
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
Optional<TemplateDeductionInfo *> Sema::isSFINAEContext() const {
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
if (InNonInstantiationSFINAEContext)
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
return Optional<TemplateDeductionInfo *>(0);
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily. llvm-svn: 185610
2013-07-04 11:08:24 +08:00
for (SmallVectorImpl<ActiveTemplateInstantiation>::const_reverse_iterator
Update LLVM. Implement support for C++ Substitution Failure Is Not An Error (SFINAE), which says that errors that occur during template argument deduction do *not* produce diagnostics and do not necessarily make a program ill-formed. Instead, template argument deduction silently fails. This is currently implemented for template argument deduction during matching of class template partial specializations, although the mechanism will also apply to template argument deduction for function templates. The scheme is simple: - If we are in a template argument deduction context, any diagnostic that is considered a SFINAE error (or warning) will be suppressed. The error will be propagated up the call stack via the normal means. - By default, all warnings and errors are SFINAE errors. Add the NoSFINAE class to a diagnostic in the .td file to make it a hard error (e.g., for access-control violations). Note that, to make this fully work, every place in Sema that emits an error *and then immediately recovers* will need to check Sema::isSFINAEContext() to determine whether it must immediately return an error rather than recovering. llvm-svn: 73332
2009-06-14 15:33:30 +08:00
Active = ActiveTemplateInstantiations.rbegin(),
ActiveEnd = ActiveTemplateInstantiations.rend();
Active != ActiveEnd;
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
++Active)
{
Update LLVM. Implement support for C++ Substitution Failure Is Not An Error (SFINAE), which says that errors that occur during template argument deduction do *not* produce diagnostics and do not necessarily make a program ill-formed. Instead, template argument deduction silently fails. This is currently implemented for template argument deduction during matching of class template partial specializations, although the mechanism will also apply to template argument deduction for function templates. The scheme is simple: - If we are in a template argument deduction context, any diagnostic that is considered a SFINAE error (or warning) will be suppressed. The error will be propagated up the call stack via the normal means. - By default, all warnings and errors are SFINAE errors. Add the NoSFINAE class to a diagnostic in the .td file to make it a hard error (e.g., for access-control violations). Note that, to make this fully work, every place in Sema that emits an error *and then immediately recovers* will need to check Sema::isSFINAEContext() to determine whether it must immediately return an error rather than recovering. llvm-svn: 73332
2009-06-14 15:33:30 +08:00
switch(Active->Kind) {
Teach the evaluation of the __is_convertible_to trait to translate access control errors into SFINAE errors, so that the trait provides enough support to implement the C++0x std::is_convertible type trait. To get there, the SFINAETrap now knows how to set up a SFINAE context independent of any template instantiations or template argument deduction steps, and (separately) can set a Sema flag to translate access control errors into SFINAE errors. The latter can also be useful if we decide that access control errors during template argument deduction should cause substitution failure (rather than a hard error) as has been proposed for C++0x. llvm-svn: 124446
2011-01-28 06:31:44 +08:00
case ActiveTemplateInstantiation::TemplateInstantiation:
PR12647: An alias template instantiation which occurs in a SFINAE context is itself a SFINAE context. llvm-svn: 155621
2012-04-26 15:24:08 +08:00
// An instantiation of an alias template may or may not be a SFINAE
// context, depending on what else is on the stack.
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change. llvm-svn: 168145
2012-11-16 16:40:59 +08:00
if (isa<TypeAliasTemplateDecl>(Active->Entity))
PR12647: An alias template instantiation which occurs in a SFINAE context is itself a SFINAE context. llvm-svn: 155621
2012-04-26 15:24:08 +08:00
break;
// Fall through.
case ActiveTemplateInstantiation::DefaultFunctionArgumentInstantiation:
Implement DR1330 in C++11 mode, to support libstdc++4.7 which uses it. We have a new flavor of exception specification, EST_Uninstantiated. A function type with this exception specification carries a pointer to a FunctionDecl, and the exception specification for that FunctionDecl is instantiated (if needed) and used in the place of the function type's exception specification. When a function template declaration with a non-trivial exception specification is instantiated, the specialization's exception specification is set to this new 'uninstantiated' kind rather than being instantiated immediately. Expr::CanThrow has migrated onto Sema, so it can instantiate exception specs on-demand. Also, any odr-use of a function triggers the instantiation of its exception specification (the exception specification could be needed by IRGen). In passing, fix two places where a DeclRefExpr was created but the corresponding function was not actually marked odr-used. We used to get away with this, but don't any more. Also fix a bug where instantiating an exception specification which refers to function parameters resulted in a crash. We still have the same bug in default arguments, which I'll be looking into next. This, plus a tiny patch to fix libstdc++'s common_type, is enough for clang to parse (and, in very limited testing, support) all of libstdc++4.7's standard headers. llvm-svn: 154886
2012-04-17 08:58:00 +08:00
case ActiveTemplateInstantiation::ExceptionSpecInstantiation:
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
// This is a template instantiation, so there is no SFINAE.
Use None rather than Optional<T>() where possible. llvm-svn: 175705
2013-02-21 09:47:18 +08:00
return None;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Update LLVM. Implement support for C++ Substitution Failure Is Not An Error (SFINAE), which says that errors that occur during template argument deduction do *not* produce diagnostics and do not necessarily make a program ill-formed. Instead, template argument deduction silently fails. This is currently implemented for template argument deduction during matching of class template partial specializations, although the mechanism will also apply to template argument deduction for function templates. The scheme is simple: - If we are in a template argument deduction context, any diagnostic that is considered a SFINAE error (or warning) will be suppressed. The error will be propagated up the call stack via the normal means. - By default, all warnings and errors are SFINAE errors. Add the NoSFINAE class to a diagnostic in the .td file to make it a hard error (e.g., for access-control violations). Note that, to make this fully work, every place in Sema that emits an error *and then immediately recovers* will need to check Sema::isSFINAEContext() to determine whether it must immediately return an error rather than recovering. llvm-svn: 73332
2009-06-14 15:33:30 +08:00
case ActiveTemplateInstantiation::DefaultTemplateArgumentInstantiation:
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
case ActiveTemplateInstantiation::PriorTemplateArgumentSubstitution:
Improve diagnostics when a default template argument does not match with its corresponding template parameter. This can happen when we performed some substitution into the default template argument and what we had doesn't match any more, e.g., template<int> struct A; template<typename T, template<T> class X = A> class B; B<long> b; Previously, we'd emit a pretty but disembodied diagnostic showing how the default argument didn't match the template parameter. The diagnostic was good, but nothing tied it to the *use* of the default argument in "B<long>". This commit fixes that. Also, tweak the counting of active template instantiations to avoid counting non-instantiation records, such as those we create for (surprise!) checking default arguments, instantiating default arguments, and performing substitutions as part of template argument deduction. llvm-svn: 86884
2009-11-12 05:54:23 +08:00
case ActiveTemplateInstantiation::DefaultTemplateArgumentChecking:
Before checking a template template argument against its corresponding template template parameter, substitute any prior template arguments into the template template parameter. This, for example, allows us to properly check the template template argument for a class such as: template<typename T, template<T Value> class X> struct Foo; The actual implementation of this feature was trivial; most of the change is dedicated to giving decent diagnostics when this substitution goes horribly wrong. We now get a note like: note: while substituting prior template arguments into template template parameter 'X' [with T = float] As part of this change, enabled some very pedantic checking when comparing template template parameter lists, which shook out a bug in our overly-eager checking of default arguments of template template parameters. We now perform only minimal checking of such default arguments when they are initially parsed. llvm-svn: 86864
2009-11-12 03:13:48 +08:00
// A default template argument instantiation and substitution into
// template parameters with arguments for prior parameters may or may
// not be a SFINAE context; look further up the stack.
Update LLVM. Implement support for C++ Substitution Failure Is Not An Error (SFINAE), which says that errors that occur during template argument deduction do *not* produce diagnostics and do not necessarily make a program ill-formed. Instead, template argument deduction silently fails. This is currently implemented for template argument deduction during matching of class template partial specializations, although the mechanism will also apply to template argument deduction for function templates. The scheme is simple: - If we are in a template argument deduction context, any diagnostic that is considered a SFINAE error (or warning) will be suppressed. The error will be propagated up the call stack via the normal means. - By default, all warnings and errors are SFINAE errors. Add the NoSFINAE class to a diagnostic in the .td file to make it a hard error (e.g., for access-control violations). Note that, to make this fully work, every place in Sema that emits an error *and then immediately recovers* will need to check Sema::isSFINAEContext() to determine whether it must immediately return an error rather than recovering. llvm-svn: 73332
2009-06-14 15:33:30 +08:00
break;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Keep track of more information within the template instantiation stack, e.g., by distinguishing between substitution that occurs for template argument deduction vs. explicitly-specifiad template arguments. This is used both to improve diagnostics and to make sure we only provide SFINAE in those cases where SFINAE should apply. In addition, deal with the sticky issue where SFINAE only considers substitution of template arguments into the *type* of a function template; we need to issue hard errors beyond this point, as test/SemaTemplate/operator-template.cpp illustrates. llvm-svn: 74651
2009-07-02 06:01:06 +08:00
case ActiveTemplateInstantiation::ExplicitTemplateArgumentSubstitution:
case ActiveTemplateInstantiation::DeducedTemplateArgumentSubstitution:
// We're either substitution explicitly-specified template arguments
// or deduced template arguments, so SFINAE applies.
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
assert(Active->DeductionInfo && "Missing deduction info pointer");
return Active->DeductionInfo;
Update LLVM. Implement support for C++ Substitution Failure Is Not An Error (SFINAE), which says that errors that occur during template argument deduction do *not* produce diagnostics and do not necessarily make a program ill-formed. Instead, template argument deduction silently fails. This is currently implemented for template argument deduction during matching of class template partial specializations, although the mechanism will also apply to template argument deduction for function templates. The scheme is simple: - If we are in a template argument deduction context, any diagnostic that is considered a SFINAE error (or warning) will be suppressed. The error will be propagated up the call stack via the normal means. - By default, all warnings and errors are SFINAE errors. Add the NoSFINAE class to a diagnostic in the .td file to make it a hard error (e.g., for access-control violations). Note that, to make this fully work, every place in Sema that emits an error *and then immediately recovers* will need to check Sema::isSFINAEContext() to determine whether it must immediately return an error rather than recovering. llvm-svn: 73332
2009-06-14 15:33:30 +08:00
}
}
Use None rather than Optional<T>() where possible. llvm-svn: 175705
2013-02-21 09:47:18 +08:00
return None;
Update LLVM. Implement support for C++ Substitution Failure Is Not An Error (SFINAE), which says that errors that occur during template argument deduction do *not* produce diagnostics and do not necessarily make a program ill-formed. Instead, template argument deduction silently fails. This is currently implemented for template argument deduction during matching of class template partial specializations, although the mechanism will also apply to template argument deduction for function templates. The scheme is simple: - If we are in a template argument deduction context, any diagnostic that is considered a SFINAE error (or warning) will be suppressed. The error will be propagated up the call stack via the normal means. - By default, all warnings and errors are SFINAE errors. Add the NoSFINAE class to a diagnostic in the .td file to make it a hard error (e.g., for access-control violations). Note that, to make this fully work, every place in Sema that emits an error *and then immediately recovers* will need to check Sema::isSFINAEContext() to determine whether it must immediately return an error rather than recovering. llvm-svn: 73332
2009-06-14 15:33:30 +08:00
}
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
/// \brief Retrieve the depth and index of a parameter pack.
static std::pair<unsigned, unsigned>
getDepthAndIndex(NamedDecl *ND) {
if (TemplateTypeParmDecl *TTP = dyn_cast<TemplateTypeParmDecl>(ND))
return std::make_pair(TTP->getDepth(), TTP->getIndex());
if (NonTypeTemplateParmDecl *NTTP = dyn_cast<NonTypeTemplateParmDecl>(ND))
return std::make_pair(NTTP->getDepth(), NTTP->getIndex());
TemplateTemplateParmDecl *TTP = cast<TemplateTemplateParmDecl>(ND);
return std::make_pair(TTP->getDepth(), TTP->getIndex());
}
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
//===----------------------------------------------------------------------===/
// Template Instantiation for Types
//===----------------------------------------------------------------------===/
Implement template instantiation for pointer, reference, and (some) array types. Semantic checking for the construction of these types has been factored out of GetTypeForDeclarator and into separate subroutines (BuildPointerType, BuildReferenceType, BuildArrayType). We'll be doing the same thing for all other types (and declarations and expressions). As part of this, moved the type-instantiation functions into a class in an anonymous namespace. llvm-svn: 65663
2009-02-28 08:25:32 +08:00
namespace {
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
class TemplateInstantiator : public TreeTransform<TemplateInstantiator> {
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs;
Implement template instantiation for pointer, reference, and (some) array types. Semantic checking for the construction of these types has been factored out of GetTypeForDeclarator and into separate subroutines (BuildPointerType, BuildReferenceType, BuildArrayType). We'll be doing the same thing for all other types (and declarations and expressions). As part of this, moved the type-instantiation functions into a class in an anonymous namespace. llvm-svn: 65663
2009-02-28 08:25:32 +08:00
SourceLocation Loc;
DeclarationName Entity;
public:
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
typedef TreeTransform<TemplateInstantiator> inherited;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
TemplateInstantiator(Sema &SemaRef,
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs,
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
SourceLocation Loc,
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
DeclarationName Entity)
: inherited(SemaRef), TemplateArgs(TemplateArgs), Loc(Loc),
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
Entity(Entity) { }
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
/// \brief Determine whether the given type \p T has already been
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
/// transformed.
///
/// For the purposes of template instantiation, a type has already been
/// transformed if it is NULL or if it is not dependent.
When we encounter a non-dependent type during template instantiation, mark any declarations we see inside of that type as "referenced". Fixes PR7079. llvm-svn: 103323
2010-05-08 07:12:07 +08:00
bool AlreadyTransformed(QualType T);
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
/// \brief Returns the location of the entity being instantiated, if known.
SourceLocation getBaseLocation() { return Loc; }
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
/// \brief Returns the name of the entity being instantiated, if any.
DeclarationName getBaseEntity() { return Entity; }
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Only set the point of instantiation for an implicit or explicit instantiation once we have committed to performing the instantiation. As part of this, make our makeshift template-instantiation location information suck slightly less. Fixes PR5264. llvm-svn: 85209
2009-10-27 14:26:26 +08:00
/// \brief Sets the "base" location and entity when that
/// information is known based on another transformation.
void setBase(SourceLocation Loc, DeclarationName Entity) {
this->Loc = Loc;
this->Entity = Entity;
}
Implement basic support for template instantiation of pack expansions whose patterns are template arguments. We can now instantiate, e.g., typedef tuple<pair<OuterTypes, InnerTypes>...> type; where OuterTypes and InnerTypes are template type parameter packs. There is a horrible inefficiency in TemplateArgumentLoc::getPackExpansionPattern(), where we need to create copies of TypeLoc data because our interfaces traffic in TypeSourceInfo pointers where they should traffic in TypeLocs instead. I've isolated in efficiency in this one routine; once we refactor our interfaces to traffic in TypeLocs, we can eliminate it. llvm-svn: 122278
2010-12-21 06:05:00 +08:00
bool TryExpandParameterPacks(SourceLocation EllipsisLoc,
SourceRange PatternRange,
Omit llvm:: before ArrayRef, as we have using llvm::ArrayRef in include/clang/Basic/LLVM.h. llvm-svn: 188089
2013-08-10 02:02:13 +08:00
ArrayRef<UnexpandedParameterPack> Unexpanded,
bool &ShouldExpand, bool &RetainExpansion,
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
Optional<unsigned> &NumExpansions) {
Implement instantiation of pack expansions whose pattern is a type-id in an exception specification. llvm-svn: 122297
2010-12-21 08:52:54 +08:00
return getSema().CheckParameterPacksForExpansion(EllipsisLoc,
PatternRange, Unexpanded,
TemplateArgs,
ShouldExpand,
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
RetainExpansion,
Implement instantiation of pack expansions whose pattern is a type-id in an exception specification. llvm-svn: 122297
2010-12-21 08:52:54 +08:00
NumExpansions);
}
Implement basic support for template instantiation of pack expansions whose patterns are template arguments. We can now instantiate, e.g., typedef tuple<pair<OuterTypes, InnerTypes>...> type; where OuterTypes and InnerTypes are template type parameter packs. There is a horrible inefficiency in TemplateArgumentLoc::getPackExpansionPattern(), where we need to create copies of TypeLoc data because our interfaces traffic in TypeSourceInfo pointers where they should traffic in TypeLocs instead. I've isolated in efficiency in this one routine; once we refactor our interfaces to traffic in TypeLocs, we can eliminate it. llvm-svn: 122278
2010-12-21 06:05:00 +08:00
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
void ExpandingFunctionParameterPack(ParmVarDecl *Pack) {
SemaRef.CurrentInstantiationScope->MakeInstantiatedLocalArgPack(Pack);
}
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
TemplateArgument ForgetPartiallySubstitutedPack() {
TemplateArgument Result;
if (NamedDecl *PartialPack
= SemaRef.CurrentInstantiationScope->getPartiallySubstitutedPack()){
MultiLevelTemplateArgumentList &TemplateArgs
= const_cast<MultiLevelTemplateArgumentList &>(this->TemplateArgs);
unsigned Depth, Index;
llvm::tie(Depth, Index) = getDepthAndIndex(PartialPack);
if (TemplateArgs.hasTemplateArgument(Depth, Index)) {
Result = TemplateArgs(Depth, Index);
TemplateArgs.setArgument(Depth, Index, TemplateArgument());
}
}
return Result;
}
void RememberPartiallySubstitutedPack(TemplateArgument Arg) {
if (Arg.isNull())
return;
if (NamedDecl *PartialPack
= SemaRef.CurrentInstantiationScope->getPartiallySubstitutedPack()){
MultiLevelTemplateArgumentList &TemplateArgs
= const_cast<MultiLevelTemplateArgumentList &>(this->TemplateArgs);
unsigned Depth, Index;
llvm::tie(Depth, Index) = getDepthAndIndex(PartialPack);
TemplateArgs.setArgument(Depth, Index, Arg);
}
}
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
/// \brief Transform the given declaration by instantiating a reference to
/// this declaration.
When looking for the instantiated declaration that corresponds to a given declaration in a template, make sure that the context we're searching through is complete. Fixes PR6376. llvm-svn: 97444
2010-03-01 23:56:25 +08:00
Decl *TransformDecl(SourceLocation Loc, Decl *D);
Refactor the template-instantiation logic for expressions into a generic tree transformation (also used for recanonicalization) and a small amount of template-instantiation-specific logic. llvm-svn: 78645
2009-08-11 13:31:07 +08:00
Introduce support for template instantiation of lambda expressions. This is mostly a simple refact, splitting the main "start a lambda expression" function into smaller chunks that are driven either from the parser (Sema::ActOnLambdaExpr) or during AST transformation (TreeTransform::TransformLambdaExpr). A few minor interesting points: - Added new entry points for TreeTransform, so that we can explicitly establish the link between the lambda closure type in the template and the lambda closure type in the instantiation. - Added a bit into LambdaExpr specifying whether it had an explicit result type or not. We should have had this anyway. This code is 'lightly' tested. llvm-svn: 150417
2012-02-14 06:00:16 +08:00
void transformAttrs(Decl *Old, Decl *New) {
SemaRef.InstantiateAttrs(TemplateArgs, Old, New);
}
void transformedLocalDecl(Decl *Old, Decl *New) {
SemaRef.CurrentInstantiationScope->InstantiatedLocal(Old, New);
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
/// \brief Transform the definition of the given declaration by
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
/// instantiating it.
Finish pushing source-location information though TreeTransform's TransformDefinition. llvm-svn: 97445
2010-03-02 01:25:41 +08:00
Decl *TransformDefinition(SourceLocation Loc, Decl *D);
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Fix a couple of Doxygen issues pointed out by -Wdocumentation. llvm-svn: 163722
2012-09-13 01:01:48 +08:00
/// \brief Transform the first qualifier within a scope by instantiating the
Handle substitutions into the "first qualifier in scope" of a qualified member access expression (e.g., t->U::member) when that first qualifier refers to a template parameters. llvm-svn: 84612
2009-10-20 13:58:46 +08:00
/// declaration.
NamedDecl *TransformFirstQualifierInScope(NamedDecl *D, SourceLocation Loc);
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
/// \brief Rebuild the exception declaration and register the declaration
/// as an instantiated local.
Simplify template instantiation for C++ exception declarations, eliminating an unnecessary use of TemporaryBase in the process. llvm-svn: 113500
2010-09-10 01:09:21 +08:00
VarDecl *RebuildExceptionDecl(VarDecl *ExceptionDecl,
DeclaratorInfo -> TypeSourceInfo. Makes an effort to rename associated variables, but the results are imperfect. For posterity, I did: cat <<EOF > $cmdfile s/DeclaratorInfo/TypeSourceInfo/g s/DInfo/TInfo/g s/TypeTypeSourceInfo/TypeSourceInfo/g s/SourceTypeSourceInfo/TypeSourceInfo/g EOF find lib -name '*.cpp' -not -path 'lib/Parse/*' -exec sed -i '' -f $cmdfile '{}' \; find lib -name '*.h' -exec sed -i '' -f $cmdfile '{}' \; find include -name '*.h' -not -path 'include/clang/Parse/*' -not -path 'include/clang/Basic/*' -exec sed -i '' -f $cmdfile '{}' \; llvm-svn: 90743
2009-12-07 10:54:59 +08:00
TypeSourceInfo *Declarator,
Fixed source range for all DeclaratorDecl's. llvm-svn: 127225
2011-03-08 16:55:46 +08:00
SourceLocation StartLoc,
SourceLocation NameLoc,
IdentifierInfo *Name);
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Implement template instantiation for Objective-C @catch statements. This is the last of the Objective-C statements. llvm-svn: 102356
2010-04-27 01:57:08 +08:00
/// \brief Rebuild the Objective-C exception declaration and register the
/// declaration as an instantiated local.
VarDecl *RebuildObjCExceptionDecl(VarDecl *ExceptionDecl,
TypeSourceInfo *TSInfo, QualType T);
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
/// \brief Check for tag mismatches when instantiating an
/// elaborated type.
Use the real keyword location when rebuilding an elaborated type instead of making something up. Fixes PR8129. llvm-svn: 118258
2010-11-05 03:04:38 +08:00
QualType RebuildElaboratedType(SourceLocation KeywordLoc,
ElaboratedTypeKeyword Keyword,
Reinstate the introduction of source-location information for nested-name-speciciers within elaborated type names, e.g., enum clang::NestedNameSpecifier::SpecifierKind Fixes in this iteration include: (1) Compute the type-source range properly for a dependent template specialization type that starts with "template template-id ::", as in a member access expression dep->template f<T>::f() This is a latent bug I triggered with this change (because now we're checking the computed source ranges for dependent template specialization types). But the real problem was... (2) Make sure to set the qualifier range on a dependent template specialization type appropriately. This will go away once we push nested-name-specifier locations into dependent template specialization types, but it was the source of the valgrind errors on the buildbots. llvm-svn: 126765
2011-03-02 02:12:44 +08:00
NestedNameSpecifierLoc QualifierLoc,
QualType T);
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
Start migrating TreeTransform's TransformTemplateName over to version that preserve source-location information. This commit adds more redundancy than it removes; WIP. llvm-svn: 126849
2011-03-03 02:07:45 +08:00
TemplateName TransformTemplateName(CXXScopeSpec &SS,
TemplateName Name,
SourceLocation NameLoc,
QualType ObjectType = QualType(),
NamedDecl *FirstQualifierInScope = 0);
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult TransformPredefinedExpr(PredefinedExpr *E);
ExprResult TransformDeclRefExpr(DeclRefExpr *E);
ExprResult TransformCXXDefaultArgExpr(CXXDefaultArgExpr *E);
PR13811: Add a FunctionParmPackExpr node to handle references to function parameter packs where the reference is not being expanded but the pack has been. Previously, Clang would segfault in such cases. llvm-svn: 163672
2012-09-12 08:56:43 +08:00
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult TransformTemplateParmRefExpr(DeclRefExpr *E,
When instantiating a non-type template parameter pack, be sure to extract the appropriate argument from the argument pack (based on the current substitution index, of course). Simple instantiation of pack expansions involving non-type template parameter packs now works. llvm-svn: 122532
2010-12-24 08:15:10 +08:00
NonTypeTemplateParmDecl *D);
Introduce a new expression kind, SubstNonTypeTemplateParmPackExpr, that captures the substitution of a non-type template argument pack for a non-type template parameter pack within a pack expansion that cannot be fully expanded. This follows the approach taken by SubstTemplateTypeParmPackType. llvm-svn: 123506
2011-01-15 09:15:58 +08:00
ExprResult TransformSubstNonTypeTemplateParmPackExpr(
SubstNonTypeTemplateParmPackExpr *E);
PR13811: Add a FunctionParmPackExpr node to handle references to function parameter packs where the reference is not being expanded but the pack has been. Previously, Clang would segfault in such cases. llvm-svn: 163672
2012-09-12 08:56:43 +08:00
/// \brief Rebuild a DeclRefExpr for a ParmVarDecl reference.
ExprResult RebuildParmVarDeclRefExpr(ParmVarDecl *PD, SourceLocation Loc);
/// \brief Transform a reference to a function parameter pack.
ExprResult TransformFunctionParmPackRefExpr(DeclRefExpr *E,
ParmVarDecl *PD);
/// \brief Transform a FunctionParmPackExpr which was built when we couldn't
/// expand a function parameter pack reference which refers to an expanded
/// pack.
ExprResult TransformFunctionParmPackExpr(FunctionParmPackExpr *E);
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
QualType TransformFunctionProtoType(TypeLocBuilder &TLB,
Replace one hack with a different hack: strip out the ObjectType parameters to the Transform*Type functions and instead call out the specific cases where an object type and the unqualified lookup results are important. Fixes an assert and failed compile on a testcase from PR7248. llvm-svn: 118887
2010-11-12 16:19:04 +08:00
FunctionProtoTypeLoc TL);
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
QualType TransformFunctionProtoType(TypeLocBuilder &TLB,
FunctionProtoTypeLoc TL,
CXXRecordDecl *ThisContext,
unsigned ThisTypeQuals);
Handle substitutions into function parameter packs whose patterns contain multiple parameter packs at different levels. llvm-svn: 123488
2011-01-15 06:40:04 +08:00
ParmVarDecl *TransformFunctionTypeParam(ParmVarDecl *OldParm,
Store a parameter index and function prototype depth in every parameter node and use this to correctly mangle parameter references in function template signatures. A follow-up patch will improve the storage usage of these fields; here I've just done the lazy thing. llvm-svn: 130669
2011-05-02 06:35:37 +08:00
int indexAdjustment,
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
Optional<unsigned> NumExpansions,
When we're substituting into a function parameter pack and expect to get a function parameter pack (but don't due to weird substitutions), complain. Fixes the last bit of PR11848. llvm-svn: 148960
2012-01-26 00:15:54 +08:00
bool ExpectParameterPack);
Maintain type source information for functions through template instantiation. Based on a patch by Enea Zaffanella! I found a way to reduce some of the redundancy between TreeTransform's "standard" FunctionProtoType transformation and TemplateInstantiator's override, and I killed off the old SubstFunctionType by adding type source info for the last cases where we were creating FunctionDecls without TSI (at least that get passed through template instantiation). llvm-svn: 98252
2010-03-11 17:03:00 +08:00
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
/// \brief Transforms a template type parameter type by performing
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
/// substitution of the corresponding template type argument.
Rewrite TreeTransform to transform types as DeclaratorInfos rather than as bare QualTypes. Don't actually exploit this yet. llvm-svn: 84716
2009-10-21 08:40:46 +08:00
QualType TransformTemplateTypeParmType(TypeLocBuilder &TLB,
Replace one hack with a different hack: strip out the ObjectType parameters to the Transform*Type functions and instead call out the specific cases where an object type and the unqualified lookup results are important. Fixes an assert and failed compile on a testcase from PR7248. llvm-svn: 118887
2010-11-12 16:19:04 +08:00
TemplateTypeParmTypeLoc TL);
Fix multiple emission of the this-> fixit for each instantiation by fixing the AST during the instantiation. Fixes PR7417! llvm-svn: 107690
2010-07-07 03:51:49 +08:00
Start implementing support for substitution into pack expansions that involve template parameter packs at multiple template levels that occur within the signatures members of class templates (and partial specializations thereof). This is a work-in-progress that is deficient in several ways, notably: - It only works for template type parameter packs, but we need to also support non-type template parameter packs and template template parameter packs. - It doesn't keep track of the lengths of the substituted argument packs in the expansion, so it can't properly diagnose length mismatches. However, this is a concrete step in the right direction. llvm-svn: 123425
2011-01-14 10:55:32 +08:00
/// \brief Transforms an already-substituted template type parameter pack
/// into either itself (if we aren't substituting into its pack expansion)
/// or the appropriate substituted argument.
QualType TransformSubstTemplateTypeParmPackType(TypeLocBuilder &TLB,
SubstTemplateTypeParmPackTypeLoc TL);
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult TransformCallExpr(CallExpr *CE) {
Fix multiple emission of the this-> fixit for each instantiation by fixing the AST during the instantiation. Fixes PR7417! llvm-svn: 107690
2010-07-07 03:51:49 +08:00
getSema().CallsUndergoingInstantiation.push_back(CE);
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult Result =
Fix multiple emission of the this-> fixit for each instantiation by fixing the AST during the instantiation. Fixes PR7417! llvm-svn: 107690
2010-07-07 03:51:49 +08:00
TreeTransform<TemplateInstantiator>::TransformCallExpr(CE);
getSema().CallsUndergoingInstantiation.pop_back();
Rip out remnants of move semantic emulation and smart pointers in Sema. These were nops for quite a while and only lead to confusion. ASTMultiPtr now behaves like a proper dumb array reference. llvm-svn: 162475
2012-08-24 05:35:17 +08:00
return Result;
Fix multiple emission of the this-> fixit for each instantiation by fixing the AST during the instantiation. Fixes PR7417! llvm-svn: 107690
2010-07-07 03:51:49 +08:00
}
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
PR12057: Allow variadic template pack expansions to cross lambda boundaries. Rather than adding a ContainsUnexpandedParameterPack bit to essentially every AST node, we tunnel the bit directly up to the surrounding lambda expression when we reach a context where an unexpanded pack can not normally appear. Thus any statement or declaration within a lambda can now potentially contain an unexpanded parameter pack. llvm-svn: 160705
2012-07-25 11:56:55 +08:00
ExprResult TransformLambdaExpr(LambdaExpr *E) {
LocalInstantiationScope Scope(SemaRef, /*CombineWithOuterScope=*/true);
return TreeTransform<TemplateInstantiator>::TransformLambdaExpr(E);
}
ExprResult TransformLambdaScope(LambdaExpr *E,
Revert "Teach TreeTransform and family how to transform generic lambdas within templates and nested within themselves." This reverts commit r191879. It caused llvm.org/pr17476. llvm-svn: 191955
2013-10-04 22:28:51 +08:00
CXXMethodDecl *CallOperator) {
CallOperator->setInstantiationOfMemberFunction(E->getCallOperator(),
TSK_ImplicitInstantiation);
return TreeTransform<TemplateInstantiator>::
TransformLambdaScope(E, CallOperator);
PR12057: Allow variadic template pack expansions to cross lambda boundaries. Rather than adding a ContainsUnexpandedParameterPack bit to essentially every AST node, we tunnel the bit directly up to the surrounding lambda expression when we reach a context where an unexpanded pack can not normally appear. Thus any statement or declaration within a lambda can now potentially contain an unexpanded parameter pack. llvm-svn: 160705
2012-07-25 11:56:55 +08:00
}
Revert "Teach TreeTransform and family how to transform generic lambdas within templates and nested within themselves." This reverts commit r191879. It caused llvm.org/pr17476. llvm-svn: 191955
2013-10-04 22:28:51 +08:00
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
private:
ExprResult transformNonTypeTemplateParmRef(NonTypeTemplateParmDecl *parm,
SourceLocation loc,
PR13136: * When substituting a reference to a non-type template parameter pack where the corresponding argument is a pack expansion, transform into an expression which contains an unexpanded parameter pack rather than into an expression which contains a pack expansion. This causes the SubstNonTypeTemplateParmExpr to be inside the PackExpansionExpr, rather than outside, so the expression still looks like a pack expansion and can be deduced. * Teach MarkUsedTemplateParameters that we can deduce a reference to a template parameter if it's wrapped in a SubstNonTypeTemplateParmExpr (such nodes are added during alias template substitution). llvm-svn: 159922
2012-07-09 11:07:20 +08:00
TemplateArgument arg);
Improve the semantic checking for explicit instantiations of templates. In particular: - An explicit instantiation can follow an implicit instantiation (we were improperly diagnosing this as an error, previously). - In C++0x, an explicit instantiation that follows an explicit specialization of the same template specialization is ignored. In C++98, we just emit an extension warning. - In C++0x, an explicit instantiation must be in a namespace enclosing the original template. C++98 has no such requirement. Also, fixed a longstanding FIXME regarding the integral type that is used for the size of a constant array type when it is being instantiated. llvm-svn: 71689
2009-05-14 02:28:20 +08:00
};
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
}
When we encounter a non-dependent type during template instantiation, mark any declarations we see inside of that type as "referenced". Fixes PR7079. llvm-svn: 103323
2010-05-08 07:12:07 +08:00
bool TemplateInstantiator::AlreadyTransformed(QualType T) {
if (T.isNull())
return true;
Introduce the notion of instantiation dependence into Clang's AST. A type/expression/template argument/etc. is instantiation-dependent if it somehow involves a template parameter, even if it doesn't meet the requirements for the more common kinds of dependence (dependent type, type-dependent expression, value-dependent expression). When we see an instantiation-dependent type, we know we always need to perform substitution into that instantiation-dependent type. This keeps us from short-circuiting evaluation in places where we shouldn't, and lets us properly implement C++0x [temp.type]p2. In theory, this would also allow us to properly mangle instantiation-dependent-but-not-dependent decltype types per the Itanium C++ ABI, but we aren't quite there because we still mangle based on the canonical type in cases like, e.g., template<unsigned> struct A { }; template<typename T> void f(A<sizeof(sizeof(decltype(T() + T())))>) { } template void f<int>(A<sizeof(sizeof(int))>); and therefore get the wrong answer. llvm-svn: 134225
2011-07-01 09:22:09 +08:00
if (T->isInstantiationDependentType() || T->isVariablyModifiedType())
When we encounter a non-dependent type during template instantiation, mark any declarations we see inside of that type as "referenced". Fixes PR7079. llvm-svn: 103323
2010-05-08 07:12:07 +08:00
return false;
getSema().MarkDeclarationsReferencedInType(Loc, T);
return true;
}
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
static TemplateArgument
getPackSubstitutedTemplateArgument(Sema &S, TemplateArgument Arg) {
assert(S.ArgumentPackSubstitutionIndex >= 0);
assert(S.ArgumentPackSubstitutionIndex < (int)Arg.pack_size());
Arg = Arg.pack_begin()[S.ArgumentPackSubstitutionIndex];
if (Arg.isPackExpansion())
Arg = Arg.getPackExpansionPattern();
return Arg;
}
When looking for the instantiated declaration that corresponds to a given declaration in a template, make sure that the context we're searching through is complete. Fixes PR6376. llvm-svn: 97444
2010-03-01 23:56:25 +08:00
Decl *TemplateInstantiator::TransformDecl(SourceLocation Loc, Decl *D) {
Improve template instantiation for member access expressions that involve qualified names, e.g., x->Base::f. We now maintain enough information in the AST to compare the results of the name lookup of "Base" in the scope of the postfix-expression (determined at template definition time) and in the type of the object expression. llvm-svn: 80953
2009-09-04 05:38:09 +08:00
if (!D)
return 0;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Improve template instantiation for member access expressions that involve qualified names, e.g., x->Base::f. We now maintain enough information in the AST to compare the results of the name lookup of "Base" in the scope of the postfix-expression (determined at template definition time) and in the type of the object expression. llvm-svn: 80953
2009-09-04 05:38:09 +08:00
if (TemplateTemplateParmDecl *TTP = dyn_cast<TemplateTemplateParmDecl>(D)) {
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
if (TTP->getDepth() < TemplateArgs.getNumLevels()) {
Fix two issues with the substitution of template template parameters when instantiating the declaration of a member template: - Only check if the have a template template argument at a specific position when we already know that we have template arguments at that level; otherwise, we're substituting for a level-reduced template template parameter. - When trying to find an instantiated declaration for a template template parameter, look into the instantiated scope. This was a typo, where we had two checks for TemplateTypeParmDecl, one of which should have been a TemplateTemplateParmDecl. With these changes, tramp3d-v4 passes -fsyntax-only. llvm-svn: 95421
2010-02-06 03:54:12 +08:00
// If the corresponding template argument is NULL or non-existent, it's
// because we are performing instantiation from explicitly-specified
// template arguments in a function template, but there were some
// arguments left unspecified.
if (!TemplateArgs.hasTemplateArgument(TTP->getDepth(),
TTP->getPosition()))
return D;
Implement support for template template parameter packs, e.g., template<template<class> class ...Metafunctions> struct apply_to_each; llvm-svn: 122874
2011-01-05 23:48:55 +08:00
TemplateArgument Arg = TemplateArgs(TTP->getDepth(), TTP->getPosition());
if (TTP->isParameterPack()) {
assert(Arg.getKind() == TemplateArgument::Pack &&
"Missing argument pack");
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
Implement support for template template parameter packs, e.g., template<template<class> class ...Metafunctions> struct apply_to_each; llvm-svn: 122874
2011-01-05 23:48:55 +08:00
}
TemplateName Template = Arg.getAsTemplate();
Introduce a new representation for template template parameters. Rather than storing them as either declarations (for the non-dependent case) or expressions (for the dependent case), we now (always) store them as TemplateNames. The primary change here is to add a new kind of TemplateArgument, which stores a TemplateName. However, making that change ripples to every switch on a TemplateArgument's kind, also affecting TemplateArgumentLocInfo/TemplateArgumentLoc, default template arguments for template template parameters, type-checking of template template arguments, etc. This change is light on testing. It should fix several pre-existing problems with template template parameters, such as: - the inability to use dependent template names as template template arguments - template template parameter default arguments cannot be instantiation However, there are enough pieces missing that more implementation is required before we can adequately test template template parameters. llvm-svn: 86777
2009-11-11 09:00:40 +08:00
assert(!Template.isNull() && Template.getAsTemplateDecl() &&
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
"Wrong kind of template template argument");
Introduce a new representation for template template parameters. Rather than storing them as either declarations (for the non-dependent case) or expressions (for the dependent case), we now (always) store them as TemplateNames. The primary change here is to add a new kind of TemplateArgument, which stores a TemplateName. However, making that change ripples to every switch on a TemplateArgument's kind, also affecting TemplateArgumentLocInfo/TemplateArgumentLoc, default template arguments for template template parameters, type-checking of template template arguments, etc. This change is light on testing. It should fix several pre-existing problems with template template parameters, such as: - the inability to use dependent template names as template template arguments - template template parameter default arguments cannot be instantiation However, there are enough pieces missing that more implementation is required before we can adequately test template template parameters. llvm-svn: 86777
2009-11-11 09:00:40 +08:00
return Template.getAsTemplateDecl();
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Introduce a new representation for template template parameters. Rather than storing them as either declarations (for the non-dependent case) or expressions (for the dependent case), we now (always) store them as TemplateNames. The primary change here is to add a new kind of TemplateArgument, which stores a TemplateName. However, making that change ripples to every switch on a TemplateArgument's kind, also affecting TemplateArgumentLocInfo/TemplateArgumentLoc, default template arguments for template template parameters, type-checking of template template arguments, etc. This change is light on testing. It should fix several pre-existing problems with template template parameters, such as: - the inability to use dependent template names as template template arguments - template template parameter default arguments cannot be instantiation However, there are enough pieces missing that more implementation is required before we can adequately test template template parameters. llvm-svn: 86777
2009-11-11 09:00:40 +08:00
// Fall through to find the instantiated declaration for this template
// template parameter.
Implement transformation of template names within the generic tree transform, then use the result for template instantiation. The generic transformation fixes a few issues: - It copes better with template template parameters and member templates (when they're implemented). - The logic used to replace template template parameters with their arguments is now centralized in TransformDecl, so that it will apply for other declaration-instantiation steps. - The error-recovery strategy is normalized now, so that any error results in a NULL TemplateName. llvm-svn: 78292
2009-08-06 14:41:21 +08:00
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
When looking for the instantiated declaration that corresponds to a given declaration in a template, make sure that the context we're searching through is complete. Fixes PR6376. llvm-svn: 97444
2010-03-01 23:56:25 +08:00
return SemaRef.FindInstantiatedDecl(Loc, cast<NamedDecl>(D), TemplateArgs);
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
}
Finish pushing source-location information though TreeTransform's TransformDefinition. llvm-svn: 97445
2010-03-02 01:25:41 +08:00
Decl *TemplateInstantiator::TransformDefinition(SourceLocation Loc, Decl *D) {
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
Decl *Inst = getSema().SubstDecl(D, getSema().CurContext, TemplateArgs);
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
if (!Inst)
return 0;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
getSema().CurrentInstantiationScope->InstantiatedLocal(D, Inst);
return Inst;
}
Handle substitutions into the "first qualifier in scope" of a qualified member access expression (e.g., t->U::member) when that first qualifier refers to a template parameters. llvm-svn: 84612
2009-10-20 13:58:46 +08:00
NamedDecl *
TemplateInstantiator::TransformFirstQualifierInScope(NamedDecl *D,
SourceLocation Loc) {
// If the first part of the nested-name-specifier was a template type
// parameter, instantiate that type parameter down to a tag type.
if (TemplateTypeParmDecl *TTPD = dyn_cast_or_null<TemplateTypeParmDecl>(D)) {
const TemplateTypeParmType *TTP
= cast<TemplateTypeParmType>(getSema().Context.getTypeDeclType(TTPD));
Handle instantiation of template type parameter packs that occur as the first qualifier in scope. We can't adequately test this test, unfortunately. llvm-svn: 122283
2010-12-21 06:48:17 +08:00
Handle substitutions into the "first qualifier in scope" of a qualified member access expression (e.g., t->U::member) when that first qualifier refers to a template parameters. llvm-svn: 84612
2009-10-20 13:58:46 +08:00
if (TTP->getDepth() < TemplateArgs.getNumLevels()) {
Handle instantiation of template type parameter packs that occur as the first qualifier in scope. We can't adequately test this test, unfortunately. llvm-svn: 122283
2010-12-21 06:48:17 +08:00
// FIXME: This needs testing w/ member access expressions.
TemplateArgument Arg = TemplateArgs(TTP->getDepth(), TTP->getIndex());
if (TTP->isParameterPack()) {
assert(Arg.getKind() == TemplateArgument::Pack &&
"Missing argument pack");
Teach template template argument pack expansions to keep track of the number of expansions, when we know it, and propagate that information through Sema. llvm-svn: 123493
2011-01-15 07:41:42 +08:00
if (getSema().ArgumentPackSubstitutionIndex == -1)
Handle instantiation of template type parameter packs that occur as the first qualifier in scope. We can't adequately test this test, unfortunately. llvm-svn: 122283
2010-12-21 06:48:17 +08:00
return 0;
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
Handle instantiation of template type parameter packs that occur as the first qualifier in scope. We can't adequately test this test, unfortunately. llvm-svn: 122283
2010-12-21 06:48:17 +08:00
}
QualType T = Arg.getAsType();
Handle substitutions into the "first qualifier in scope" of a qualified member access expression (e.g., t->U::member) when that first qualifier refers to a template parameters. llvm-svn: 84612
2009-10-20 13:58:46 +08:00
if (T.isNull())
When looking for the instantiated declaration that corresponds to a given declaration in a template, make sure that the context we're searching through is complete. Fixes PR6376. llvm-svn: 97444
2010-03-01 23:56:25 +08:00
return cast_or_null<NamedDecl>(TransformDecl(Loc, D));
Handle substitutions into the "first qualifier in scope" of a qualified member access expression (e.g., t->U::member) when that first qualifier refers to a template parameters. llvm-svn: 84612
2009-10-20 13:58:46 +08:00
if (const TagType *Tag = T->getAs<TagType>())
return Tag->getDecl();
// The resulting type is not a tag; complain.
getSema().Diag(Loc, diag::err_nested_name_spec_non_tag) << T;
return 0;
}
}
When looking for the instantiated declaration that corresponds to a given declaration in a template, make sure that the context we're searching through is complete. Fixes PR6376. llvm-svn: 97444
2010-03-01 23:56:25 +08:00
return cast_or_null<NamedDecl>(TransformDecl(Loc, D));
Handle substitutions into the "first qualifier in scope" of a qualified member access expression (e.g., t->U::member) when that first qualifier refers to a template parameters. llvm-svn: 84612
2009-10-20 13:58:46 +08:00
}
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
VarDecl *
TemplateInstantiator::RebuildExceptionDecl(VarDecl *ExceptionDecl,
DeclaratorInfo -> TypeSourceInfo. Makes an effort to rename associated variables, but the results are imperfect. For posterity, I did: cat <<EOF > $cmdfile s/DeclaratorInfo/TypeSourceInfo/g s/DInfo/TInfo/g s/TypeTypeSourceInfo/TypeSourceInfo/g s/SourceTypeSourceInfo/TypeSourceInfo/g EOF find lib -name '*.cpp' -not -path 'lib/Parse/*' -exec sed -i '' -f $cmdfile '{}' \; find lib -name '*.h' -exec sed -i '' -f $cmdfile '{}' \; find include -name '*.h' -not -path 'include/clang/Parse/*' -not -path 'include/clang/Basic/*' -exec sed -i '' -f $cmdfile '{}' \; llvm-svn: 90743
2009-12-07 10:54:59 +08:00
TypeSourceInfo *Declarator,
Fixed source range for all DeclaratorDecl's. llvm-svn: 127225
2011-03-08 16:55:46 +08:00
SourceLocation StartLoc,
SourceLocation NameLoc,
IdentifierInfo *Name) {
Simplify template instantiation for C++ exception declarations, eliminating an unnecessary use of TemporaryBase in the process. llvm-svn: 113500
2010-09-10 01:09:21 +08:00
VarDecl *Var = inherited::RebuildExceptionDecl(ExceptionDecl, Declarator,
Fixed source range for all DeclaratorDecl's. llvm-svn: 127225
2011-03-08 16:55:46 +08:00
StartLoc, NameLoc, Name);
Implement template instantiation for Objective-C @catch statements. This is the last of the Objective-C statements. llvm-svn: 102356
2010-04-27 01:57:08 +08:00
if (Var)
getSema().CurrentInstantiationScope->InstantiatedLocal(ExceptionDecl, Var);
return Var;
}
VarDecl *TemplateInstantiator::RebuildObjCExceptionDecl(VarDecl *ExceptionDecl,
TypeSourceInfo *TSInfo,
QualType T) {
VarDecl *Var = inherited::RebuildObjCExceptionDecl(ExceptionDecl, TSInfo, T);
if (Var)
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
getSema().CurrentInstantiationScope->InstantiatedLocal(ExceptionDecl, Var);
return Var;
}
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
QualType
Use the real keyword location when rebuilding an elaborated type instead of making something up. Fixes PR8129. llvm-svn: 118258
2010-11-05 03:04:38 +08:00
TemplateInstantiator::RebuildElaboratedType(SourceLocation KeywordLoc,
ElaboratedTypeKeyword Keyword,
Reinstate the introduction of source-location information for nested-name-speciciers within elaborated type names, e.g., enum clang::NestedNameSpecifier::SpecifierKind Fixes in this iteration include: (1) Compute the type-source range properly for a dependent template specialization type that starts with "template template-id ::", as in a member access expression dep->template f<T>::f() This is a latent bug I triggered with this change (because now we're checking the computed source ranges for dependent template specialization types). But the real problem was... (2) Make sure to set the qualifier range on a dependent template specialization type appropriately. This will go away once we push nested-name-specifier locations into dependent template specialization types, but it was the source of the valgrind errors on the buildbots. llvm-svn: 126765
2011-03-02 02:12:44 +08:00
NestedNameSpecifierLoc QualifierLoc,
Merged Elaborated and QualifiedName types. llvm-svn: 103517
2010-05-12 05:36:43 +08:00
QualType T) {
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
if (const TagType *TT = T->getAs<TagType>()) {
TagDecl* TD = TT->getDecl();
Use the real keyword location when rebuilding an elaborated type instead of making something up. Fixes PR8129. llvm-svn: 118258
2010-11-05 03:04:38 +08:00
SourceLocation TagLocation = KeywordLoc;
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
IdentifierInfo *Id = TD->getIdentifier();
// TODO: should we even warn on struct/class mismatches for this? Seems
// like it's likely to produce a lot of spurious errors.
Don't form a null reference when checking for validity of an anonymous elaborated type specifier in template instantiation: such a specifier is always valid because it must be specified within the definition of the type. llvm-svn: 162068
2012-08-17 08:12:27 +08:00
if (Id && Keyword != ETK_None && Keyword != ETK_Typename) {
Merged Elaborated and QualifiedName types. llvm-svn: 103517
2010-05-12 05:36:43 +08:00
TagTypeKind Kind = TypeWithKeyword::getTagTypeKindForKeyword(Keyword);
Made changes to how 'struct'/'class' mismatches are handled in -Wmismatched-tags. - Removed fix-it hints from template instaniations since changes to the templates are rarely helpful. - Changed the caret in template instaniations from the class/struct name to the class/struct keyword, matching the other warnings. - Do not offer fix-it hints when multiple declarations disagree. Warnings are still given. - Once a definition is found, offer a fix-it hint to all previous declarations with wrong tag. - Declarations that disagree with a previous definition will get a fix-it hint to change the declaration. llvm-svn: 132831
2011-06-10 11:11:26 +08:00
if (!SemaRef.isAcceptableTagRedeclaration(TD, Kind, /*isDefinition*/false,
TagLocation, *Id)) {
Merged Elaborated and QualifiedName types. llvm-svn: 103517
2010-05-12 05:36:43 +08:00
SemaRef.Diag(TagLocation, diag::err_use_with_wrong_tag)
<< Id
<< FixItHint::CreateReplacement(SourceRange(TagLocation),
TD->getKindName());
SemaRef.Diag(TD->getLocation(), diag::note_previous_use);
}
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
}
}
Use the real keyword location when rebuilding an elaborated type instead of making something up. Fixes PR8129. llvm-svn: 118258
2010-11-05 03:04:38 +08:00
return TreeTransform<TemplateInstantiator>::RebuildElaboratedType(KeywordLoc,
Keyword,
Reinstate the introduction of source-location information for nested-name-speciciers within elaborated type names, e.g., enum clang::NestedNameSpecifier::SpecifierKind Fixes in this iteration include: (1) Compute the type-source range properly for a dependent template specialization type that starts with "template template-id ::", as in a member access expression dep->template f<T>::f() This is a latent bug I triggered with this change (because now we're checking the computed source ranges for dependent template specialization types). But the real problem was... (2) Make sure to set the qualifier range on a dependent template specialization type appropriately. This will go away once we push nested-name-specifier locations into dependent template specialization types, but it was the source of the valgrind errors on the buildbots. llvm-svn: 126765
2011-03-02 02:12:44 +08:00
QualifierLoc,
T);
Support elaborated dependent types and diagnose tag mismatches. llvm-svn: 81504
2009-09-11 12:59:25 +08:00
}
Start migrating TreeTransform's TransformTemplateName over to version that preserve source-location information. This commit adds more redundancy than it removes; WIP. llvm-svn: 126849
2011-03-03 02:07:45 +08:00
TemplateName TemplateInstantiator::TransformTemplateName(CXXScopeSpec &SS,
TemplateName Name,
SourceLocation NameLoc,
QualType ObjectType,
NamedDecl *FirstQualifierInScope) {
if (TemplateTemplateParmDecl *TTP
= dyn_cast_or_null<TemplateTemplateParmDecl>(Name.getAsTemplateDecl())) {
if (TTP->getDepth() < TemplateArgs.getNumLevels()) {
// If the corresponding template argument is NULL or non-existent, it's
// because we are performing instantiation from explicitly-specified
// template arguments in a function template, but there were some
// arguments left unspecified.
if (!TemplateArgs.hasTemplateArgument(TTP->getDepth(),
TTP->getPosition()))
return Name;
TemplateArgument Arg = TemplateArgs(TTP->getDepth(), TTP->getPosition());
if (TTP->isParameterPack()) {
assert(Arg.getKind() == TemplateArgument::Pack &&
"Missing argument pack");
if (getSema().ArgumentPackSubstitutionIndex == -1) {
// We have the template argument pack to substitute, but we're not
// actually expanding the enclosing pack expansion yet. So, just
// keep the entire argument pack.
return getSema().Context.getSubstTemplateTemplateParmPack(TTP, Arg);
}
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
Start migrating TreeTransform's TransformTemplateName over to version that preserve source-location information. This commit adds more redundancy than it removes; WIP. llvm-svn: 126849
2011-03-03 02:07:45 +08:00
}
TemplateName Template = Arg.getAsTemplate();
Implement support for C++0x alias templates. llvm-svn: 130953
2011-05-06 05:57:07 +08:00
assert(!Template.isNull() && "Null template template argument");
Preserve that a TemplateName was arrived at by substituting for a template template parameter. Uses to follow. I've also made the uniquing of SubstTemplateTemplateParmPacks use a ContextualFoldingSet as a minor space efficiency. llvm-svn: 134137
2011-06-30 16:33:18 +08:00
When substituting in for a template name, do not produce a qualified template name as the result of substitution. The qualifier is handled separately by the tree transformer, so we would end up in an inconsistent state. This is actually the last bit of PR9016, and possibly also fixes PR8965. It takes Boost.Icl from "epic fail" down to a single failure. llvm-svn: 127108
2011-03-06 04:06:51 +08:00
// We don't ever want to substitute for a qualified template name, since
// the qualifier is handled separately. So, look through the qualified
// template name to its underlying declaration.
if (QualifiedTemplateName *QTN = Template.getAsQualifiedTemplateName())
Template = TemplateName(QTN->getTemplateDecl());
Preserve that a TemplateName was arrived at by substituting for a template template parameter. Uses to follow. I've also made the uniquing of SubstTemplateTemplateParmPacks use a ContextualFoldingSet as a minor space efficiency. llvm-svn: 134137
2011-06-30 16:33:18 +08:00
Template = getSema().Context.getSubstTemplateTemplateParm(TTP, Template);
Start migrating TreeTransform's TransformTemplateName over to version that preserve source-location information. This commit adds more redundancy than it removes; WIP. llvm-svn: 126849
2011-03-03 02:07:45 +08:00
return Template;
}
}
if (SubstTemplateTemplateParmPackStorage *SubstPack
= Name.getAsSubstTemplateTemplateParmPack()) {
if (getSema().ArgumentPackSubstitutionIndex == -1)
return Name;
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
TemplateArgument Arg = SubstPack->getArgumentPack();
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
return Arg.getAsTemplate();
Start migrating TreeTransform's TransformTemplateName over to version that preserve source-location information. This commit adds more redundancy than it removes; WIP. llvm-svn: 126849
2011-03-03 02:07:45 +08:00
}
return inherited::TransformTemplateName(SS, Name, NameLoc, ObjectType,
FirstQualifierInScope);
}
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult
The refactor of implicit member access expressions means we don't need this horrible isAddressOfOperand hack in TreeTransform, since that syntactic information is managed by the initial parser callbacks now. That's enough insomniac commits for one night. llvm-svn: 90849
2009-12-08 17:21:05 +08:00
TemplateInstantiator::TransformPredefinedExpr(PredefinedExpr *E) {
Instantiate PredefinedExprs correctly. Patch by Sam Weinig! llvm-svn: 81498
2009-09-11 09:22:35 +08:00
if (!E->isTypeDependent())
No really, we don't have a retain/release system for statements/expressions anymore. llvm-svn: 117357
2010-10-26 15:05:15 +08:00
return SemaRef.Owned(E);
Instantiate PredefinedExprs correctly. Patch by Sam Weinig! llvm-svn: 81498
2009-09-11 09:22:35 +08:00
Handle PredefinedExpr with templates and lambdas Summary: - lambdas, blocks or captured statements in templates were not handled which causes codegen crashes. Differential Revision: http://llvm-reviews.chandlerc.com/D1628 llvm-svn: 190784
2013-09-16 21:57:27 +08:00
return getSema().BuildPredefinedExpr(E->getLocation(), E->getIdentType());
Instantiate PredefinedExprs correctly. Patch by Sam Weinig! llvm-svn: 81498
2009-09-11 09:22:35 +08:00
}
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
TemplateInstantiator::TransformTemplateParmRefExpr(DeclRefExpr *E,
When substituting a declaration non-type template argument for a non-type template parameter that has reference type, augment the qualifiers of the non-type template argument with those of the referenced type. Fixes PR6250. llvm-svn: 95607
2010-02-09 07:41:45 +08:00
NonTypeTemplateParmDecl *NTTP) {
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
// If the corresponding template argument is NULL or non-existent, it's
// because we are performing instantiation from explicitly-specified
// template arguments in a function template, but there were some
// arguments left unspecified.
if (!TemplateArgs.hasTemplateArgument(NTTP->getDepth(),
NTTP->getPosition()))
No really, we don't have a retain/release system for statements/expressions anymore. llvm-svn: 117357
2010-10-26 15:05:15 +08:00
return SemaRef.Owned(E);
Implement "incremental" template instantiation for non-type template parameters and template type parameters, which occurs when substituting into the declarations of member templates inside class templates. This eliminates errors about our inability to "reduce non-type template parameter depth", fixing PR5311. Also fixes a bug when instantiating a template type parameter declaration in a member template, where we weren't properly reducing the template parameter's depth. LLVM's StringSwitch header now parses. llvm-svn: 85669
2009-11-01 01:21:17 +08:00
When instantiating a non-type template parameter pack, be sure to extract the appropriate argument from the argument pack (based on the current substitution index, of course). Simple instantiation of pack expansions involving non-type template parameter packs now works. llvm-svn: 122532
2010-12-24 08:15:10 +08:00
TemplateArgument Arg = TemplateArgs(NTTP->getDepth(), NTTP->getPosition());
if (NTTP->isParameterPack()) {
assert(Arg.getKind() == TemplateArgument::Pack &&
"Missing argument pack");
if (getSema().ArgumentPackSubstitutionIndex == -1) {
Introduce a new expression kind, SubstNonTypeTemplateParmPackExpr, that captures the substitution of a non-type template argument pack for a non-type template parameter pack within a pack expansion that cannot be fully expanded. This follows the approach taken by SubstTemplateTypeParmPackType. llvm-svn: 123506
2011-01-15 09:15:58 +08:00
// We have an argument pack, but we can't select a particular argument
// out of it yet. Therefore, we'll build an expression to hold on to that
// argument pack.
QualType TargetType = SemaRef.SubstType(NTTP->getType(), TemplateArgs,
E->getLocation(),
NTTP->getDeclName());
if (TargetType.isNull())
return ExprError();
return new (SemaRef.Context) SubstNonTypeTemplateParmPackExpr(TargetType,
NTTP,
E->getLocation(),
Arg);
When instantiating a non-type template parameter pack, be sure to extract the appropriate argument from the argument pack (based on the current substitution index, of course). Simple instantiation of pack expansions involving non-type template parameter packs now works. llvm-svn: 122532
2010-12-24 08:15:10 +08:00
}
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
When instantiating a non-type template parameter pack, be sure to extract the appropriate argument from the argument pack (based on the current substitution index, of course). Simple instantiation of pack expansions involving non-type template parameter packs now works. llvm-svn: 122532
2010-12-24 08:15:10 +08:00
}
Implement "incremental" template instantiation for non-type template parameters and template type parameters, which occurs when substituting into the declarations of member templates inside class templates. This eliminates errors about our inability to "reduce non-type template parameter depth", fixing PR5311. Also fixes a bug when instantiating a template type parameter declaration in a member template, where we weren't properly reducing the template parameter's depth. LLVM's StringSwitch header now parses. llvm-svn: 85669
2009-11-01 01:21:17 +08:00
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
return transformNonTypeTemplateParmRef(NTTP, E->getLocation(), Arg);
}
ExprResult TemplateInstantiator::transformNonTypeTemplateParmRef(
NonTypeTemplateParmDecl *parm,
SourceLocation loc,
PR13136: * When substituting a reference to a non-type template parameter pack where the corresponding argument is a pack expansion, transform into an expression which contains an unexpanded parameter pack rather than into an expression which contains a pack expansion. This causes the SubstNonTypeTemplateParmExpr to be inside the PackExpansionExpr, rather than outside, so the expression still looks like a pack expansion and can be deduced. * Teach MarkUsedTemplateParameters that we can deduce a reference to a template parameter if it's wrapped in a SubstNonTypeTemplateParmExpr (such nodes are added during alias template substitution). llvm-svn: 159922
2012-07-09 11:07:20 +08:00
TemplateArgument arg) {
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
ExprResult result;
QualType type;
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
// The template argument itself might be an expression, in which
// case we just return that expression.
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
if (arg.getKind() == TemplateArgument::Expression) {
Expr *argExpr = arg.getAsExpr();
result = SemaRef.Owned(argExpr);
type = argExpr->getType();
Implement "incremental" template instantiation for non-type template parameters and template type parameters, which occurs when substituting into the declarations of member templates inside class templates. This eliminates errors about our inability to "reduce non-type template parameter depth", fixing PR5311. Also fixes a bug when instantiating a template type parameter declaration in a member template, where we weren't properly reducing the template parameter's depth. LLVM's StringSwitch header now parses. llvm-svn: 85669
2009-11-01 01:21:17 +08:00
Fix the AST representation for non-type template arguments to encode enough information so we can mangle them correctly in cases involving dependent parameter types. (This specifically impacts cases involving null pointers and cases involving parameters of reference type.) Fix the mangler to use this information instead of trying to scavenge it out of the parameter declaration. <rdar://problem/12296776>. llvm-svn: 164656
2012-09-26 10:36:12 +08:00
} else if (arg.getKind() == TemplateArgument::Declaration ||
arg.getKind() == TemplateArgument::NullPtr) {
Implement support for null non-type template arguments for non-type template parameters of pointer, pointer-to-member, or nullptr_t type in C++11. Fixes PR9700 / <rdar://problem/11193097>. llvm-svn: 154219
2012-04-07 06:40:38 +08:00
ValueDecl *VD;
Fix the AST representation for non-type template arguments to encode enough information so we can mangle them correctly in cases involving dependent parameter types. (This specifically impacts cases involving null pointers and cases involving parameters of reference type.) Fix the mangler to use this information instead of trying to scavenge it out of the parameter declaration. <rdar://problem/12296776>. llvm-svn: 164656
2012-09-26 10:36:12 +08:00
if (arg.getKind() == TemplateArgument::Declaration) {
VD = cast<ValueDecl>(arg.getAsDecl());
Implement support for null non-type template arguments for non-type template parameters of pointer, pointer-to-member, or nullptr_t type in C++11. Fixes PR9700 / <rdar://problem/11193097>. llvm-svn: 154219
2012-04-07 06:40:38 +08:00
// Find the instantiation of the template argument. This is
// required for nested templates.
VD = cast_or_null<ValueDecl>(
getSema().FindInstantiatedDecl(loc, VD, TemplateArgs));
if (!VD)
return ExprError();
} else {
// Propagate NULL template argument.
VD = 0;
}
Use a substituted type when determining how to substitute in non-type template params. Don't insert addrof operations when matching against a pointer; array/function conversions should take care of this for us, assuming the argument type-checked in the first place. Add a fixme where we seem to be using a less-restrictive reference type than we should. Fixes PR 6249. llvm-svn: 95495
2010-02-06 18:23:53 +08:00
// Derive the type we want the substituted decl to have. This had
// better be non-dependent, or these checks will have serious problems.
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
if (parm->isExpandedParameterPack()) {
type = parm->getExpansionType(SemaRef.ArgumentPackSubstitutionIndex);
} else if (parm->isParameterPack() &&
isa<PackExpansionType>(parm->getType())) {
type = SemaRef.SubstType(
cast<PackExpansionType>(parm->getType())->getPattern(),
TemplateArgs, loc, parm->getDeclName());
} else {
type = SemaRef.SubstType(parm->getType(), TemplateArgs,
loc, parm->getDeclName());
}
assert(!type.isNull() && "type substitution failed for param type");
assert(!type->isDependentType() && "param type still dependent");
result = SemaRef.BuildExpressionFromDeclTemplateArgument(arg, type, loc);
if (!result.isInvalid()) type = result.get()->getType();
} else {
result = SemaRef.BuildExpressionFromIntegralTemplateArgument(arg, loc);
// Note that this type can be different from the type of 'result',
// e.g. if it's an enum type.
type = arg.getIntegralType();
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
}
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
if (result.isInvalid()) return ExprError();
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
Expr *resultExpr = result.take();
return SemaRef.Owned(new (SemaRef.Context)
SubstNonTypeTemplateParmExpr(type,
resultExpr->getValueKind(),
loc, parm, resultExpr));
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
}
Introduce a new expression kind, SubstNonTypeTemplateParmPackExpr, that captures the substitution of a non-type template argument pack for a non-type template parameter pack within a pack expansion that cannot be fully expanded. This follows the approach taken by SubstTemplateTypeParmPackType. llvm-svn: 123506
2011-01-15 09:15:58 +08:00
ExprResult
TemplateInstantiator::TransformSubstNonTypeTemplateParmPackExpr(
SubstNonTypeTemplateParmPackExpr *E) {
if (getSema().ArgumentPackSubstitutionIndex == -1) {
// We aren't expanding the parameter pack, so just return ourselves.
return getSema().Owned(E);
}
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
TemplateArgument Arg = E->getArgumentPack();
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
Create a new expression node, SubstNonTypeTemplateParmExpr, to represent a fully-substituted non-type template parameter. This should improve source fidelity, as well as being generically useful for diagnostics and such. llvm-svn: 135243
2011-07-15 13:09:51 +08:00
return transformNonTypeTemplateParmRef(E->getParameterPack(),
E->getParameterPackLocation(),
Arg);
Introduce a new expression kind, SubstNonTypeTemplateParmPackExpr, that captures the substitution of a non-type template argument pack for a non-type template parameter pack within a pack expansion that cannot be fully expanded. This follows the approach taken by SubstTemplateTypeParmPackType. llvm-svn: 123506
2011-01-15 09:15:58 +08:00
}
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
PR13811: Add a FunctionParmPackExpr node to handle references to function parameter packs where the reference is not being expanded but the pack has been. Previously, Clang would segfault in such cases. llvm-svn: 163672
2012-09-12 08:56:43 +08:00
ExprResult
TemplateInstantiator::RebuildParmVarDeclRefExpr(ParmVarDecl *PD,
SourceLocation Loc) {
DeclarationNameInfo NameInfo(PD->getDeclName(), Loc);
return getSema().BuildDeclarationNameExpr(CXXScopeSpec(), NameInfo, PD);
}
ExprResult
TemplateInstantiator::TransformFunctionParmPackExpr(FunctionParmPackExpr *E) {
if (getSema().ArgumentPackSubstitutionIndex != -1) {
// We can expand this parameter pack now.
ParmVarDecl *D = E->getExpansion(getSema().ArgumentPackSubstitutionIndex);
ValueDecl *VD = cast_or_null<ValueDecl>(TransformDecl(E->getExprLoc(), D));
if (!VD)
return ExprError();
return RebuildParmVarDeclRefExpr(cast<ParmVarDecl>(VD), E->getExprLoc());
}
QualType T = TransformType(E->getType());
if (T.isNull())
return ExprError();
// Transform each of the parameter expansions into the corresponding
// parameters in the instantiation of the function decl.
Remove useless 'llvm::' qualifier from names like StringRef and others that are brought into 'clang' namespace by clang/Basic/LLVM.h llvm-svn: 172323
2013-01-13 03:30:44 +08:00
SmallVector<Decl *, 8> Parms;
PR13811: Add a FunctionParmPackExpr node to handle references to function parameter packs where the reference is not being expanded but the pack has been. Previously, Clang would segfault in such cases. llvm-svn: 163672
2012-09-12 08:56:43 +08:00
Parms.reserve(E->getNumExpansions());
for (FunctionParmPackExpr::iterator I = E->begin(), End = E->end();
I != End; ++I) {
ParmVarDecl *D =
cast_or_null<ParmVarDecl>(TransformDecl(E->getExprLoc(), *I));
if (!D)
return ExprError();
Parms.push_back(D);
}
return FunctionParmPackExpr::Create(getSema().Context, T,
E->getParameterPack(),
E->getParameterPackLocation(), Parms);
}
ExprResult
TemplateInstantiator::TransformFunctionParmPackRefExpr(DeclRefExpr *E,
ParmVarDecl *PD) {
typedef LocalInstantiationScope::DeclArgumentPack DeclArgumentPack;
llvm::PointerUnion<Decl *, DeclArgumentPack *> *Found
= getSema().CurrentInstantiationScope->findInstantiationOf(PD);
assert(Found && "no instantiation for parameter pack");
Decl *TransformedDecl;
if (DeclArgumentPack *Pack = Found->dyn_cast<DeclArgumentPack *>()) {
// If this is a reference to a function parameter pack which we can substitute
// but can't yet expand, build a FunctionParmPackExpr for it.
if (getSema().ArgumentPackSubstitutionIndex == -1) {
QualType T = TransformType(E->getType());
if (T.isNull())
return ExprError();
return FunctionParmPackExpr::Create(getSema().Context, T, PD,
E->getExprLoc(), *Pack);
}
TransformedDecl = (*Pack)[getSema().ArgumentPackSubstitutionIndex];
} else {
TransformedDecl = Found->get<Decl*>();
}
// We have either an unexpanded pack or a specific expansion.
return RebuildParmVarDeclRefExpr(cast<ParmVarDecl>(TransformedDecl),
E->getExprLoc());
}
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
TemplateInstantiator::TransformDeclRefExpr(DeclRefExpr *E) {
NamedDecl *D = E->getDecl();
PR13811: Add a FunctionParmPackExpr node to handle references to function parameter packs where the reference is not being expanded but the pack has been. Previously, Clang would segfault in such cases. llvm-svn: 163672
2012-09-12 08:56:43 +08:00
// Handle references to non-type template parameters and non-type template
// parameter packs.
Extract a function to instantiate references to value template parameters. llvm-svn: 95491
2010-02-06 16:42:39 +08:00
if (NonTypeTemplateParmDecl *NTTP = dyn_cast<NonTypeTemplateParmDecl>(D)) {
if (NTTP->getDepth() < TemplateArgs.getNumLevels())
return TransformTemplateParmRefExpr(E, NTTP);
Implement "incremental" template instantiation for non-type template parameters and template type parameters, which occurs when substituting into the declarations of member templates inside class templates. This eliminates errors about our inability to "reduce non-type template parameter depth", fixing PR5311. Also fixes a bug when instantiating a template type parameter declaration in a member template, where we weren't properly reducing the template parameter's depth. LLVM's StringSwitch header now parses. llvm-svn: 85669
2009-11-01 01:21:17 +08:00
// We have a non-type template parameter that isn't fully substituted;
// FindInstantiatedDecl will find it in the local instantiation scope.
Refactor the template-instantiation logic for expressions into a generic tree transformation (also used for recanonicalization) and a small amount of template-instantiation-specific logic. llvm-svn: 78645
2009-08-11 13:31:07 +08:00
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
PR13811: Add a FunctionParmPackExpr node to handle references to function parameter packs where the reference is not being expanded but the pack has been. Previously, Clang would segfault in such cases. llvm-svn: 163672
2012-09-12 08:56:43 +08:00
// Handle references to function parameter packs.
if (ParmVarDecl *PD = dyn_cast<ParmVarDecl>(D))
if (PD->isParameterPack())
return TransformFunctionParmPackRefExpr(E, PD);
The refactor of implicit member access expressions means we don't need this horrible isAddressOfOperand hack in TreeTransform, since that syntactic information is managed by the initial parser callbacks now. That's enough insomniac commits for one night. llvm-svn: 90849
2009-12-08 17:21:05 +08:00
return TreeTransform<TemplateInstantiator>::TransformDeclRefExpr(E);
Refactor the template-instantiation logic for expressions into a generic tree transformation (also used for recanonicalization) and a small amount of template-instantiation-specific logic. llvm-svn: 78645
2009-08-11 13:31:07 +08:00
}
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult TemplateInstantiator::TransformCXXDefaultArgExpr(
The refactor of implicit member access expressions means we don't need this horrible isAddressOfOperand hack in TreeTransform, since that syntactic information is managed by the initial parser callbacks now. That's enough insomniac commits for one night. llvm-svn: 90849
2009-12-08 17:21:05 +08:00
CXXDefaultArgExpr *E) {
Special-case default argument expression in instantiation. This should fix PR4301. Doug, please double-check my assumptions. Read the PR for more details. llvm-svn: 86465
2009-11-08 21:56:19 +08:00
assert(!cast<FunctionDecl>(E->getParam()->getDeclContext())->
getDescribedFunctionTemplate() &&
"Default arg expressions are never formed in dependent cases.");
When we see a CXXDefaultArgExpr during template instantiation, rebuild the default argument so that we're sure to mark any referenced declarations. This gets us another little step closer to fixing PR5810. llvm-svn: 92078
2009-12-24 07:03:06 +08:00
return SemaRef.BuildCXXDefaultArgExpr(E->getUsedLocation(),
cast<FunctionDecl>(E->getParam()->getDeclContext()),
E->getParam());
Special-case default argument expression in instantiation. This should fix PR4301. Doug, please double-check my assumptions. Read the PR for more details. llvm-svn: 86465
2009-11-08 21:56:19 +08:00
}
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
QualType TemplateInstantiator::TransformFunctionProtoType(TypeLocBuilder &TLB,
Replace one hack with a different hack: strip out the ObjectType parameters to the Transform*Type functions and instead call out the specific cases where an object type and the unqualified lookup results are important. Fixes an assert and failed compile on a testcase from PR7248. llvm-svn: 118887
2010-11-12 16:19:04 +08:00
FunctionProtoTypeLoc TL) {
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
// We need a local instantiation scope for this function prototype.
Move more stuff out of Sema.h. llvm-svn: 112026
2010-08-25 13:32:35 +08:00
LocalInstantiationScope Scope(SemaRef, /*CombineWithOuterScope=*/true);
Replace one hack with a different hack: strip out the ObjectType parameters to the Transform*Type functions and instead call out the specific cases where an object type and the unqualified lookup results are important. Fixes an assert and failed compile on a testcase from PR7248. llvm-svn: 118887
2010-11-12 16:19:04 +08:00
return inherited::TransformFunctionProtoType(TLB, TL);
Maintain type source information for functions through template instantiation. Based on a patch by Enea Zaffanella! I found a way to reduce some of the redundancy between TreeTransform's "standard" FunctionProtoType transformation and TemplateInstantiator's override, and I killed off the old SubstFunctionType by adding type source info for the last cases where we were creating FunctionDecls without TSI (at least that get passed through template instantiation). llvm-svn: 98252
2010-03-11 17:03:00 +08:00
}
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
QualType TemplateInstantiator::TransformFunctionProtoType(TypeLocBuilder &TLB,
FunctionProtoTypeLoc TL,
CXXRecordDecl *ThisContext,
unsigned ThisTypeQuals) {
// We need a local instantiation scope for this function prototype.
LocalInstantiationScope Scope(SemaRef, /*CombineWithOuterScope=*/true);
return inherited::TransformFunctionProtoType(TLB, TL, ThisContext,
ThisTypeQuals);
}
Maintain type source information for functions through template instantiation. Based on a patch by Enea Zaffanella! I found a way to reduce some of the redundancy between TreeTransform's "standard" FunctionProtoType transformation and TemplateInstantiator's override, and I killed off the old SubstFunctionType by adding type source info for the last cases where we were creating FunctionDecls without TSI (at least that get passed through template instantiation). llvm-svn: 98252
2010-03-11 17:03:00 +08:00
ParmVarDecl *
Handle substitutions into function parameter packs whose patterns contain multiple parameter packs at different levels. llvm-svn: 123488
2011-01-15 06:40:04 +08:00
TemplateInstantiator::TransformFunctionTypeParam(ParmVarDecl *OldParm,
Store a parameter index and function prototype depth in every parameter node and use this to correctly mangle parameter references in function template signatures. A follow-up patch will improve the storage usage of these fields; here I've just done the lazy thing. llvm-svn: 130669
2011-05-02 06:35:37 +08:00
int indexAdjustment,
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
Optional<unsigned> NumExpansions,
When we're substituting into a function parameter pack and expect to get a function parameter pack (but don't due to weird substitutions), complain. Fixes the last bit of PR11848. llvm-svn: 148960
2012-01-26 00:15:54 +08:00
bool ExpectParameterPack) {
Store a parameter index and function prototype depth in every parameter node and use this to correctly mangle parameter references in function template signatures. A follow-up patch will improve the storage usage of these fields; here I've just done the lazy thing. llvm-svn: 130669
2011-05-02 06:35:37 +08:00
return SemaRef.SubstParmVarDecl(OldParm, TemplateArgs, indexAdjustment,
When we're substituting into a function parameter pack and expect to get a function parameter pack (but don't due to weird substitutions), complain. Fixes the last bit of PR11848. llvm-svn: 148960
2012-01-26 00:15:54 +08:00
NumExpansions, ExpectParameterPack);
Maintain type source information for functions through template instantiation. Based on a patch by Enea Zaffanella! I found a way to reduce some of the redundancy between TreeTransform's "standard" FunctionProtoType transformation and TemplateInstantiator's override, and I killed off the old SubstFunctionType by adding type source info for the last cases where we were creating FunctionDecls without TSI (at least that get passed through template instantiation). llvm-svn: 98252
2010-03-11 17:03:00 +08:00
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
QualType
Rewrite TreeTransform to transform types as DeclaratorInfos rather than as bare QualTypes. Don't actually exploit this yet. llvm-svn: 84716
2009-10-21 08:40:46 +08:00
TemplateInstantiator::TransformTemplateTypeParmType(TypeLocBuilder &TLB,
Replace one hack with a different hack: strip out the ObjectType parameters to the Transform*Type functions and instead call out the specific cases where an object type and the unqualified lookup results are important. Fixes an assert and failed compile on a testcase from PR7248. llvm-svn: 118887
2010-11-12 16:19:04 +08:00
TemplateTypeParmTypeLoc TL) {
Change QualType::getTypePtr() to return a const pointer, then change a thousand other things which were (generally inadvertantly) relying on that. llvm-svn: 123814
2011-01-19 14:33:43 +08:00
const TemplateTypeParmType *T = TL.getTypePtr();
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
if (T->getDepth() < TemplateArgs.getNumLevels()) {
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
// Replace the template type parameter with its corresponding
// template argument.
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
// If the corresponding template argument is NULL or doesn't exist, it's
// because we are performing instantiation from explicitly-specified
// template arguments in a function template class, but there were some
Cope with explicitly-specified function template arguments when there are fewer template arguments than there are template parameters for that function. llvm-svn: 74578
2009-07-01 08:28:38 +08:00
// arguments left unspecified.
Rewrite TreeTransform to transform types as DeclaratorInfos rather than as bare QualTypes. Don't actually exploit this yet. llvm-svn: 84716
2009-10-21 08:40:46 +08:00
if (!TemplateArgs.hasTemplateArgument(T->getDepth(), T->getIndex())) {
TemplateTypeParmTypeLoc NewTL
= TLB.push<TemplateTypeParmTypeLoc>(TL.getType());
NewTL.setNameLoc(TL.getNameLoc());
return TL.getType();
}
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Implement basic support for template instantiation of pack expansions whose patterns are template arguments. We can now instantiate, e.g., typedef tuple<pair<OuterTypes, InnerTypes>...> type; where OuterTypes and InnerTypes are template type parameter packs. There is a horrible inefficiency in TemplateArgumentLoc::getPackExpansionPattern(), where we need to create copies of TypeLoc data because our interfaces traffic in TypeSourceInfo pointers where they should traffic in TypeLocs instead. I've isolated in efficiency in this one routine; once we refactor our interfaces to traffic in TypeLocs, we can eliminate it. llvm-svn: 122278
2010-12-21 06:05:00 +08:00
TemplateArgument Arg = TemplateArgs(T->getDepth(), T->getIndex());
if (T->isParameterPack()) {
assert(Arg.getKind() == TemplateArgument::Pack &&
"Missing argument pack");
if (getSema().ArgumentPackSubstitutionIndex == -1) {
Start implementing support for substitution into pack expansions that involve template parameter packs at multiple template levels that occur within the signatures members of class templates (and partial specializations thereof). This is a work-in-progress that is deficient in several ways, notably: - It only works for template type parameter packs, but we need to also support non-type template parameter packs and template template parameter packs. - It doesn't keep track of the lengths of the substituted argument packs in the expansion, so it can't properly diagnose length mismatches. However, this is a concrete step in the right direction. llvm-svn: 123425
2011-01-14 10:55:32 +08:00
// We have the template argument pack, but we're not expanding the
// enclosing pack expansion yet. Just save the template argument
// pack for later substitution.
QualType Result
= getSema().Context.getSubstTemplateTypeParmPackType(T, Arg);
SubstTemplateTypeParmPackTypeLoc NewTL
= TLB.push<SubstTemplateTypeParmPackTypeLoc>(Result);
NewTL.setNameLoc(TL.getNameLoc());
return Result;
Implement basic support for template instantiation of pack expansions whose patterns are template arguments. We can now instantiate, e.g., typedef tuple<pair<OuterTypes, InnerTypes>...> type; where OuterTypes and InnerTypes are template type parameter packs. There is a horrible inefficiency in TemplateArgumentLoc::getPackExpansionPattern(), where we need to create copies of TypeLoc data because our interfaces traffic in TypeSourceInfo pointers where they should traffic in TypeLocs instead. I've isolated in efficiency in this one routine; once we refactor our interfaces to traffic in TypeLocs, we can eliminate it. llvm-svn: 122278
2010-12-21 06:05:00 +08:00
}
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
Implement basic support for template instantiation of pack expansions whose patterns are template arguments. We can now instantiate, e.g., typedef tuple<pair<OuterTypes, InnerTypes>...> type; where OuterTypes and InnerTypes are template type parameter packs. There is a horrible inefficiency in TemplateArgumentLoc::getPackExpansionPattern(), where we need to create copies of TypeLoc data because our interfaces traffic in TypeSourceInfo pointers where they should traffic in TypeLocs instead. I've isolated in efficiency in this one routine; once we refactor our interfaces to traffic in TypeLocs, we can eliminate it. llvm-svn: 122278
2010-12-21 06:05:00 +08:00
}
assert(Arg.getKind() == TemplateArgument::Type &&
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
"Template argument kind mismatch");
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
Implement basic support for template instantiation of pack expansions whose patterns are template arguments. We can now instantiate, e.g., typedef tuple<pair<OuterTypes, InnerTypes>...> type; where OuterTypes and InnerTypes are template type parameter packs. There is a horrible inefficiency in TemplateArgumentLoc::getPackExpansionPattern(), where we need to create copies of TypeLoc data because our interfaces traffic in TypeSourceInfo pointers where they should traffic in TypeLocs instead. I've isolated in efficiency in this one routine; once we refactor our interfaces to traffic in TypeLocs, we can eliminate it. llvm-svn: 122278
2010-12-21 06:05:00 +08:00
QualType Replacement = Arg.getAsType();
When performing template-substitution into a type, don't just replace the TemplateTypeParmType with the substituted type directly; instead, replace it with a SubstTemplateTypeParmType which will note that the type was originally written as a template type parameter. This makes it reasonable to preserve source information even through template substitution. Also define the new SubstTemplateTypeParmType class, obviously. For consistency with current behavior, we stringize these types as if they were the underlying type. I'm not sure this is the right thing to do. At any rate, I paled at adding yet another clause to the don't-desugar 'if' statement, so I extracted a function to do it. The new function also does The Right Thing more often, I think: e.g. if we have a chain of typedefs leading to a vector type, we will now desugar all but the last one. llvm-svn: 84412
2009-10-18 17:09:24 +08:00
// TODO: only do this uniquing once, at the start of instantiation.
Rewrite TreeTransform to transform types as DeclaratorInfos rather than as bare QualTypes. Don't actually exploit this yet. llvm-svn: 84716
2009-10-21 08:40:46 +08:00
QualType Result
= getSema().Context.getSubstTemplateTypeParmType(T, Replacement);
SubstTemplateTypeParmTypeLoc NewTL
= TLB.push<SubstTemplateTypeParmTypeLoc>(Result);
NewTL.setNameLoc(TL.getNameLoc());
return Result;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
}
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
// The template type parameter comes from an inner template (e.g.,
// the template parameter list of a member template inside the
// template we are instantiating). Create a new template type
// parameter with the template "level" reduced by one.
Re-applies the patch first applied way back in r106099, with accompanying fixes to make it work today. The core of this patch is to provide a link from a TemplateTypeParmType back to the TemplateTypeParmDecl node which declared it. This in turn provides much more precise information about the type, where it came from, and how it functions for AST consumers. To make the patch work almost a year after its first attempt, it needed serialization support, and it now retains the old getName() interface. Finally, it requires us to not attempt to instantiate the type in an unsupported friend decl -- specifically those coming from template friend decls but which refer to a specific type through a dependent name. A cleaner representation of the last item would be to build FriendTemplateDecl nodes for these, storing their template parameters etc, and to perform proper instantation of them like any other template declaration. They can still be flagged as unsupported for the purpose of access checking, etc. This passed an asserts-enabled bootstrap for me, and the reduced test case mentioned in the original review thread no longer causes issues, likely fixed at somewhere amidst the 24k revisions that have elapsed. llvm-svn: 130628
2011-05-01 08:51:33 +08:00
TemplateTypeParmDecl *NewTTPDecl = 0;
if (TemplateTypeParmDecl *OldTTPDecl = T->getDecl())
NewTTPDecl = cast_or_null<TemplateTypeParmDecl>(
TransformDecl(TL.getNameLoc(), OldTTPDecl));
Rewrite TreeTransform to transform types as DeclaratorInfos rather than as bare QualTypes. Don't actually exploit this yet. llvm-svn: 84716
2009-10-21 08:40:46 +08:00
QualType Result
= getSema().Context.getTemplateTypeParmType(T->getDepth()
- TemplateArgs.getNumLevels(),
T->getIndex(),
T->isParameterPack(),
Re-applies the patch first applied way back in r106099, with accompanying fixes to make it work today. The core of this patch is to provide a link from a TemplateTypeParmType back to the TemplateTypeParmDecl node which declared it. This in turn provides much more precise information about the type, where it came from, and how it functions for AST consumers. To make the patch work almost a year after its first attempt, it needed serialization support, and it now retains the old getName() interface. Finally, it requires us to not attempt to instantiate the type in an unsupported friend decl -- specifically those coming from template friend decls but which refer to a specific type through a dependent name. A cleaner representation of the last item would be to build FriendTemplateDecl nodes for these, storing their template parameters etc, and to perform proper instantation of them like any other template declaration. They can still be flagged as unsupported for the purpose of access checking, etc. This passed an asserts-enabled bootstrap for me, and the reduced test case mentioned in the original review thread no longer causes issues, likely fixed at somewhere amidst the 24k revisions that have elapsed. llvm-svn: 130628
2011-05-01 08:51:33 +08:00
NewTTPDecl);
Rewrite TreeTransform to transform types as DeclaratorInfos rather than as bare QualTypes. Don't actually exploit this yet. llvm-svn: 84716
2009-10-21 08:40:46 +08:00
TemplateTypeParmTypeLoc NewTL = TLB.push<TemplateTypeParmTypeLoc>(Result);
NewTL.setNameLoc(TL.getNameLoc());
return Result;
Start implementing support for substitution into pack expansions that involve template parameter packs at multiple template levels that occur within the signatures members of class templates (and partial specializations thereof). This is a work-in-progress that is deficient in several ways, notably: - It only works for template type parameter packs, but we need to also support non-type template parameter packs and template template parameter packs. - It doesn't keep track of the lengths of the substituted argument packs in the expansion, so it can't properly diagnose length mismatches. However, this is a concrete step in the right direction. llvm-svn: 123425
2011-01-14 10:55:32 +08:00
}
QualType
TemplateInstantiator::TransformSubstTemplateTypeParmPackType(
TypeLocBuilder &TLB,
SubstTemplateTypeParmPackTypeLoc TL) {
if (getSema().ArgumentPackSubstitutionIndex == -1) {
// We aren't expanding the parameter pack, so just return ourselves.
SubstTemplateTypeParmPackTypeLoc NewTL
= TLB.push<SubstTemplateTypeParmPackTypeLoc>(TL.getType());
NewTL.setNameLoc(TL.getNameLoc());
return TL.getType();
}
Fix for template substitution with packs. When we see a pack, and replace it with a template argument which is also a pack, we want to use the pack pattern, not the expanded pack. The caller should take care of expanding the pack afterwards. Fixes PR16646. llvm-svn: 186713
2013-07-20 03:40:38 +08:00
TemplateArgument Arg = TL.getTypePtr()->getArgumentPack();
Arg = getPackSubstitutedTemplateArgument(getSema(), Arg);
QualType Result = Arg.getAsType();
Start implementing support for substitution into pack expansions that involve template parameter packs at multiple template levels that occur within the signatures members of class templates (and partial specializations thereof). This is a work-in-progress that is deficient in several ways, notably: - It only works for template type parameter packs, but we need to also support non-type template parameter packs and template template parameter packs. - It doesn't keep track of the lengths of the substituted argument packs in the expansion, so it can't properly diagnose length mismatches. However, this is a concrete step in the right direction. llvm-svn: 123425
2011-01-14 10:55:32 +08:00
Result = getSema().Context.getSubstTemplateTypeParmType(
TL.getTypePtr()->getReplacedParameter(),
Result);
SubstTemplateTypeParmTypeLoc NewTL
= TLB.push<SubstTemplateTypeParmTypeLoc>(Result);
NewTL.setNameLoc(TL.getNameLoc());
return Result;
Implement template instantiation for pointer, reference, and (some) array types. Semantic checking for the construction of these types has been factored out of GetTypeForDeclarator and into separate subroutines (BuildPointerType, BuildReferenceType, BuildArrayType). We'll be doing the same thing for all other types (and declarations and expressions). As part of this, moved the type-instantiation functions into a class in an anonymous namespace. llvm-svn: 65663
2009-02-28 08:25:32 +08:00
}
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
/// \brief Perform substitution on the type T with a given set of template
/// arguments.
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
///
/// This routine substitutes the given template arguments into the
/// type T and produces the instantiated type.
///
/// \param T the type into which the template arguments will be
/// substituted. If this type is not dependent, it will be returned
/// immediately.
///
Still more Doxygen documentation fixes: * Escape #, < and @ symbols where Doxygen would try to interpret them; * Fix several function param documentation where names had got out of sync; * Delete param documentation referring to parameters that no longer exist. llvm-svn: 158472
2012-06-15 05:40:34 +08:00
/// \param Args the template arguments that will be
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
/// substituted for the top-level template parameters within T.
///
/// \param Loc the location in the source code where this substitution
/// is being performed. It will typically be the location of the
/// declarator (if we're instantiating the type of some declaration)
/// or the location of the type in the source code (if, e.g., we're
/// instantiating the type of a cast expression).
///
/// \param Entity the name of the entity associated with a declaration
/// being instantiated (if any). May be empty to indicate that there
/// is no such entity (if, e.g., this is a type that occurs as part of
/// a cast expression) or that the entity has no name (e.g., an
/// unnamed function parameter).
///
/// \returns If the instantiation succeeds, the instantiated
/// type. Otherwise, produces diagnostics and returns a NULL type.
DeclaratorInfo -> TypeSourceInfo. Makes an effort to rename associated variables, but the results are imperfect. For posterity, I did: cat <<EOF > $cmdfile s/DeclaratorInfo/TypeSourceInfo/g s/DInfo/TInfo/g s/TypeTypeSourceInfo/TypeSourceInfo/g s/SourceTypeSourceInfo/TypeSourceInfo/g EOF find lib -name '*.cpp' -not -path 'lib/Parse/*' -exec sed -i '' -f $cmdfile '{}' \; find lib -name '*.h' -exec sed -i '' -f $cmdfile '{}' \; find include -name '*.h' -not -path 'include/clang/Parse/*' -not -path 'include/clang/Basic/*' -exec sed -i '' -f $cmdfile '{}' \; llvm-svn: 90743
2009-12-07 10:54:59 +08:00
TypeSourceInfo *Sema::SubstType(TypeSourceInfo *T,
Clone Sema::SubstType for DeclaratorInfos. llvm-svn: 84724
2009-10-21 08:58:09 +08:00
const MultiLevelTemplateArgumentList &Args,
SourceLocation Loc,
DeclarationName Entity) {
assert(!ActiveTemplateInstantiations.empty() &&
"Cannot perform an instantiation without some context on the "
"instantiation stack");
Introduce the notion of instantiation dependence into Clang's AST. A type/expression/template argument/etc. is instantiation-dependent if it somehow involves a template parameter, even if it doesn't meet the requirements for the more common kinds of dependence (dependent type, type-dependent expression, value-dependent expression). When we see an instantiation-dependent type, we know we always need to perform substitution into that instantiation-dependent type. This keeps us from short-circuiting evaluation in places where we shouldn't, and lets us properly implement C++0x [temp.type]p2. In theory, this would also allow us to properly mangle instantiation-dependent-but-not-dependent decltype types per the Itanium C++ ABI, but we aren't quite there because we still mangle based on the canonical type in cases like, e.g., template<unsigned> struct A { }; template<typename T> void f(A<sizeof(sizeof(decltype(T() + T())))>) { } template void f<int>(A<sizeof(sizeof(int))>); and therefore get the wrong answer. llvm-svn: 134225
2011-07-01 09:22:09 +08:00
if (!T->getType()->isInstantiationDependentType() &&
Make sure that we instantiate variably modified types, even if they aren't dependent. Fixes <rdar://problem/8020206>. llvm-svn: 104511
2010-05-25 01:22:01 +08:00
!T->getType()->isVariablyModifiedType())
Clone Sema::SubstType for DeclaratorInfos. llvm-svn: 84724
2009-10-21 08:58:09 +08:00
return T;
TemplateInstantiator Instantiator(*this, Args, Loc, Entity);
return Instantiator.TransformType(T);
}
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
TypeSourceInfo *Sema::SubstType(TypeLoc TL,
const MultiLevelTemplateArgumentList &Args,
SourceLocation Loc,
DeclarationName Entity) {
assert(!ActiveTemplateInstantiations.empty() &&
"Cannot perform an instantiation without some context on the "
"instantiation stack");
if (TL.getType().isNull())
return 0;
Introduce the notion of instantiation dependence into Clang's AST. A type/expression/template argument/etc. is instantiation-dependent if it somehow involves a template parameter, even if it doesn't meet the requirements for the more common kinds of dependence (dependent type, type-dependent expression, value-dependent expression). When we see an instantiation-dependent type, we know we always need to perform substitution into that instantiation-dependent type. This keeps us from short-circuiting evaluation in places where we shouldn't, and lets us properly implement C++0x [temp.type]p2. In theory, this would also allow us to properly mangle instantiation-dependent-but-not-dependent decltype types per the Itanium C++ ABI, but we aren't quite there because we still mangle based on the canonical type in cases like, e.g., template<unsigned> struct A { }; template<typename T> void f(A<sizeof(sizeof(decltype(T() + T())))>) { } template void f<int>(A<sizeof(sizeof(int))>); and therefore get the wrong answer. llvm-svn: 134225
2011-07-01 09:22:09 +08:00
if (!TL.getType()->isInstantiationDependentType() &&
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
!TL.getType()->isVariablyModifiedType()) {
// FIXME: Make a copy of the TypeLoc data here, so that we can
// return a new TypeSourceInfo. Inefficient!
TypeLocBuilder TLB;
TLB.pushFullCopy(TL);
return TLB.getTypeSourceInfo(Context, TL.getType());
}
TemplateInstantiator Instantiator(*this, Args, Loc, Entity);
TypeLocBuilder TLB;
TLB.reserve(TL.getFullDataSize());
QualType Result = Instantiator.TransformType(TLB, TL);
if (Result.isNull())
return 0;
return TLB.getTypeSourceInfo(Context, Result);
}
Clone Sema::SubstType for DeclaratorInfos. llvm-svn: 84724
2009-10-21 08:58:09 +08:00
/// Deprecated form of the above.
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
QualType Sema::SubstType(QualType T,
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs,
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
SourceLocation Loc, DeclarationName Entity) {
Extend the notion of active template instantiations to include the context of a template-id for which we need to instantiate default template arguments. In the TextDiagnosticPrinter, don't suppress the caret diagnostic if we are producing a non-note diagnostic that follows a note diagnostic with the same location, because notes are (conceptually) a part of the warning or error that comes before them. llvm-svn: 66572
2009-03-11 04:44:00 +08:00
assert(!ActiveTemplateInstantiations.empty() &&
"Cannot perform an instantiation without some context on the "
"instantiation stack");
Make sure that we instantiate variably modified types, even if they aren't dependent. Fixes <rdar://problem/8020206>. llvm-svn: 104511
2010-05-25 01:22:01 +08:00
// If T is not a dependent type or a variably-modified type, there
// is nothing to do.
Introduce the notion of instantiation dependence into Clang's AST. A type/expression/template argument/etc. is instantiation-dependent if it somehow involves a template parameter, even if it doesn't meet the requirements for the more common kinds of dependence (dependent type, type-dependent expression, value-dependent expression). When we see an instantiation-dependent type, we know we always need to perform substitution into that instantiation-dependent type. This keeps us from short-circuiting evaluation in places where we shouldn't, and lets us properly implement C++0x [temp.type]p2. In theory, this would also allow us to properly mangle instantiation-dependent-but-not-dependent decltype types per the Itanium C++ ABI, but we aren't quite there because we still mangle based on the canonical type in cases like, e.g., template<unsigned> struct A { }; template<typename T> void f(A<sizeof(sizeof(decltype(T() + T())))>) { } template void f<int>(A<sizeof(sizeof(int))>); and therefore get the wrong answer. llvm-svn: 134225
2011-07-01 09:22:09 +08:00
if (!T->isInstantiationDependentType() && !T->isVariablyModifiedType())
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
return T;
Refactor template instantiation for types into a generic tree transformation template (TreeTransform) that handles the transformation and reconstruction of AST nodes. Template instantiation for types is a (relatively small) customization of the generic tree transformation. llvm-svn: 78071
2009-08-05 00:50:30 +08:00
TemplateInstantiator Instantiator(*this, TemplateArgs, Loc, Entity);
return Instantiator.TransformType(T);
Implement the basic approach for instantiating types, with a lot of FIXME'd stubs for those types we don't yet know how to instantiate (everything that isn't a template parameter!). We now instantiate default arguments for template type parameters when needed. This will be our testbed while I fill out the remaining type-instantiation logic. llvm-svn: 65649
2009-02-28 03:31:52 +08:00
}
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
static bool NeedsInstantiationAsFunctionType(TypeSourceInfo *T) {
Introduce the notion of instantiation dependence into Clang's AST. A type/expression/template argument/etc. is instantiation-dependent if it somehow involves a template parameter, even if it doesn't meet the requirements for the more common kinds of dependence (dependent type, type-dependent expression, value-dependent expression). When we see an instantiation-dependent type, we know we always need to perform substitution into that instantiation-dependent type. This keeps us from short-circuiting evaluation in places where we shouldn't, and lets us properly implement C++0x [temp.type]p2. In theory, this would also allow us to properly mangle instantiation-dependent-but-not-dependent decltype types per the Itanium C++ ABI, but we aren't quite there because we still mangle based on the canonical type in cases like, e.g., template<unsigned> struct A { }; template<typename T> void f(A<sizeof(sizeof(decltype(T() + T())))>) { } template void f<int>(A<sizeof(sizeof(int))>); and therefore get the wrong answer. llvm-svn: 134225
2011-07-01 09:22:09 +08:00
if (T->getType()->isInstantiationDependentType() ||
T->getType()->isVariablyModifiedType())
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
return true;
Added missing IgnoreParens(). llvm-svn: 121795
2010-12-15 06:11:44 +08:00
TypeLoc TL = T->getTypeLoc().IgnoreParens();
Replace TypeLoc llvm::cast support to be well-defined. The TypeLoc hierarchy used the llvm::cast machinery to perform undefined behavior by casting pointers/references to TypeLoc objects to derived types and then using the derived copy constructors (or even returning pointers to derived types that actually point to the original TypeLoc object). Some context is in this thread: http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-December/056804.html Though it's spread over a few months which can be hard to read in the mail archive. llvm-svn: 175462
2013-02-19 06:06:02 +08:00
if (!TL.getAs<FunctionProtoTypeLoc>())
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
return false;
Replace TypeLoc llvm::cast support to be well-defined. The TypeLoc hierarchy used the llvm::cast machinery to perform undefined behavior by casting pointers/references to TypeLoc objects to derived types and then using the derived copy constructors (or even returning pointers to derived types that actually point to the original TypeLoc object). Some context is in this thread: http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-December/056804.html Though it's spread over a few months which can be hard to read in the mail archive. llvm-svn: 175462
2013-02-19 06:06:02 +08:00
FunctionProtoTypeLoc FP = TL.castAs<FunctionProtoTypeLoc>();
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
for (unsigned I = 0, E = FP.getNumArgs(); I != E; ++I) {
ParmVarDecl *P = FP.getArg(I);
Fix declaring class template methods with an attributed typedef This change unifies the logic for template instantiation of methods and functions declared with typedefs. It ensures that SubstFunctionType() always fills the Params out param with non-null ParmVarDecls or returns null. Reviewers: rsmith Differential Revision: http://llvm-reviews.chandlerc.com/D1135 llvm-svn: 187528
2013-08-01 05:00:18 +08:00
// This must be synthesized from a typedef.
if (!P) continue;
When determining whether we need to instantiate a function type, also consider whether any of the parameter types (as written, prior to decay) are dependent. Fixes PR9880 and <rdar://problem/9408413>. llvm-svn: 131099
2011-05-10 04:45:16 +08:00
// The parameter's type as written might be dependent even if the
// decayed type was not dependent.
if (TypeSourceInfo *TSInfo = P->getTypeSourceInfo())
Introduce the notion of instantiation dependence into Clang's AST. A type/expression/template argument/etc. is instantiation-dependent if it somehow involves a template parameter, even if it doesn't meet the requirements for the more common kinds of dependence (dependent type, type-dependent expression, value-dependent expression). When we see an instantiation-dependent type, we know we always need to perform substitution into that instantiation-dependent type. This keeps us from short-circuiting evaluation in places where we shouldn't, and lets us properly implement C++0x [temp.type]p2. In theory, this would also allow us to properly mangle instantiation-dependent-but-not-dependent decltype types per the Itanium C++ ABI, but we aren't quite there because we still mangle based on the canonical type in cases like, e.g., template<unsigned> struct A { }; template<typename T> void f(A<sizeof(sizeof(decltype(T() + T())))>) { } template void f<int>(A<sizeof(sizeof(int))>); and therefore get the wrong answer. llvm-svn: 134225
2011-07-01 09:22:09 +08:00
if (TSInfo->getType()->isInstantiationDependentType())
When determining whether we need to instantiate a function type, also consider whether any of the parameter types (as written, prior to decay) are dependent. Fixes PR9880 and <rdar://problem/9408413>. llvm-svn: 131099
2011-05-10 04:45:16 +08:00
return true;
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
// TODO: currently we always rebuild expressions. When we
// properly get lazier about this, we should use the same
// logic to avoid rebuilding prototypes here.
hasInit() -> hasDefaultArg() llvm-svn: 122905
2011-01-06 05:14:17 +08:00
if (P->hasDefaultArg())
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
return true;
}
return false;
}
/// A form of SubstType intended specifically for instantiating the
/// type of a FunctionDecl. Its purpose is solely to force the
/// instantiation of default-argument expressions.
TypeSourceInfo *Sema::SubstFunctionDeclType(TypeSourceInfo *T,
const MultiLevelTemplateArgumentList &Args,
SourceLocation Loc,
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
DeclarationName Entity,
CXXRecordDecl *ThisContext,
unsigned ThisTypeQuals) {
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
assert(!ActiveTemplateInstantiations.empty() &&
"Cannot perform an instantiation without some context on the "
"instantiation stack");
if (!NeedsInstantiationAsFunctionType(T))
return T;
TemplateInstantiator Instantiator(*this, Args, Loc, Entity);
TypeLocBuilder TLB;
TypeLoc TL = T->getTypeLoc();
TLB.reserve(TL.getFullDataSize());
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
QualType Result;
Replace TypeLoc llvm::cast support to be well-defined. The TypeLoc hierarchy used the llvm::cast machinery to perform undefined behavior by casting pointers/references to TypeLoc objects to derived types and then using the derived copy constructors (or even returning pointers to derived types that actually point to the original TypeLoc object). Some context is in this thread: http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-December/056804.html Though it's spread over a few months which can be hard to read in the mail archive. llvm-svn: 175462
2013-02-19 06:06:02 +08:00
if (FunctionProtoTypeLoc Proto = TL.getAs<FunctionProtoTypeLoc>()) {
Result = Instantiator.TransformFunctionProtoType(TLB, Proto, ThisContext,
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
ThisTypeQuals);
} else {
Result = Instantiator.TransformType(TLB, TL);
}
Instantiate default argument expressions even if their associated parameter type isn't dependent. Fixes rdar://problem/7838962. llvm-svn: 100871
2010-04-10 01:38:44 +08:00
if (Result.isNull())
return 0;
return TLB.getTypeSourceInfo(Context, Result);
}
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
ParmVarDecl *Sema::SubstParmVarDecl(ParmVarDecl *OldParm,
Handle substitutions into function parameter packs whose patterns contain multiple parameter packs at different levels. llvm-svn: 123488
2011-01-15 06:40:04 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs,
Store a parameter index and function prototype depth in every parameter node and use this to correctly mangle parameter references in function template signatures. A follow-up patch will improve the storage usage of these fields; here I've just done the lazy thing. llvm-svn: 130669
2011-05-02 06:35:37 +08:00
int indexAdjustment,
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
Optional<unsigned> NumExpansions,
When we're substituting into a function parameter pack and expect to get a function parameter pack (but don't due to weird substitutions), complain. Fixes the last bit of PR11848. llvm-svn: 148960
2012-01-26 00:15:54 +08:00
bool ExpectParameterPack) {
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
TypeSourceInfo *OldDI = OldParm->getTypeSourceInfo();
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
TypeSourceInfo *NewDI = 0;
TypeLoc OldTL = OldDI->getTypeLoc();
Replace TypeLoc llvm::cast support to be well-defined. The TypeLoc hierarchy used the llvm::cast machinery to perform undefined behavior by casting pointers/references to TypeLoc objects to derived types and then using the derived copy constructors (or even returning pointers to derived types that actually point to the original TypeLoc object). Some context is in this thread: http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-December/056804.html Though it's spread over a few months which can be hard to read in the mail archive. llvm-svn: 175462
2013-02-19 06:06:02 +08:00
if (PackExpansionTypeLoc ExpansionTL = OldTL.getAs<PackExpansionTypeLoc>()) {
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
// We have a function parameter pack. Substitute into the pattern of the
// expansion.
NewDI = SubstType(ExpansionTL.getPatternLoc(), TemplateArgs,
OldParm->getLocation(), OldParm->getDeclName());
if (!NewDI)
return 0;
if (NewDI->getType()->containsUnexpandedParameterPack()) {
// We still have unexpanded parameter packs, which means that
// our function parameter is still a function parameter pack.
// Therefore, make its type a pack expansion type.
Keep track of the number of expansions to be produced from a type pack expansion, when it is known due to the substitution of an out parameter pack. This allows us to properly handle substitution into pack expansions that involve multiple parameter packs at different template parameter levels, even when this substitution happens one level at a time (as with partial specializations of member class templates and the signatures of member function templates). Note that the diagnostic we provide when there is an arity mismatch between an outer parameter pack and an inner parameter pack in this case isn't as clear as the normal diagnostic for an arity mismatch. However, this doesn't matter because these cases are very, very rare and (even then) only typically occur in a SFINAE context. The other kinds of pack expansions (expression, template, etc.) still need to support optional tracking of the number of expansions, and we need the moral equivalent of SubstTemplateTypeParmPackType for substituted argument packs of template template and non-type template parameters. llvm-svn: 123448
2011-01-15 01:04:44 +08:00
NewDI = CheckPackExpansion(NewDI, ExpansionTL.getEllipsisLoc(),
Handle substitutions into function parameter packs whose patterns contain multiple parameter packs at different levels. llvm-svn: 123488
2011-01-15 06:40:04 +08:00
NumExpansions);
When we're substituting into a function parameter pack and expect to get a function parameter pack (but don't due to weird substitutions), complain. Fixes the last bit of PR11848. llvm-svn: 148960
2012-01-26 00:15:54 +08:00
} else if (ExpectParameterPack) {
// We expected to get a parameter pack but didn't (because the type
// itself is not a pack expansion type), so complain. This can occur when
// the substitution goes through an alias template that "loses" the
// pack expansion.
Diag(OldParm->getLocation(),
diag::err_function_parameter_pack_without_parameter_packs)
<< NewDI->getType();
return 0;
}
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
} else {
NewDI = SubstType(OldDI, TemplateArgs, OldParm->getLocation(),
OldParm->getDeclName());
}
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
if (!NewDI)
return 0;
if (NewDI->getType()->isVoidType()) {
Diag(OldParm->getLocation(), diag::err_param_with_void_type);
return 0;
}
ParmVarDecl *NewParm = CheckParameter(Context.getTranslationUnitDecl(),
Fixed source range for all DeclaratorDecl's. llvm-svn: 127225
2011-03-08 16:55:46 +08:00
OldParm->getInnerLocStart(),
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
OldParm->getLocation(),
Fixed source range for all DeclaratorDecl's. llvm-svn: 127225
2011-03-08 16:55:46 +08:00
OldParm->getIdentifier(),
NewDI->getType(), NewDI,
Add 178663 back. http://lab.llvm.org:8011/builders/clang-x86_64-darwin10-gdb went back green before it processed the reverted 178663, so it could not have been the culprit. Revert "Revert 178663." This reverts commit 4f8a3eb2ce5d4ba422483439e20c8cbb4d953a41. llvm-svn: 178682
2013-04-04 03:27:57 +08:00
OldParm->getStorageClass());
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
if (!NewParm)
return 0;
Revert r104106; it's breaking linking of Boost.Serialization. llvm-svn: 104121
2010-05-20 01:02:24 +08:00
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
// Mark the (new) default argument as uninstantiated (if any).
if (OldParm->hasUninstantiatedDefaultArg()) {
Expr *Arg = OldParm->getUninstantiatedDefaultArg();
NewParm->setUninstantiatedDefaultArg(Arg);
If we end up instantiating a function parameter whose default argument has not yet been parsed, note that the default argument hasn't been parsed and keep track of all of the instantiations of that function parameter. When its default argument does get parsed, imbue the instantiations with that default argument. Fixes PR8245. llvm-svn: 116324
2010-10-13 02:23:32 +08:00
} else if (OldParm->hasUnparsedDefaultArg()) {
NewParm->setUnparsedDefaultArg();
UnparsedDefaultArgInstantiations[OldParm].push_back(NewParm);
PR12710 - broken default argument handling for templates. I broke this in r155838 by not actually instantiating non-dependent default arg expressions. The motivation for that change was to avoid producing duplicate conversion warnings for such default args (we produce them once when we parse the template - there's no need to produce them at each instantiation) but without actually instantiating the default arg, things break in weird ways. Technically, I think we could still get the right diagnostic experience without the bugs if we instantiated the non-dependent args (for non-dependent params only) immediately, rather than lazily. But I'm not sure if such a refactoring/ change would be desirable so here's the conservative fix for now. llvm-svn: 155893
2012-05-01 14:05:57 +08:00
} else if (Expr *Arg = OldParm->getDefaultArg())
// FIXME: if we non-lazily instantiated non-dependent default args for
// non-dependent parameter types we could remove a bunch of duplicate
// conversion warnings for such arguments.
NewParm->setUninstantiatedDefaultArg(Arg);
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
NewParm->setHasInheritedDefaultArg(OldParm->hasInheritedDefaultArg());
When we're substituting into a function parameter pack and expect to get a function parameter pack (but don't due to weird substitutions), complain. Fixes the last bit of PR11848. llvm-svn: 148960
2012-01-26 00:15:54 +08:00
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
if (OldParm->isParameterPack() && !NewParm->isParameterPack()) {
Fix PR11848: decree that an alias template contains an unexpanded parameter pack iff its substitution contains an unexpanded parameter pack. This has the effect that we now reject declarations such as this (which we used to crash when expanding): template<typename T> using Int = int; template<typename ...Ts> void f(Int<Ts> ...ints); The standard is inconsistent on how this case should be treated. llvm-svn: 148905
2012-01-25 10:14:59 +08:00
// Add the new parameter to the instantiated parameter pack.
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
CurrentInstantiationScope->InstantiatedLocalPackArg(OldParm, NewParm);
} else {
// Introduce an Old -> New mapping
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
CurrentInstantiationScope->InstantiatedLocal(OldParm, NewParm);
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
}
Initial implementation of function parameter packs. This implementation allows: 1) Declaration of function parameter packs 2) Instantiation of function parameter packs within function types. 3) Template argument deduction of function parameter packs when matching two function types. We're missing all of the important template-instantiation logic for function template definitions, along with template argument deduction from the argument list of a function call, so don't even think of trying to use these for real yet. llvm-svn: 122926
2011-01-06 07:12:31 +08:00
Fix http://llvm.org/PR7660 A ParmVarDecl instantiated from a FunctionProtoType may have Record as DeclContext, in which case isStaticDataMember() will erroneously return true. llvm-svn: 108692
2010-07-19 18:14:41 +08:00
// FIXME: OldParm may come from a FunctionProtoType, in which case CurContext
// can be anything, is this right ?
Set DeclContext of ParamVarDecl only. No need to set that of VarDecl for block variables (they are already set). Per Doug's comment. llvm-svn: 108273
2010-07-14 05:05:02 +08:00
NewParm->setDeclContext(CurContext);
Store a parameter index and function prototype depth in every parameter node and use this to correctly mangle parameter references in function template signatures. A follow-up patch will improve the storage usage of these fields; here I've just done the lazy thing. llvm-svn: 130669
2011-05-02 06:35:37 +08:00
NewParm->setScopeInfo(OldParm->getFunctionScopeDepth(),
OldParm->getFunctionScopeIndex() + indexAdjustment);
Sema: Preserve attributes on parameters in instantiated function templates. This was causing correctness issues for ARC and the static analyzer when a function template has "consumed" Objective-C object parameters (i.e. parameters that will be released by the function before returning). The fix is threefold: (1) Actually copy over the attributes from old ParmVarDecls to new ones. (2) Have Sema::BuildFunctionType only work for building FunctionProtoTypes, which it was doing anyway. This allows us to pass an ExtProtoInfo instead of a plain ExtInfo and several flags. (3) Drop param attributes as part of StripImplicitInstantiation, which is used when an implicit instantiation is followed by an explicit one. <rdar://problem/12685622> llvm-svn: 176728
2013-03-09 06:25:36 +08:00
InstantiateAttrs(TemplateArgs, OldParm, NewParm);
Be sure to instantiate the parameters of a function, even when the function's type is (strictly speaking) non-dependent. This ensures that, e.g., default function arguments get instantiated properly. And, since I couldn't resist, collapse the two implementations of function-parameter instantiation into calls to a single, new function (Sema::SubstParmVarDecl), since the two had nearly identical code (and each had bugs the other didn't!). More importantly, factored out the semantic analysis of a parameter declaration into Sema::CheckParameter, which is called both by Sema::ActOnParamDeclarator (when parameters are parsed) and when a parameter is instantiated. Previously, we were missing some Objective-C and address-space checks on instantiated function parameters. Fixes PR6733. llvm-svn: 101029
2010-04-12 15:48:19 +08:00
return NewParm;
}
Factor out the template transformation of a sequence of function parameters into parameter types, so that substitution of explicitly-specified function template arguments uses the same path. This enables the use of explicitly-specified function template arguments with variadic templates. llvm-svn: 122986
2011-01-07 08:20:55 +08:00
/// \brief Substitute the given template arguments into the given set of
/// parameters, producing the set of parameter types that would be generated
/// from such a substitution.
bool Sema::SubstParmTypes(SourceLocation Loc,
ParmVarDecl **Params, unsigned NumParams,
const MultiLevelTemplateArgumentList &TemplateArgs,
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVectorImpl<QualType> &ParamTypes,
SmallVectorImpl<ParmVarDecl *> *OutParams) {
Factor out the template transformation of a sequence of function parameters into parameter types, so that substitution of explicitly-specified function template arguments uses the same path. This enables the use of explicitly-specified function template arguments with variadic templates. llvm-svn: 122986
2011-01-07 08:20:55 +08:00
assert(!ActiveTemplateInstantiations.empty() &&
"Cannot perform an instantiation without some context on the "
"instantiation stack");
TemplateInstantiator Instantiator(*this, TemplateArgs, Loc,
DeclarationName());
return Instantiator.TransformFunctionTypeParams(Loc, Params, NumParams, 0,
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
ParamTypes, OutParams);
Factor out the template transformation of a sequence of function parameters into parameter types, so that substitution of explicitly-specified function template arguments uses the same path. This enables the use of explicitly-specified function template arguments with variadic templates. llvm-svn: 122986
2011-01-07 08:20:55 +08:00
}
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
/// \brief Perform substitution on the base class specifiers of the
/// given class template specialization.
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
///
/// Produces a diagnostic and returns true on error, returns false and
/// attaches the instantiated base classes to the class template
/// specialization if successful.
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
bool
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
Sema::SubstBaseSpecifiers(CXXRecordDecl *Instantiation,
CXXRecordDecl *Pattern,
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs) {
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
bool Invalid = false;
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVector<CXXBaseSpecifier*, 4> InstantiatedBases;
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
for (ClassTemplateSpecializationDecl::base_class_iterator
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
Base = Pattern->bases_begin(), BaseEnd = Pattern->bases_end();
If we run into multiple errors within the same template instantiation, only print the template instantiation backtrace for the first error. Also, if a base class has failed to type-check during instantiation, just drop that base class and continue on to check other base classes. llvm-svn: 66563
2009-03-11 02:52:44 +08:00
Base != BaseEnd; ++Base) {
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
if (!Base->getType()->isDependentType()) {
Propagate the invalid bit from bases to derived template classes. Fixes PR16292. llvm-svn: 184581
2013-06-22 02:58:32 +08:00
if (const CXXRecordDecl *RD = Base->getType()->getAsCXXRecordDecl()) {
if (RD->isInvalidDecl())
Instantiation->setInvalidDecl();
}
Allocate BaseOrMemberInitializers and CXXBaseSpecifier nodes via ASTContext. llvm-svn: 76758
2009-07-23 01:41:53 +08:00
InstantiatedBases.push_back(new (Context) CXXBaseSpecifier(*Base));
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
continue;
}
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
SourceLocation EllipsisLoc;
When we're substituting into a parameter-type-list nested inside the pattern of an expansion, and we have a paramameter that is not a parameter pack, don't suppress substitution of parameter packs within this context. llvm-svn: 126819
2011-03-02 10:04:06 +08:00
TypeSourceInfo *BaseTypeLoc;
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
if (Base->isPackExpansion()) {
// This is a pack expansion. See whether we should expand it now, or
// wait until later.
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVector<UnexpandedParameterPack, 2> Unexpanded;
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
collectUnexpandedParameterPacks(Base->getTypeSourceInfo()->getTypeLoc(),
Unexpanded);
bool ShouldExpand = false;
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
bool RetainExpansion = false;
Include llvm::Optional in clang/Basic/LLVM.h Post-commit CR feedback from Jordan Rose regarding r175594. llvm-svn: 175679
2013-02-21 06:23:23 +08:00
Optional<unsigned> NumExpansions;
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
if (CheckParameterPacksForExpansion(Base->getEllipsisLoc(),
Base->getSourceRange(),
ArrayRef-ifying the UnexpandedParameterPacks passed to Sema::CheckParameterPacksForExpansion llvm-svn: 140290
2011-09-22 10:34:54 +08:00
Unexpanded,
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
TemplateArgs, ShouldExpand,
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
RetainExpansion,
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
NumExpansions)) {
Invalid = true;
Implement pack expansion of base initializers, so that we can initialize those lovely mixins that come from pack expansions of base specifiers. llvm-svn: 122793
2011-01-04 08:32:56 +08:00
continue;
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
}
// If we should expand this pack expansion now, do so.
if (ShouldExpand) {
Keep track of the number of expansions to be produced from a type pack expansion, when it is known due to the substitution of an out parameter pack. This allows us to properly handle substitution into pack expansions that involve multiple parameter packs at different template parameter levels, even when this substitution happens one level at a time (as with partial specializations of member class templates and the signatures of member function templates). Note that the diagnostic we provide when there is an arity mismatch between an outer parameter pack and an inner parameter pack in this case isn't as clear as the normal diagnostic for an arity mismatch. However, this doesn't matter because these cases are very, very rare and (even then) only typically occur in a SFINAE context. The other kinds of pack expansions (expression, template, etc.) still need to support optional tracking of the number of expansions, and we need the moral equivalent of SubstTemplateTypeParmPackType for substituted argument packs of template template and non-type template parameters. llvm-svn: 123448
2011-01-15 01:04:44 +08:00
for (unsigned I = 0; I != *NumExpansions; ++I) {
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
Sema::ArgumentPackSubstitutionIndexRAII SubstIndex(*this, I);
TypeSourceInfo *BaseTypeLoc = SubstType(Base->getTypeSourceInfo(),
TemplateArgs,
Base->getSourceRange().getBegin(),
DeclarationName());
if (!BaseTypeLoc) {
Invalid = true;
continue;
}
if (CXXBaseSpecifier *InstantiatedBase
= CheckBaseSpecifier(Instantiation,
Base->getSourceRange(),
Base->isVirtual(),
Base->getAccessSpecifierAsWritten(),
BaseTypeLoc,
SourceLocation()))
InstantiatedBases.push_back(InstantiatedBase);
else
Invalid = true;
}
continue;
}
// The resulting base specifier will (still) be a pack expansion.
EllipsisLoc = Base->getEllipsisLoc();
When we're substituting into a parameter-type-list nested inside the pattern of an expansion, and we have a paramameter that is not a parameter pack, don't suppress substitution of parameter packs within this context. llvm-svn: 126819
2011-03-02 10:04:06 +08:00
Sema::ArgumentPackSubstitutionIndexRAII SubstIndex(*this, -1);
BaseTypeLoc = SubstType(Base->getTypeSourceInfo(),
TemplateArgs,
Base->getSourceRange().getBegin(),
DeclarationName());
} else {
BaseTypeLoc = SubstType(Base->getTypeSourceInfo(),
TemplateArgs,
Base->getSourceRange().getBegin(),
DeclarationName());
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
}
Add source location information to C++ base specifiers. llvm-svn: 109396
2010-07-27 00:56:01 +08:00
if (!BaseTypeLoc) {
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Invalid = true;
continue;
}
if (CXXBaseSpecifier *InstantiatedBase
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
= CheckBaseSpecifier(Instantiation,
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Base->getSourceRange(),
Base->isVirtual(),
Base->getAccessSpecifierAsWritten(),
Implement pack expansions whose pattern is a base-specifier. llvm-svn: 122782
2011-01-04 06:36:02 +08:00
BaseTypeLoc,
EllipsisLoc))
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
InstantiatedBases.push_back(InstantiatedBase);
else
Invalid = true;
}
If we run into multiple errors within the same template instantiation, only print the template instantiation backtrace for the first error. Also, if a base class has failed to type-check during instantiation, just drop that base class and continue on to check other base classes. llvm-svn: 66563
2009-03-11 02:52:44 +08:00
if (!Invalid &&
Use v.data() instead of &v[0] when SmallVector v might be empty. llvm-svn: 72210
2009-05-21 17:52:38 +08:00
AttachBaseSpecifiers(Instantiation, InstantiatedBases.data(),
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
InstantiatedBases.size()))
Invalid = true;
return Invalid;
}
Delayed template instantiation of late-parsed attributes. llvm-svn: 148595
2012-01-21 06:50:54 +08:00
// Defined via #include from SemaTemplateInstantiateDecl.cpp
Move instantiateTemplateAttribute into the sema namespace, make helpers static. llvm-svn: 149864
2012-02-06 19:13:08 +08:00
namespace clang {
namespace sema {
Attr *instantiateTemplateAttribute(const Attr *At, ASTContext &C, Sema &S,
const MultiLevelTemplateArgumentList &TemplateArgs);
}
}
Delayed template instantiation of late-parsed attributes. llvm-svn: 148595
2012-01-21 06:50:54 +08:00
Instantiating a class template should not instantiate the definition of any scoped enumeration members. Later uses of an enumeration temploid as a nested name specifier should cause its instantiation. Plus some groundwork for explicit specialization of member enumerations of class templates. llvm-svn: 152750
2012-03-15 07:13:10 +08:00
/// Determine whether we would be unable to instantiate this template (because
/// it either has no definition, or is in the process of being instantiated).
static bool DiagnoseUninstantiableTemplate(Sema &S,
SourceLocation PointOfInstantiation,
TagDecl *Instantiation,
bool InstantiatedFromMember,
TagDecl *Pattern,
TagDecl *PatternDef,
TemplateSpecializationKind TSK,
bool Complain = true) {
if (PatternDef && !PatternDef->isBeingDefined())
return false;
if (!Complain || (PatternDef && PatternDef->isInvalidDecl())) {
// Say nothing
} else if (PatternDef) {
assert(PatternDef->isBeingDefined());
S.Diag(PointOfInstantiation,
diag::err_template_instantiate_within_definition)
<< (TSK != TSK_ImplicitInstantiation)
<< S.Context.getTypeDeclType(Instantiation);
// Not much point in noting the template declaration here, since
// we're lexically inside it.
Instantiation->setInvalidDecl();
} else if (InstantiatedFromMember) {
S.Diag(PointOfInstantiation,
diag::err_implicit_instantiate_member_undefined)
<< S.Context.getTypeDeclType(Instantiation);
S.Diag(Pattern->getLocation(), diag::note_member_of_template_here);
} else {
S.Diag(PointOfInstantiation, diag::err_template_instantiate_undefined)
<< (TSK != TSK_ImplicitInstantiation)
<< S.Context.getTypeDeclType(Instantiation);
S.Diag(Pattern->getLocation(), diag::note_template_decl_here);
}
// In general, Instantiation isn't marked invalid to get more than one
// error for multiple undefined instantiations. But the code that does
// explicit declaration -> explicit definition conversion can't handle
// invalid declarations, so mark as invalid in that case.
if (TSK == TSK_ExplicitInstantiationDeclaration)
Instantiation->setInvalidDecl();
return true;
}
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
/// \brief Instantiate the definition of a class from a given pattern.
///
/// \param PointOfInstantiation The point of instantiation within the
/// source code.
///
/// \param Instantiation is the declaration whose definition is being
/// instantiated. This will be either a class template specialization
/// or a member class of a class template specialization.
///
/// \param Pattern is the pattern from which the instantiation
/// occurs. This will be either the declaration of a class template or
/// the declaration of a member class of a class template.
///
/// \param TemplateArgs The template arguments to be substituted into
/// the pattern.
///
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
/// \param TSK the kind of implicit or explicit instantiation to perform.
Try to complete a type before looking for conversion functions within that type. Note that we do not produce a diagnostic if the type is incomplete; rather, we just don't look for conversion functions. Fixes PR4660. llvm-svn: 79919
2009-08-24 23:23:48 +08:00
///
/// \param Complain whether to complain if the class cannot be instantiated due
/// to the lack of a definition.
///
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
/// \returns true if an error occurred, false otherwise.
bool
Sema::InstantiateClass(SourceLocation PointOfInstantiation,
CXXRecordDecl *Instantiation, CXXRecordDecl *Pattern,
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs,
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
TemplateSpecializationKind TSK,
Try to complete a type before looking for conversion functions within that type. Note that we do not produce a diagnostic if the type is incomplete; rather, we just don't look for conversion functions. Fixes PR4660. llvm-svn: 79919
2009-08-24 23:23:48 +08:00
bool Complain) {
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
CXXRecordDecl *PatternDef
Eliminate the ASTContext parameter from RecordDecl::getDefinition() and CXXRecordDecl::getDefinition(); it's totally unnecessary. No functionality change. llvm-svn: 95836
2010-02-11 09:04:33 +08:00
= cast_or_null<CXXRecordDecl>(Pattern->getDefinition());
Instantiating a class template should not instantiate the definition of any scoped enumeration members. Later uses of an enumeration temploid as a nested name specifier should cause its instantiation. Plus some groundwork for explicit specialization of member enumerations of class templates. llvm-svn: 152750
2012-03-15 07:13:10 +08:00
if (DiagnoseUninstantiableTemplate(*this, PointOfInstantiation, Instantiation,
Instantiation->getInstantiatedFromMemberClass(),
Pattern, PatternDef, TSK, Complain))
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
return true;
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
Pattern = PatternDef;
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Check the interactions between explicit instantiations and template specializations. Work in progress; there's more cleanup required to actually use the new CheckSpecializationInstantiationRedecl checker uniformly. llvm-svn: 84185
2009-10-15 23:54:05 +08:00
// \brief Record the point of instantiation.
if (MemberSpecializationInfo *MSInfo
= Instantiation->getMemberSpecializationInfo()) {
MSInfo->setTemplateSpecializationKind(TSK);
MSInfo->setPointOfInstantiation(PointOfInstantiation);
Only set the point of instantiation for an implicit or explicit instantiation once we have committed to performing the instantiation. As part of this, make our makeshift template-instantiation location information suck slightly less. Fixes PR5264. llvm-svn: 85209
2009-10-27 14:26:26 +08:00
} else if (ClassTemplateSpecializationDecl *Spec
Fix a crash on invalid, http://llvm.org/pr11599 llvm-svn: 146988
2011-12-21 04:32:49 +08:00
= dyn_cast<ClassTemplateSpecializationDecl>(Instantiation)) {
Only set the point of instantiation for an implicit or explicit instantiation once we have committed to performing the instantiation. As part of this, make our makeshift template-instantiation location information suck slightly less. Fixes PR5264. llvm-svn: 85209
2009-10-27 14:26:26 +08:00
Spec->setTemplateSpecializationKind(TSK);
Spec->setPointOfInstantiation(PointOfInstantiation);
Check the interactions between explicit instantiations and template specializations. Work in progress; there's more cleanup required to actually use the new CheckSpecializationInstantiationRedecl checker uniformly. llvm-svn: 84185
2009-10-15 23:54:05 +08:00
}
Fix notes regarding the instantiation of member classes (and test 'em). llvm-svn: 67708
2009-03-26 05:23:52 +08:00
InstantiatingTemplate Inst(*this, PointOfInstantiation, Instantiation);
Limit the template instantiation depth to some user-configurable value (default: 99). Beyond this limit, produce an error and consider the current template instantiation a failure. The stack we're building to track the instantiations will, eventually, be used to produce instantiation backtraces from diagnostics within template instantiation. However, we're not quite there yet. This adds a new Clang driver option -ftemplate-depth=NNN, which should eventually be generated from the GCC command-line operation -ftemplate-depth-NNN (note the '-' rather than the '='!). I did not make the driver changes to do this mapping. llvm-svn: 66513
2009-03-10 08:06:19 +08:00
if (Inst)
return true;
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
// Enter the scope of this instantiation. We don't use
// PushDeclContext because we don't have a scope.
Properly switch into the declaring scope of a template when performing template argument deduction or (more importantly) the final substitution required by such deduction. Makes access control magically work in these cases. Fixes PR6967. llvm-svn: 102572
2010-04-29 08:35:03 +08:00
ContextRAII SavedContext(*this, Instantiation);
Whenever we instantiate a function definition or class, enter a new potentially-evaluated expression context, to ensure that used declarations get properly marked. Fixes PR7123. llvm-svn: 103624
2010-05-13 01:27:19 +08:00
EnterExpressionEvaluationContext EvalContext(*this,
One who seeks knowledge learns something new every day. One who seeks the Tao unlearns something new every day. Less and less remains until you arrive at non-action. When you arrive at non-action, nothing will be left undone. llvm-svn: 112244
2010-08-27 07:41:50 +08:00
Sema::PotentiallyEvaluated);
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Each non-local class instantiation is its own local instantiation scope. Fixes PR6619. llvm-svn: 99377
2010-03-24 09:33:17 +08:00
// If this is an instantiation of a local class, merge this local
// instantiation scope with the enclosing scope. Otherwise, every
// instantiation of a class has its own local instantiation scope.
bool MergeWithParentScope = !Instantiation->isDefinedOutsideFunctionOrMethod();
Move more stuff out of Sema.h. llvm-svn: 112026
2010-08-25 13:32:35 +08:00
LocalInstantiationScope Scope(*this, MergeWithParentScope);
Each non-local class instantiation is its own local instantiation scope. Fixes PR6619. llvm-svn: 99377
2010-03-24 09:33:17 +08:00
Instantiate attributes from the pattern record when instantiating a class template. Fixes rdar://problem/8243419. llvm-svn: 109967
2010-08-01 10:01:53 +08:00
// Pull attributes from the pattern onto the instantiation.
InstantiateAttrs(TemplateArgs, Pattern, Instantiation);
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
// Start the definition of this instantiation.
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
Instantiation->startDefinition();
Partial and full specializations of a class template may have a different tag kind ("struct" vs. "class") than the primary template, which has an affect on access control. Should fix the last remaining Boost.Accumulors failure. llvm-svn: 103144
2010-05-06 08:28:52 +08:00
Instantiation->setTagKind(Pattern->getTagKind());
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
// Do substitution on the base class specifiers.
if (SubstBaseSpecifiers(Instantiation, Pattern, TemplateArgs))
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
Instantiation->setInvalidDecl();
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Instantiate class member template partial specialization declarations in the order they occur within the class template, delaying out-of-line member template partial specializations until after the class has been fully instantiated. This fixes a regression introduced by r118454 (itself a fix for PR8001). llvm-svn: 118704
2010-11-11 03:44:59 +08:00
TemplateDeclInstantiator Instantiator(*this, Instantiation, TemplateArgs);
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVector<Decl*, 4> Fields;
SmallVector<std::pair<FieldDecl*, FieldDecl*>, 4>
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
FieldsWithMemberInitializers;
Delayed template instantiation of late-parsed attributes. llvm-svn: 148595
2012-01-21 06:50:54 +08:00
// Delay instantiation of late parsed attributes.
LateInstantiatedAttrVec LateAttrs;
Instantiator.enableLateAttributeInstantiation(&LateAttrs);
De-ASTContext-ify DeclContext. Remove ASTContext parameter from DeclContext's methods. This change cascaded down to other Decl's methods and changes to call sites started "escalating". Timings using pre-tokenized "cocoa.h" showed only a ~1% increase in time run between and after this commit. llvm-svn: 74506
2009-06-30 10:36:12 +08:00
for (RecordDecl::decl_iterator Member = Pattern->decls_begin(),
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
MemberEnd = Pattern->decls_end();
Propagate the ASTContext to various AST traversal and lookup functions. No functionality change (really). llvm-svn: 68726
2009-04-10 05:40:53 +08:00
Member != MemberEnd; ++Member) {
Don't instantiate members not belonging in the semantic context of the template. e.g. for: template <int i> class A { class B *g; }; 'class B' has the template as lexical context but semantically it is introduced in namespace scope. Fixes rdar://8611125 & http://llvm.org/PR8505 llvm-svn: 118235
2010-11-04 11:18:57 +08:00
// Don't instantiate members not belonging in this semantic context.
// e.g. for:
// @code
// template <int i> class A {
// class B *g;
// };
// @endcode
// 'class B' has the template as lexical context but semantically it is
// introduced in namespace scope.
if ((*Member)->getDeclContext() != Pattern)
continue;
Instantiate class member template partial specialization declarations in the order they occur within the class template, delaying out-of-line member template partial specializations until after the class has been fully instantiated. This fixes a regression introduced by r118454 (itself a fix for PR8001). llvm-svn: 118704
2010-11-11 03:44:59 +08:00
if ((*Member)->isInvalidDecl()) {
Stop instantiating a class if we hit a static_assert failure. Also, if the static_assert fails when parsing the template, don't diagnose it again on every instantiation. llvm-svn: 160088
2012-07-12 06:37:56 +08:00
Instantiation->setInvalidDecl();
Instantiate class member template partial specialization declarations in the order they occur within the class template, delaying out-of-line member template partial specializations until after the class has been fully instantiated. This fixes a regression introduced by r118454 (itself a fix for PR8001). llvm-svn: 118704
2010-11-11 03:44:59 +08:00
continue;
}
Decl *NewMember = Instantiator.Visit(*Member);
Refactor instantiation of declarations within a template into a much cleaner visitor framework. Added a visitor for declarations, which is quite similar to the visitor for statatements. llvm-svn: 67104
2009-03-18 05:15:40 +08:00
if (NewMember) {
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
if (FieldDecl *Field = dyn_cast<FieldDecl>(NewMember)) {
DeclPtrTy -> Decl * llvm-svn: 111733
2010-08-21 17:40:31 +08:00
Fields.push_back(Field);
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
FieldDecl *OldField = cast<FieldDecl>(*Member);
if (OldField->getInClassInitializer())
FieldsWithMemberInitializers.push_back(std::make_pair(OldField,
Field));
Support for definitions of member enumerations of class templates outside the class template's definition, and for explicit specializations of such enum members. llvm-svn: 153304
2012-03-23 11:33:32 +08:00
} else if (EnumDecl *Enum = dyn_cast<EnumDecl>(NewMember)) {
// C++11 [temp.inst]p1: The implicit instantiation of a class template
// specialization causes the implicit instantiation of the definitions
// of unscoped member enumerations.
// Record a point of instantiation for this implicit instantiation.
When defining a forward-declared enum, don't try to attach the definition to a previous declaration if the redeclaration is invalid. That way lies madness. Fixes a crash-on-invalid reported by Abramo. llvm-svn: 153349
2012-03-24 07:09:08 +08:00
if (TSK == TSK_ImplicitInstantiation && !Enum->isScoped() &&
Enum->isCompleteDefinition()) {
Support for definitions of member enumerations of class templates outside the class template's definition, and for explicit specializations of such enum members. llvm-svn: 153304
2012-03-23 11:33:32 +08:00
MemberSpecializationInfo *MSInfo =Enum->getMemberSpecializationInfo();
assert(MSInfo && "no spec info for member enum specialization");
MSInfo->setTemplateSpecializationKind(TSK_ImplicitInstantiation);
MSInfo->setPointOfInstantiation(PointOfInstantiation);
}
Stop instantiating a class if we hit a static_assert failure. Also, if the static_assert fails when parsing the template, don't diagnose it again on every instantiation. llvm-svn: 160088
2012-07-12 06:37:56 +08:00
} else if (StaticAssertDecl *SA = dyn_cast<StaticAssertDecl>(NewMember)) {
if (SA->isFailed()) {
// A static_assert failed. Bail out; instantiating this
// class is probably not meaningful.
Instantiation->setInvalidDecl();
break;
}
Support for definitions of member enumerations of class templates outside the class template's definition, and for explicit specializations of such enum members. llvm-svn: 153304
2012-03-23 11:33:32 +08:00
}
if (NewMember->isInvalidDecl())
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
Instantiation->setInvalidDecl();
Refactor instantiation of declarations within a template into a much cleaner visitor framework. Added a visitor for declarations, which is quite similar to the visitor for statatements. llvm-svn: 67104
2009-03-18 05:15:40 +08:00
} else {
// FIXME: Eventually, a NULL return will mean that one of the
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
// instantiations was a semantic disaster, and we'll want to mark the
// declaration invalid.
// For now, we expect to skip some members that we can't yet handle.
Add basic, hackish support for instantiation of typedefs in a class template. More importantly, start to sort out the issues regarding complete types and nested-name-specifiers, especially the question of: when do we instantiate a class template specialization that occurs to the left of a '::' in a nested-name-specifier? llvm-svn: 66662
2009-03-12 00:48:53 +08:00
}
}
Implement basic template instantiation for fields. Reshuffle checking for FieldDecls so that the parser and the template instantiation make use of the same semantic checking module. llvm-svn: 66685
2009-03-12 02:59:21 +08:00
// Finish checking fields.
ArrayRef-ifying the fields passed to Sema::ActOnFields llvm-svn: 140293
2011-09-22 10:58:26 +08:00
ActOnFields(0, Instantiation->getLocation(), Instantiation, Fields,
SourceLocation(), SourceLocation(), 0);
Move the "current scope" state from the Parser into Action. This allows Sema some limited access to the current scope, which we only use in one way: when Sema is performing some kind of declaration that is not directly driven by the parser (e.g., due to template instantiatio or lazy declaration of a member), we can find the Scope associated with a DeclContext, if that DeclContext is still in the process of being parsed. Use this to make the implicit declaration of special member functions in a C++ class more "scope-less", rather than using the NULL Scope hack. llvm-svn: 107491
2010-07-03 01:43:08 +08:00
CheckCompletedCXXClass(Instantiation);
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
// Attach any in-class member initializers now the class is complete.
Add a FIXME. llvm-svn: 169664
2012-12-08 10:13:02 +08:00
// FIXME: We are supposed to defer instantiating these until they are needed.
CXXThisScopeRAII objects aren't free, don't compute one if it's unused. llvm-svn: 156987
2012-05-17 20:01:52 +08:00
if (!FieldsWithMemberInitializers.empty()) {
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
// C++11 [expr.prim.general]p4:
// Otherwise, if a member-declarator declares a non-static data member
// (9.2) of a class X, the expression this is a prvalue of type "pointer
// to X" within the optional brace-or-equal-initializer. It shall not
// appear elsewhere in the member-declarator.
CXXThisScopeRAII ThisScope(*this, Instantiation, (unsigned)0);
for (unsigned I = 0, N = FieldsWithMemberInitializers.size(); I != N; ++I) {
FieldDecl *OldField = FieldsWithMemberInitializers[I].first;
FieldDecl *NewField = FieldsWithMemberInitializers[I].second;
Expr *OldInit = OldField->getInClassInitializer();
ExprResult NewInit = SubstInitializer(OldInit, TemplateArgs,
/*CXXDirectInit=*/false);
if (NewInit.isInvalid())
NewField->setInvalidDecl();
else {
Expr *Init = NewInit.take();
assert(Init && "no-argument initializer in class");
assert(!isa<ParenListExpr>(Init) && "call-style init in class");
PR13064: Store whether an in-class initializer uses direct or copy initialization, and use that information to produce the right kind of initialization during template instantiation. llvm-svn: 158288
2012-06-10 11:12:00 +08:00
ActOnCXXInClassMemberInitializer(NewField, Init->getLocStart(), Init);
Implement C++11 [expr.prim.general]p3, which permits the use of 'this' in the declaration of a non-static member function after the (optional) cv-qualifier-seq, which in practice means in the exception specification and late-specified return type. The new scheme here used to manage 'this' outside of a member function scope is more general than the Scope-based mechanism previously used for non-static data member initializers and late-parsesd attributes, because it can also handle the cv-qualifiers on the member function. Note, however, that a separate pass is required for static member functions to determine whether 'this' was used, because we might not know that we have a static function until after declaration matching. Finally, this introduces name mangling for 'this' and for the implicit 'this', which is intended to match GCC's mangling. Independent verification for the new mangling test case would be appreciated. Fixes PR10036 and PR12450. llvm-svn: 154799
2012-04-16 15:05:22 +08:00
}
Replace r134583's fix for PR10290 with one which also works for non-value-dependent cases. llvm-svn: 135543
2011-07-20 08:12:52 +08:00
}
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
}
Delayed template instantiation of late-parsed attributes. llvm-svn: 148595
2012-01-21 06:50:54 +08:00
// Instantiate late parsed attributes, and attach them to their decls.
// See Sema::InstantiateAttrs
for (LateInstantiatedAttrVec::iterator I = LateAttrs.begin(),
E = LateAttrs.end(); I != E; ++I) {
assert(CurrentInstantiationScope == Instantiator.getStartingScope());
CurrentInstantiationScope = I->Scope;
PR16243: Use CXXThisOverride during template instantiation, and fix up the places which weren't setting it up properly. This allows us to get the right cv-qualifiers for 'this' when it appears outside a method body in a class template. llvm-svn: 183483
2013-06-07 10:33:37 +08:00
// Allow 'this' within late-parsed attributes.
NamedDecl *ND = dyn_cast<NamedDecl>(I->NewDecl);
CXXRecordDecl *ThisContext =
dyn_cast_or_null<CXXRecordDecl>(ND->getDeclContext());
CXXThisScopeRAII ThisScope(*this, ThisContext, /*TypeQuals*/0,
ND && ND->isCXXInstanceMember());
Delayed template instantiation of late-parsed attributes. llvm-svn: 148595
2012-01-21 06:50:54 +08:00
Attr *NewAttr =
instantiateTemplateAttribute(I->TmplAttr, Context, *this, TemplateArgs);
I->NewDecl->addAttr(NewAttr);
LocalInstantiationScope::deleteScopes(I->Scope,
Instantiator.getStartingScope());
}
Instantiator.disableLateAttributeInstantiation();
LateAttrs.clear();
Final piece of core issue 1330: delay computing the exception specification of a defaulted special member function until the exception specification is needed (using the same criteria used for the delayed instantiation of exception specifications for function temploids). EST_Delayed is now EST_Unevaluated (using 1330's terminology), and, like EST_Uninstantiated, carries a pointer to the FunctionDecl which will be used to resolve the exception specification. This is enabled for all C++ modes: it's a little faster in the case where the exception specification isn't used, allows our C++11-in-C++98 extensions to work, and is still correct for C++98, since in that mode the computation of the exception specification can't fail. The diagnostics here aren't great (in particular, we should include implicit evaluation of exception specifications for defaulted special members in the template instantiation backtraces), but they're not much worse than before. Our approach to the problem of cycles between in-class initializers and the exception specification for a defaulted default constructor is modified a little by this change -- we now reject any odr-use of a defaulted default constructor if that constructor uses an in-class initializer and the use is in an in-class initialzer which is declared lexically earlier. This is a closer approximation to the current draft solution in core issue 1351, but isn't an exact match (but the current draft wording isn't reasonable, so that's to be expected). llvm-svn: 160847
2012-07-27 12:22:15 +08:00
ActOnFinishDelayedMemberInitializers(Instantiation);
Implement support for C++11 in-class initialization of non-static data members. llvm-svn: 132878
2011-06-12 01:19:42 +08:00
Fixed implicit instantiations source range. llvm-svn: 144977
2011-11-18 16:08:52 +08:00
if (TSK == TSK_ImplicitInstantiation) {
For class template implicit instantiation, also update its location to point to the pattern template that it came from, otherwise we had this situation: template <typename T1, typename T2> struct S { }; template <typename T> struct S<T, int> { }; void f() { S<int, int> s; // location of declaration "S<int, int>" was of "S<T1, T2>" not "S<T, int>" } llvm-svn: 150290
2012-02-11 09:59:57 +08:00
Instantiation->setLocation(Pattern->getLocation());
Fixed implicit instantiations source range. llvm-svn: 144977
2011-11-18 16:08:52 +08:00
Instantiation->setLocStart(Pattern->getInnerLocStart());
Fixed source range for template implicit instantiations. llvm-svn: 141018
2011-10-04 04:34:03 +08:00
Instantiation->setRBraceLoc(Pattern->getRBraceLoc());
Fixed implicit instantiations source range. llvm-svn: 144977
2011-11-18 16:08:52 +08:00
}
Fixed source range for template implicit instantiations. llvm-svn: 141018
2011-10-04 04:34:03 +08:00
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
if (!Instantiation->isInvalidDecl()) {
Check access to friend declarations. There's a number of different things going on here that were problematic: - We were missing the actual access check, or rather, it was suppressed on account of being a redeclaration lookup. - The access check would naturally happen during delay, which isn't appropriate in this case. - We weren't actually emitting dependent diagnostics associated with class templates, which was unfortunate. - Access was being propagated incorrectly for friend method declarations that couldn't be matched at parse-time. llvm-svn: 161652
2012-08-10 11:15:35 +08:00
// Perform any dependent diagnostics from the pattern.
PerformDependentDiagnostics(Pattern, TemplateArgs);
Instantiate class member template partial specialization declarations in the order they occur within the class template, delaying out-of-line member template partial specializations until after the class has been fully instantiated. This fixes a regression introduced by r118454 (itself a fix for PR8001). llvm-svn: 118704
2010-11-11 03:44:59 +08:00
// Instantiate any out-of-line class template partial
// specializations now.
If a partial specialization of a member template is declared within a class template and defined outside it, don't instantiate it twice when instantiating the surrounding class template specialization. That would cause us to reject the code because we think two partial specializations instantiated to produce the same signature. llvm-svn: 191418
2013-09-26 11:49:48 +08:00
for (TemplateDeclInstantiator::delayed_partial_spec_iterator
Instantiate class member template partial specialization declarations in the order they occur within the class template, delaying out-of-line member template partial specializations until after the class has been fully instantiated. This fixes a regression introduced by r118454 (itself a fix for PR8001). llvm-svn: 118704
2010-11-11 03:44:59 +08:00
P = Instantiator.delayed_partial_spec_begin(),
PEnd = Instantiator.delayed_partial_spec_end();
P != PEnd; ++P) {
if (!Instantiator.InstantiateClassTemplatePartialSpecialization(
If a partial specialization of a member template is declared within a class template and defined outside it, don't instantiate it twice when instantiating the surrounding class template specialization. That would cause us to reject the code because we think two partial specializations instantiated to produce the same signature. llvm-svn: 191418
2013-09-26 11:49:48 +08:00
P->first, P->second)) {
Instantiation->setInvalidDecl();
break;
}
}
// Instantiate any out-of-line variable template partial
// specializations now.
for (TemplateDeclInstantiator::delayed_var_partial_spec_iterator
P = Instantiator.delayed_var_partial_spec_begin(),
PEnd = Instantiator.delayed_var_partial_spec_end();
P != PEnd; ++P) {
if (!Instantiator.InstantiateVarTemplatePartialSpecialization(
P->first, P->second)) {
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
Instantiation->setInvalidDecl();
Instantiate class member template partial specialization declarations in the order they occur within the class template, delaying out-of-line member template partial specializations until after the class has been fully instantiated. This fixes a regression introduced by r118454 (itself a fix for PR8001). llvm-svn: 118704
2010-11-11 03:44:59 +08:00
break;
}
}
}
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
// Exit the scope of this instantiation.
Properly switch into the declaring scope of a template when performing template argument deduction or (more importantly) the final substitution required by such deduction. Makes access control magically work in these cases. Fixes PR6967. llvm-svn: 102572
2010-04-29 08:35:03 +08:00
SavedContext.pop();
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
if (!Instantiation->isInvalidDecl()) {
Make sure that CodeGen sees template instantiations. llvm-svn: 72433
2009-05-27 04:50:29 +08:00
Consumer.HandleTagDeclDefinition(Instantiation);
Rework when and how vtables are emitted, by tracking where vtables are "used" (e.g., we will refer to the vtable in the generated code) and when they are defined (i.e., because we've seen the key function definition). Previously, we were effectively tracking "potential definitions" rather than uses, so we were a bit too eager about emitting vtables for classes without key functions. The new scheme: - For every use of a vtable, Sema calls MarkVTableUsed() to indicate the use. For example, this occurs when calling a virtual member function of the class, defining a constructor of that class type, dynamic_cast'ing from that type to a derived class, casting to/through a virtual base class, etc. - For every definition of a vtable, Sema calls MarkVTableUsed() to indicate the definition. This happens at the end of the translation unit for classes whose key function has been defined (so we can delay computation of the key function; see PR6564), and will also occur with explicit template instantiation definitions. - For every vtable defined/used, we mark all of the virtual member functions of that vtable as defined/used, unless we know that the key function is in another translation unit. This instantiates virtual member functions when needed. - At the end of the translation unit, Sema tells CodeGen (via the ASTConsumer) which vtables must be defined (CodeGen will define them) and which may be used (for which CodeGen will define the vtables lazily). From a language perspective, both the old and the new schemes are permissible: we're allowed to instantiate virtual member functions whenever we want per the standard. However, all other C++ compilers were more lazy than we were, and our eagerness was both a performance issue (we instantiated too much) and a portability problem (we broke Boost test cases, which now pass). Notes: (1) There's a ton of churn in the tests, because the order in which vtables get emitted to IR has changed. I've tried to isolate some of the larger tests from these issues. (2) Some diagnostics related to implicitly-instantiated/implicitly-defined virtual member functions have moved to the point of first use/definition. It's better this way. (3) I could use a review of the places where we MarkVTableUsed, to see if I missed any place where the language effectively requires a vtable. Fixes PR7114 and PR6564. llvm-svn: 103718
2010-05-14 00:44:06 +08:00
// Always emit the vtable for an explicit instantiation definition
// of a polymorphic class template specialization.
if (TSK == TSK_ExplicitInstantiationDefinition)
MarkVTableUsed(PointOfInstantiation, Instantiation, true);
}
Be more eager about setting the 'Invalid' bit on an invalid class template instantiation. I wasn't able to reproduce this down to anything small enough to put in our test suite, but it's "obviously" okay to set the invalid bit earlier and precludes a known-broken-but-not-marked-broken class from being used elsewhere. llvm-svn: 159584
2012-07-03 05:00:41 +08:00
return Instantiation->isInvalidDecl();
Implement the basics of implicit instantiation of class templates, in response to attempts to diagnose an "incomplete" type. This will force us to use DiagnoseIncompleteType more regularly (rather than looking at isIncompleteType), but that's also a good thing. Implicit instantiation is still very simplistic, and will create a new definition for the class template specialization (as it should) but it only actually instantiates the base classes and attaches those. Actually instantiating class members will follow. Also, instantiate the types of non-type template parameters before checking them, allowing, e.g., template<typename T, T Value> struct Constant; to work properly. llvm-svn: 65924
2009-03-03 12:44:36 +08:00
}
Introduce a new expression type, UnresolvedDeclRefExpr, that describes dependent qualified-ids such as Fibonacci<N - 1>::value where N is a template parameter. These references are "unresolved" because the name is dependent and, therefore, cannot be resolved to a declaration node (as we would do for a DeclRefExpr or QualifiedDeclRefExpr). UnresolvedDeclRefExprs instantiate to DeclRefExprs, QualifiedDeclRefExprs, etc. Also, be a bit more careful about keeping only a single set of specializations for a class template, and instantiating from the definition of that template rather than a previous declaration. In general, we need a better solution for this for all TagDecls, because it's too easy to accidentally look at a declaration that isn't the definition. We can now process a simple Fibonacci computation described as a template metaprogram. llvm-svn: 67308
2009-03-20 01:26:29 +08:00
Instantiating a class template should not instantiate the definition of any scoped enumeration members. Later uses of an enumeration temploid as a nested name specifier should cause its instantiation. Plus some groundwork for explicit specialization of member enumerations of class templates. llvm-svn: 152750
2012-03-15 07:13:10 +08:00
/// \brief Instantiate the definition of an enum from a given pattern.
///
/// \param PointOfInstantiation The point of instantiation within the
/// source code.
/// \param Instantiation is the declaration whose definition is being
/// instantiated. This will be a member enumeration of a class
/// temploid specialization, or a local enumeration within a
/// function temploid specialization.
/// \param Pattern The templated declaration from which the instantiation
/// occurs.
/// \param TemplateArgs The template arguments to be substituted into
/// the pattern.
/// \param TSK The kind of implicit or explicit instantiation to perform.
///
/// \return \c true if an error occurred, \c false otherwise.
bool Sema::InstantiateEnum(SourceLocation PointOfInstantiation,
EnumDecl *Instantiation, EnumDecl *Pattern,
const MultiLevelTemplateArgumentList &TemplateArgs,
TemplateSpecializationKind TSK) {
EnumDecl *PatternDef = Pattern->getDefinition();
if (DiagnoseUninstantiableTemplate(*this, PointOfInstantiation, Instantiation,
Instantiation->getInstantiatedFromMemberEnum(),
Pattern, PatternDef, TSK,/*Complain*/true))
return true;
Pattern = PatternDef;
// Record the point of instantiation.
if (MemberSpecializationInfo *MSInfo
= Instantiation->getMemberSpecializationInfo()) {
MSInfo->setTemplateSpecializationKind(TSK);
MSInfo->setPointOfInstantiation(PointOfInstantiation);
}
InstantiatingTemplate Inst(*this, PointOfInstantiation, Instantiation);
if (Inst)
return true;
// Enter the scope of this instantiation. We don't use
// PushDeclContext because we don't have a scope.
ContextRAII SavedContext(*this, Instantiation);
EnterExpressionEvaluationContext EvalContext(*this,
Sema::PotentiallyEvaluated);
LocalInstantiationScope Scope(*this, /*MergeWithParentScope*/true);
// Pull attributes from the pattern onto the instantiation.
InstantiateAttrs(TemplateArgs, Pattern, Instantiation);
TemplateDeclInstantiator Instantiator(*this, Instantiation, TemplateArgs);
Instantiator.InstantiateEnumDefinition(Instantiation, Pattern);
// Exit the scope of this instantiation.
SavedContext.pop();
return Instantiation->isInvalidDecl();
}
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
namespace {
/// \brief A partial specialization whose template arguments have matched
/// a given template-id.
struct PartialSpecMatchResult {
ClassTemplatePartialSpecializationDecl *Partial;
TemplateArgumentList *Args;
};
}
Refactor for clarity and simplicity. llvm-svn: 188974
2013-08-22 08:59:14 +08:00
bool Sema::InstantiateClassTemplateSpecialization(
SourceLocation PointOfInstantiation,
ClassTemplateSpecializationDecl *ClassTemplateSpec,
TemplateSpecializationKind TSK, bool Complain) {
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
// Perform the actual instantiation on the canonical declaration.
ClassTemplateSpec = cast<ClassTemplateSpecializationDecl>(
Remove ASTContext::getCanonicalDecl() and use Decl::getCanonicalDecl in its place. llvm-svn: 76274
2009-07-18 08:34:25 +08:00
ClassTemplateSpec->getCanonicalDecl());
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
Slight improvement for extern templates, so that an explicit instantiation definition can follow an explicit instantiation declaration. This is as far as I want to go with extern templates now, but they will still need quite a bit more work to get all of the C++0x semantics right. llvm-svn: 81573
2009-09-12 05:19:12 +08:00
// Check whether we have already instantiated or specialized this class
// template specialization.
if (ClassTemplateSpec->getSpecializationKind() != TSK_Undeclared) {
if (ClassTemplateSpec->getSpecializationKind() ==
TSK_ExplicitInstantiationDeclaration &&
TSK == TSK_ExplicitInstantiationDefinition) {
// An explicit instantiation definition follows an explicit instantiation
// declaration (C++0x [temp.explicit]p10); go ahead and perform the
// explicit instantiation.
ClassTemplateSpec->setSpecializationKind(TSK);
Rework when and how vtables are emitted, by tracking where vtables are "used" (e.g., we will refer to the vtable in the generated code) and when they are defined (i.e., because we've seen the key function definition). Previously, we were effectively tracking "potential definitions" rather than uses, so we were a bit too eager about emitting vtables for classes without key functions. The new scheme: - For every use of a vtable, Sema calls MarkVTableUsed() to indicate the use. For example, this occurs when calling a virtual member function of the class, defining a constructor of that class type, dynamic_cast'ing from that type to a derived class, casting to/through a virtual base class, etc. - For every definition of a vtable, Sema calls MarkVTableUsed() to indicate the definition. This happens at the end of the translation unit for classes whose key function has been defined (so we can delay computation of the key function; see PR6564), and will also occur with explicit template instantiation definitions. - For every vtable defined/used, we mark all of the virtual member functions of that vtable as defined/used, unless we know that the key function is in another translation unit. This instantiates virtual member functions when needed. - At the end of the translation unit, Sema tells CodeGen (via the ASTConsumer) which vtables must be defined (CodeGen will define them) and which may be used (for which CodeGen will define the vtables lazily). From a language perspective, both the old and the new schemes are permissible: we're allowed to instantiate virtual member functions whenever we want per the standard. However, all other C++ compilers were more lazy than we were, and our eagerness was both a performance issue (we instantiated too much) and a portability problem (we broke Boost test cases, which now pass). Notes: (1) There's a ton of churn in the tests, because the order in which vtables get emitted to IR has changed. I've tried to isolate some of the larger tests from these issues. (2) Some diagnostics related to implicitly-instantiated/implicitly-defined virtual member functions have moved to the point of first use/definition. It's better this way. (3) I could use a review of the places where we MarkVTableUsed, to see if I missed any place where the language effectively requires a vtable. Fixes PR7114 and PR6564. llvm-svn: 103718
2010-05-14 00:44:06 +08:00
// If this is an explicit instantiation definition, mark the
// vtable as used.
Fix a crash on invalid, http://llvm.org/pr11599 llvm-svn: 146988
2011-12-21 04:32:49 +08:00
if (TSK == TSK_ExplicitInstantiationDefinition &&
!ClassTemplateSpec->isInvalidDecl())
Rework when and how vtables are emitted, by tracking where vtables are "used" (e.g., we will refer to the vtable in the generated code) and when they are defined (i.e., because we've seen the key function definition). Previously, we were effectively tracking "potential definitions" rather than uses, so we were a bit too eager about emitting vtables for classes without key functions. The new scheme: - For every use of a vtable, Sema calls MarkVTableUsed() to indicate the use. For example, this occurs when calling a virtual member function of the class, defining a constructor of that class type, dynamic_cast'ing from that type to a derived class, casting to/through a virtual base class, etc. - For every definition of a vtable, Sema calls MarkVTableUsed() to indicate the definition. This happens at the end of the translation unit for classes whose key function has been defined (so we can delay computation of the key function; see PR6564), and will also occur with explicit template instantiation definitions. - For every vtable defined/used, we mark all of the virtual member functions of that vtable as defined/used, unless we know that the key function is in another translation unit. This instantiates virtual member functions when needed. - At the end of the translation unit, Sema tells CodeGen (via the ASTConsumer) which vtables must be defined (CodeGen will define them) and which may be used (for which CodeGen will define the vtables lazily). From a language perspective, both the old and the new schemes are permissible: we're allowed to instantiate virtual member functions whenever we want per the standard. However, all other C++ compilers were more lazy than we were, and our eagerness was both a performance issue (we instantiated too much) and a portability problem (we broke Boost test cases, which now pass). Notes: (1) There's a ton of churn in the tests, because the order in which vtables get emitted to IR has changed. I've tried to isolate some of the larger tests from these issues. (2) Some diagnostics related to implicitly-instantiated/implicitly-defined virtual member functions have moved to the point of first use/definition. It's better this way. (3) I could use a review of the places where we MarkVTableUsed, to see if I missed any place where the language effectively requires a vtable. Fixes PR7114 and PR6564. llvm-svn: 103718
2010-05-14 00:44:06 +08:00
MarkVTableUsed(PointOfInstantiation, ClassTemplateSpec, true);
Slight improvement for extern templates, so that an explicit instantiation definition can follow an explicit instantiation declaration. This is as far as I want to go with extern templates now, but they will still need quite a bit more work to get all of the C++0x semantics right. llvm-svn: 81573
2009-09-12 05:19:12 +08:00
return false;
}
// We can only instantiate something that hasn't already been
// instantiated or specialized. Fail without any diagnostics: our
// caller will provide an error message.
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
return true;
Slight improvement for extern templates, so that an explicit instantiation definition can follow an explicit instantiation declaration. This is as far as I want to go with extern templates now, but they will still need quite a bit more work to get all of the C++0x semantics right. llvm-svn: 81573
2009-09-12 05:19:12 +08:00
}
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
Slightly improved template argument deduction for use in partial ordering, along with another test case for partial ordering of partial specializations. llvm-svn: 81869
2009-09-16 00:51:42 +08:00
if (ClassTemplateSpec->isInvalidDecl())
return true;
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
ClassTemplateDecl *Template = ClassTemplateSpec->getSpecializedTemplate();
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
CXXRecordDecl *Pattern = 0;
Initial infrastructure for class template partial specialization. Here we have the basics of declaring and storing class template partial specializations, matching class template partial specializations at instantiation time via (limited) template argument deduction, and using the class template partial specialization's pattern for instantiation. This patch is enough to make a simple is_pointer type trait work, but not much else. llvm-svn: 72662
2009-05-31 17:31:02 +08:00
It looks like we've finished off matching of class template partial specializations; add comments and update the C++ status page llvm-svn: 73263
2009-06-13 06:31:52 +08:00
// C++ [temp.class.spec.match]p1:
// When a class template is used in a context that requires an
// instantiation of the class, it is necessary to determine
// whether the instantiation is to be generated using the primary
// template or one of the partial specializations. This is done by
// matching the template arguments of the class template
// specialization with the template argument lists of the partial
// specializations.
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
typedef PartialSpecMatchResult MatchResult;
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVector<MatchResult, 4> Matched;
SmallVector<ClassTemplatePartialSpecializationDecl *, 4> PartialSpecs;
Introduce a sequence number into class template partial specializations, which keeps track of the order in which they were originally declared. We use this number so that we can always walk the list of partial specializations in a predictable order during matching or template instantiation. This also fixes a failure in Boost.Proto, where SourceManager::isBeforeInTranslationUnit was behaving poorly in inconsistent ways. llvm-svn: 102693
2010-04-30 13:56:50 +08:00
Template->getPartialSpecializations(PartialSpecs);
FIXME fix: improving diagnostics for template arguments deduction of class templates and explicit specializations This patch essentially removes all the FIXMEs following calls to DeduceTemplateArguments() that want to keep track of deduction failure info. llvm-svn: 186730
2013-07-20 07:00:19 +08:00
TemplateSpecCandidateSet FailedCandidates(PointOfInstantiation);
Introduce a sequence number into class template partial specializations, which keeps track of the order in which they were originally declared. We use this number so that we can always walk the list of partial specializations in a predictable order during matching or template instantiation. This also fixes a failure in Boost.Proto, where SourceManager::isBeforeInTranslationUnit was behaving poorly in inconsistent ways. llvm-svn: 102693
2010-04-30 13:56:50 +08:00
for (unsigned I = 0, N = PartialSpecs.size(); I != N; ++I) {
ClassTemplatePartialSpecializationDecl *Partial = PartialSpecs[I];
FIXME fix: improving diagnostics for template arguments deduction of class templates and explicit specializations This patch essentially removes all the FIXMEs following calls to DeduceTemplateArguments() that want to keep track of deduction failure info. llvm-svn: 186730
2013-07-20 07:00:19 +08:00
TemplateDeductionInfo Info(FailedCandidates.getLocation());
Improve template argument deduction to keep track of why template argument deduction failed. For example, given template<typename T> struct is_same<T, T> { ... }; template argument deduction will fail for is_same<int, float>, and now reports enough information Right now, we don't do anything with this extra information, but it can be used for informative diagnostics that say, e.g., "template argument deduction failed because T was deduced to 'int' in one context and 'float' in another". llvm-svn: 73237
2009-06-13 02:26:56 +08:00
if (TemplateDeductionResult Result
Introduce a sequence number into class template partial specializations, which keeps track of the order in which they were originally declared. We use this number so that we can always walk the list of partial specializations in a predictable order during matching or template instantiation. This also fixes a failure in Boost.Proto, where SourceManager::isBeforeInTranslationUnit was behaving poorly in inconsistent ways. llvm-svn: 102693
2010-04-30 13:56:50 +08:00
= DeduceTemplateArguments(Partial,
Improve template argument deduction to keep track of why template argument deduction failed. For example, given template<typename T> struct is_same<T, T> { ... }; template argument deduction will fail for is_same<int, float>, and now reports enough information Right now, we don't do anything with this extra information, but it can be used for informative diagnostics that say, e.g., "template argument deduction failed because T was deduced to 'int' in one context and 'float' in another". llvm-svn: 73237
2009-06-13 02:26:56 +08:00
ClassTemplateSpec->getTemplateArgs(),
Info)) {
FIXME fix: improving diagnostics for template arguments deduction of class templates and explicit specializations This patch essentially removes all the FIXMEs following calls to DeduceTemplateArguments() that want to keep track of deduction failure info. llvm-svn: 186730
2013-07-20 07:00:19 +08:00
// Store the failed-deduction information for use in diagnostics, later.
// TODO: Actually use the failed-deduction info?
FailedCandidates.addCandidate()
.set(Partial, MakeDeductionFailureInfo(Context, Result, Info));
Improve template argument deduction to keep track of why template argument deduction failed. For example, given template<typename T> struct is_same<T, T> { ... }; template argument deduction will fail for is_same<int, float>, and now reports enough information Right now, we don't do anything with this extra information, but it can be used for informative diagnostics that say, e.g., "template argument deduction failed because T was deduced to 'int' in one context and 'float' in another". llvm-svn: 73237
2009-06-13 02:26:56 +08:00
(void)Result;
} else {
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
Matched.push_back(PartialSpecMatchResult());
Matched.back().Partial = Partial;
Matched.back().Args = Info.take();
Improve template argument deduction to keep track of why template argument deduction failed. For example, given template<typename T> struct is_same<T, T> { ... }; template argument deduction will fail for is_same<int, float>, and now reports enough information Right now, we don't do anything with this extra information, but it can be used for informative diagnostics that say, e.g., "template argument deduction failed because T was deduced to 'int' in one context and 'float' in another". llvm-svn: 73237
2009-06-13 02:26:56 +08:00
}
Initial infrastructure for class template partial specialization. Here we have the basics of declaring and storing class template partial specializations, matching class template partial specializations at instantiation time via (limited) template argument deduction, and using the class template partial specialization's pattern for instantiation. This patch is enough to make a simple is_pointer type trait work, but not much else. llvm-svn: 72662
2009-05-31 17:31:02 +08:00
}
Implement support for non-type template parameter packs whose type is a pack expansion, e.g., the parameter pack Values in: template<typename ...Types> struct Outer { template<Types ...Values> struct Inner; }; This new implementation approach introduces the notion of an "expanded" non-type template parameter pack, for which we have already expanded the types of the parameter pack (to, say, "int*, float*", for Outer<int*, float*>) but have not yet expanded the values. Aside from creating these expanded non-type template parameter packs, this patch updates template argument checking and non-type template parameter pack instantiation to make use of the appropriate types in the parameter pack. llvm-svn: 123845
2011-01-20 04:10:05 +08:00
// If we're dealing with a member template where the template parameters
// have been instantiated, this provides the original template parameters
// from which the member template's parameters were instantiated.
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVector<const NamedDecl *, 4> InstantiatedTemplateParameters;
Implement support for non-type template parameter packs whose type is a pack expansion, e.g., the parameter pack Values in: template<typename ...Types> struct Outer { template<Types ...Values> struct Inner; }; This new implementation approach introduces the notion of an "expanded" non-type template parameter pack, for which we have already expanded the types of the parameter pack (to, say, "int*, float*", for Outer<int*, float*>) but have not yet expanded the values. Aside from creating these expanded non-type template parameter packs, this patch updates template argument checking and non-type template parameter pack instantiation to make use of the appropriate types in the parameter pack. llvm-svn: 123845
2011-01-20 04:10:05 +08:00
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
if (Matched.size() >= 1) {
Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily. llvm-svn: 185610
2013-07-04 11:08:24 +08:00
SmallVectorImpl<MatchResult>::iterator Best = Matched.begin();
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
if (Matched.size() == 1) {
// -- If exactly one matching specialization is found, the
// instantiation is generated from that specialization.
// We don't need to do anything for this.
} else {
// -- If more than one matching specialization is found, the
// partial order rules (14.5.4.2) are used to determine
// whether one of the specializations is more specialized
// than the others. If none of the specializations is more
// specialized than all of the other matching
// specializations, then the use of the class template is
// ambiguous and the program is ill-formed.
Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily. llvm-svn: 185610
2013-07-04 11:08:24 +08:00
for (SmallVectorImpl<MatchResult>::iterator P = Best + 1,
PEnd = Matched.end();
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
P != PEnd; ++P) {
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
if (getMoreSpecializedPartialSpecialization(P->Partial, Best->Partial,
Thread a source location into the template-argument deduction routines. There may be some other places that could take advantage of this new information, but I haven't really looked yet. llvm-svn: 95600
2010-02-09 07:07:23 +08:00
PointOfInstantiation)
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
== P->Partial)
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
Best = P;
Implement partial ordering of class template partial specializations (C++ [temp.class.order]). llvm-svn: 81866
2009-09-16 00:23:51 +08:00
}
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
// Determine if the best partial specialization is more specialized than
// the others.
bool Ambiguous = false;
Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily. llvm-svn: 185610
2013-07-04 11:08:24 +08:00
for (SmallVectorImpl<MatchResult>::iterator P = Matched.begin(),
PEnd = Matched.end();
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
P != PEnd; ++P) {
if (P != Best &&
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
getMoreSpecializedPartialSpecialization(P->Partial, Best->Partial,
Thread a source location into the template-argument deduction routines. There may be some other places that could take advantage of this new information, but I haven't really looked yet. llvm-svn: 95600
2010-02-09 07:07:23 +08:00
PointOfInstantiation)
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
!= Best->Partial) {
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
Ambiguous = true;
break;
}
}
if (Ambiguous) {
// Partial ordering did not produce a clear winner. Complain.
ClassTemplateSpec->setInvalidDecl();
Diag(PointOfInstantiation, diag::err_partial_spec_ordering_ambiguous)
<< ClassTemplateSpec;
// Print the matching partial specializations.
Use SmallVectorImpl instead of SmallVector for iterators and references to avoid specifying the vector size unnecessarily. llvm-svn: 185610
2013-07-04 11:08:24 +08:00
for (SmallVectorImpl<MatchResult>::iterator P = Matched.begin(),
PEnd = Matched.end();
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
P != PEnd; ++P)
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
Diag(P->Partial->getLocation(), diag::note_partial_spec_match)
<< getTemplateArgumentBindingsText(
P->Partial->getTemplateParameters(),
*P->Args);
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
return true;
}
Implement partial ordering of class template partial specializations (C++ [temp.class.order]). llvm-svn: 81866
2009-09-16 00:23:51 +08:00
}
// Instantiate using the best class template partial specialization.
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
ClassTemplatePartialSpecializationDecl *OrigPartialSpec = Best->Partial;
Implement support for semantic checking and template instantiation of class template partial specializations of member templates. Also, fixes a silly little bug in the marking of "used" template parameters in member templates. Fixes PR5236. llvm-svn: 85447
2009-10-29 08:04:11 +08:00
while (OrigPartialSpec->getInstantiatedFromMember()) {
// If we've found an explicit specialization of this class template,
// stop here and use that as the pattern.
if (OrigPartialSpec->isMemberSpecialization())
break;
OrigPartialSpec = OrigPartialSpec->getInstantiatedFromMember();
}
Pattern = OrigPartialSpec;
Introduce support for emitting diagnostics (warnings + their notes) that are suppressed during template argument deduction. This change queues diagnostics computed during template argument deduction. Then, if the resulting function template specialization or partial specialization is chosen by overload resolution or partial ordering (respectively), we will emit the queued diagnostics at that point. This addresses most of PR6784. However, the check for unnamed/local template arguments (which existed before this change) is still only skin-deep, and needs to be extended to look deeper into types. It must be improved to finish PR6784. llvm-svn: 116373
2010-10-13 07:32:35 +08:00
ClassTemplateSpec->setInstantiationOf(Best->Partial, Best->Args);
It looks like we've finished off matching of class template partial specializations; add comments and update the C++ status page llvm-svn: 73263
2009-06-13 06:31:52 +08:00
} else {
// -- If no matches are found, the instantiation is generated
// from the primary template.
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
ClassTemplateDecl *OrigTemplate = Template;
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
while (OrigTemplate->getInstantiatedFromMemberTemplate()) {
// If we've found an explicit specialization of this class template,
// stop here and use that as the pattern.
if (OrigTemplate->isMemberSpecialization())
break;
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
OrigTemplate = OrigTemplate->getInstantiatedFromMemberTemplate();
When explicitly specializing a member that is a template, mark the template as a specialization. For example, this occurs with: template<typename T> struct X { template<typename U> struct Inner { /* ... */ }; }; template<> template<typename T> struct X<int>::Inner { T member; }; We need to treat templates that are member specializations as special in two contexts: - When looking for a definition of a member template, we look through the instantiation chain until we hit the primary template *or a member specialization*. This allows us to distinguish between the primary "Inner" definition and the X<int>::Inner definition, above. - When computing all of the levels of template arguments needed to instantiate a member template, don't add template arguments from contexts outside of the instantiation of a member specialization, since the user has already manually substituted those arguments. Fix up the existing test for p18, which was actually wrong (but we didn't diagnose it because of our poor handling of member specializations of templates), and add a new test for member specializations of templates. llvm-svn: 83974
2009-10-14 00:30:37 +08:00
}
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
Pattern = OrigTemplate->getTemplatedDecl();
Initial infrastructure for class template partial specialization. Here we have the basics of declaring and storing class template partial specializations, matching class template partial specializations at instantiation time via (limited) template argument deduction, and using the class template partial specialization's pattern for instantiation. This patch is enough to make a simple is_pointer type trait work, but not much else. llvm-svn: 72662
2009-05-31 17:31:02 +08:00
}
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
Only set the point of instantiation for an implicit or explicit instantiation once we have committed to performing the instantiation. As part of this, make our makeshift template-instantiation location information suck slightly less. Fixes PR5264. llvm-svn: 85209
2009-10-27 14:26:26 +08:00
bool Result = InstantiateClass(PointOfInstantiation, ClassTemplateSpec,
Pattern,
getTemplateInstantiationArgs(ClassTemplateSpec),
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
TSK,
Try to complete a type before looking for conversion functions within that type. Note that we do not produce a diagnostic if the type is incomplete; rather, we just don't look for conversion functions. Fixes PR4660. llvm-svn: 79919
2009-08-24 23:23:48 +08:00
Complain);
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Several improvements to template argument deduction: - Once we have deduced template arguments for a class template partial specialization, we use exactly those template arguments for instantiating the definition of the class template partial specialization. - Added template argument deduction for non-type template parameters. - Added template argument deduction for dependently-sized array types. With these changes, we can now implement, e.g., the remove_reference type trait. Also, Daniel's Ackermann template metaprogram now compiles properly. llvm-svn: 72909
2009-06-05 08:53:49 +08:00
return Result;
Instantiation for member classes of class templates. Note that only the declarations of member classes are instantiated when the owning class template is instantiated. The definitions of such member classes are instantiated when a complete type is required. This change also introduces the injected-class-name into a class template specialization. llvm-svn: 67707
2009-03-26 05:17:03 +08:00
}
Clarify the difference between substitution and instantiation by renaming functions that don't instantiate definitions. llvm-svn: 80037
2009-08-26 06:02:44 +08:00
/// \brief Instantiates the definitions of all of the member
/// of the given class, which is an instantiation of a class template
/// or a member class of a template.
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
void
Sema::InstantiateClassMembers(SourceLocation PointOfInstantiation,
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
CXXRecordDecl *Instantiation,
const MultiLevelTemplateArgumentList &TemplateArgs,
TemplateSpecializationKind TSK) {
De-ASTContext-ify DeclContext. Remove ASTContext parameter from DeclContext's methods. This change cascaded down to other Decl's methods and changes to call sites started "escalating". Timings using pre-tokenized "cocoa.h" showed only a ~1% increase in time run between and after this commit. llvm-svn: 74506
2009-06-30 10:36:12 +08:00
for (DeclContext::decl_iterator D = Instantiation->decls_begin(),
DEnd = Instantiation->decls_end();
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
D != DEnd; ++D) {
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
bool SuppressNew = false;
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
if (FunctionDecl *Function = dyn_cast<FunctionDecl>(*D)) {
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
if (FunctionDecl *Pattern
= Function->getInstantiatedFromMemberFunction()) {
MemberSpecializationInfo *MSInfo
= Function->getMemberSpecializationInfo();
assert(MSInfo && "No member specialization information?");
Only complain about explicit instantiations following explicit specializations when the explicit instantiation was... explicitly written, i.e., not the product of an explicit instantiation of an enclosing class. Fixes this spurious warning when Clang builds LLVM: /Volumes/Data/dgregor/Projects/llvm/lib/CodeGen/MachineDominators.cpp:22:1: warning: explicit instantiation of 'addRoot' that occurs after an explicit specialization will be ignored (C++0x extension) [-pedantic] llvm-svn: 100900
2010-04-10 05:02:29 +08:00
if (MSInfo->getTemplateSpecializationKind()
== TSK_ExplicitSpecialization)
continue;
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
if (CheckSpecializationInstantiationRedecl(PointOfInstantiation, TSK,
Function,
MSInfo->getTemplateSpecializationKind(),
Correctly set "explicit template instantiation" kind on inner structs of templates whose explicit instantiation is first declared and then defined. Fixes http://llvm.org/pr8207 llvm-svn: 114874
2010-09-28 05:02:09 +08:00
MSInfo->getPointOfInstantiation(),
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
SuppressNew) ||
SuppressNew)
Improve checking for specializations of member classes of class templates, and keep track of how those member classes were instantiated or specialized. Make sure that we don't try to instantiate an explicitly-specialized member class of a class template, when that explicit specialization was a declaration rather than a definition. llvm-svn: 83547
2009-10-08 23:14:33 +08:00
continue;
Modify some deleted function methods to better reflect reality: - New isDefined() function checks for deletedness - isThisDeclarationADefinition checks for deletedness - New doesThisDeclarationHaveABody() does what isThisDeclarationADefinition() used to do - The IsDeleted bit is not propagated across redeclarations - isDeleted() now checks the canoncial declaration - New isDeletedAsWritten() does what it says on the tin. - isUserProvided() now correct (thanks Richard!) This fixes the bug that we weren't catching void foo() = delete; void foo() {} as being a redefinition. llvm-svn: 131013
2011-05-07 04:44:56 +08:00
if (Function->isDefined())
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
continue;
if (TSK == TSK_ExplicitInstantiationDefinition) {
// C++0x [temp.explicit]p8:
// An explicit instantiation definition that names a class template
// specialization explicitly instantiates the class template
// specialization and is only an explicit instantiation definition
// of members whose definition is visible at the point of
// instantiation.
Modify some deleted function methods to better reflect reality: - New isDefined() function checks for deletedness - isThisDeclarationADefinition checks for deletedness - New doesThisDeclarationHaveABody() does what isThisDeclarationADefinition() used to do - The IsDeleted bit is not propagated across redeclarations - isDeleted() now checks the canoncial declaration - New isDeletedAsWritten() does what it says on the tin. - isUserProvided() now correct (thanks Richard!) This fixes the bug that we weren't catching void foo() = delete; void foo() {} as being a redefinition. llvm-svn: 131013
2011-05-07 04:44:56 +08:00
if (!Pattern->isDefined())
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
continue;
Function->setTemplateSpecializationKind(TSK, PointOfInstantiation);
InstantiateFunctionDefinition(PointOfInstantiation, Function);
} else {
Function->setTemplateSpecializationKind(TSK, PointOfInstantiation);
}
Improve checking for specializations of member classes of class templates, and keep track of how those member classes were instantiated or specialized. Make sure that we don't try to instantiate an explicitly-specialized member class of a class template, when that explicit specialization was a declaration rather than a definition. llvm-svn: 83547
2009-10-08 23:14:33 +08:00
}
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
} else if (VarDecl *Var = dyn_cast<VarDecl>(*D)) {
Variable templates: handle instantiation of static data member templates appropriately, especially when they appear within class templates. llvm-svn: 191548
2013-09-28 04:14:12 +08:00
if (isa<VarTemplateSpecializationDecl>(Var))
continue;
For instantiations of static data members of class templates, keep track of the kind of specialization or instantiation. Also, check the scope of the specialization and ensure that a specialization declaration without an initializer is not a definition. llvm-svn: 83533
2009-10-08 15:24:58 +08:00
if (Var->isStaticDataMember()) {
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
MemberSpecializationInfo *MSInfo = Var->getMemberSpecializationInfo();
assert(MSInfo && "No member specialization information?");
Only complain about explicit instantiations following explicit specializations when the explicit instantiation was... explicitly written, i.e., not the product of an explicit instantiation of an enclosing class. Fixes this spurious warning when Clang builds LLVM: /Volumes/Data/dgregor/Projects/llvm/lib/CodeGen/MachineDominators.cpp:22:1: warning: explicit instantiation of 'addRoot' that occurs after an explicit specialization will be ignored (C++0x extension) [-pedantic] llvm-svn: 100900
2010-04-10 05:02:29 +08:00
if (MSInfo->getTemplateSpecializationKind()
== TSK_ExplicitSpecialization)
continue;
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
if (CheckSpecializationInstantiationRedecl(PointOfInstantiation, TSK,
Var,
MSInfo->getTemplateSpecializationKind(),
Correctly set "explicit template instantiation" kind on inner structs of templates whose explicit instantiation is first declared and then defined. Fixes http://llvm.org/pr8207 llvm-svn: 114874
2010-09-28 05:02:09 +08:00
MSInfo->getPointOfInstantiation(),
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
SuppressNew) ||
SuppressNew)
Improve checking for specializations of member classes of class templates, and keep track of how those member classes were instantiated or specialized. Make sure that we don't try to instantiate an explicitly-specialized member class of a class template, when that explicit specialization was a declaration rather than a definition. llvm-svn: 83547
2009-10-08 23:14:33 +08:00
continue;
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
if (TSK == TSK_ExplicitInstantiationDefinition) {
// C++0x [temp.explicit]p8:
// An explicit instantiation definition that names a class template
// specialization explicitly instantiates the class template
// specialization and is only an explicit instantiation definition
// of members whose definition is visible at the point of
// instantiation.
if (!Var->getInstantiatedFromStaticDataMember()
->getOutOfLineDefinition())
continue;
Var->setTemplateSpecializationKind(TSK, PointOfInstantiation);
For instantiations of static data members of class templates, keep track of the kind of specialization or instantiation. Also, check the scope of the specialization and ensure that a specialization declaration without an initializer is not a definition. llvm-svn: 83533
2009-10-08 15:24:58 +08:00
InstantiateStaticDataMemberDefinition(PointOfInstantiation, Var);
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
} else {
Var->setTemplateSpecializationKind(TSK, PointOfInstantiation);
}
}
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
} else if (CXXRecordDecl *Record = dyn_cast<CXXRecordDecl>(*D)) {
Make sure that we don't visit redeclarations of nested classes while instantiating class members as part of an explicit instantiation. Addresses a compilation problem in Boost.Serialization. llvm-svn: 101725
2010-04-19 02:11:38 +08:00
// Always skip the injected-class-name, along with any
// redeclarations of nested classes, since both would cause us
// to try to instantiate the members of a class twice.
De-virtualize getPreviousDecl() and getMostRecentDecl() when we know we have a redeclarable type, and only use the new virtual versions (getPreviousDeclImpl() and getMostRecentDeclImpl()) when we don't have that type information. This keeps us from penalizing users with strict type information (and is the moral equivalent of a "final" method). Plus, settle on the names getPreviousDecl() and getMostRecentDecl() throughout. llvm-svn: 148187
2012-01-15 00:38:05 +08:00
if (Record->isInjectedClassName() || Record->getPreviousDecl())
Keep track of whether a member function instantiated from a member function of a class template was implicitly instantiated, explicitly instantiated (declaration or definition), or explicitly specialized. The same MemberSpecializationInfo structure will be used for static data members and member classes as well. llvm-svn: 83509
2009-10-08 07:56:10 +08:00
continue;
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
MemberSpecializationInfo *MSInfo = Record->getMemberSpecializationInfo();
assert(MSInfo && "No member specialization information?");
Only complain about explicit instantiations following explicit specializations when the explicit instantiation was... explicitly written, i.e., not the product of an explicit instantiation of an enclosing class. Fixes this spurious warning when Clang builds LLVM: /Volumes/Data/dgregor/Projects/llvm/lib/CodeGen/MachineDominators.cpp:22:1: warning: explicit instantiation of 'addRoot' that occurs after an explicit specialization will be ignored (C++0x extension) [-pedantic] llvm-svn: 100900
2010-04-10 05:02:29 +08:00
if (MSInfo->getTemplateSpecializationKind()
== TSK_ExplicitSpecialization)
continue;
Correctly set "explicit template instantiation" kind on inner structs of templates whose explicit instantiation is first declared and then defined. Fixes http://llvm.org/pr8207 llvm-svn: 114874
2010-09-28 05:02:09 +08:00
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
if (CheckSpecializationInstantiationRedecl(PointOfInstantiation, TSK,
Record,
MSInfo->getTemplateSpecializationKind(),
Correctly set "explicit template instantiation" kind on inner structs of templates whose explicit instantiation is first declared and then defined. Fixes http://llvm.org/pr8207 llvm-svn: 114874
2010-09-28 05:02:09 +08:00
MSInfo->getPointOfInstantiation(),
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
SuppressNew) ||
SuppressNew)
Improve checking for specializations of member classes of class templates, and keep track of how those member classes were instantiated or specialized. Make sure that we don't try to instantiate an explicitly-specialized member class of a class template, when that explicit specialization was a declaration rather than a definition. llvm-svn: 83547
2009-10-08 23:14:33 +08:00
continue;
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
CXXRecordDecl *Pattern = Record->getInstantiatedFromMemberClass();
assert(Pattern && "Missing instantiated-from-template information");
Eliminate the ASTContext parameter from RecordDecl::getDefinition() and CXXRecordDecl::getDefinition(); it's totally unnecessary. No functionality change. llvm-svn: 95836
2010-02-11 09:04:33 +08:00
if (!Record->getDefinition()) {
if (!Pattern->getDefinition()) {
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
// C++0x [temp.explicit]p8:
// An explicit instantiation definition that names a class template
// specialization explicitly instantiates the class template
// specialization and is only an explicit instantiation definition
// of members whose definition is visible at the point of
// instantiation.
if (TSK == TSK_ExplicitInstantiationDeclaration) {
MSInfo->setTemplateSpecializationKind(TSK);
MSInfo->setPointOfInstantiation(PointOfInstantiation);
}
continue;
}
InstantiateClass(PointOfInstantiation, Record, Pattern,
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
TemplateArgs,
TSK);
Correctly set "explicit template instantiation" kind on inner structs of templates whose explicit instantiation is first declared and then defined. Fixes http://llvm.org/pr8207 llvm-svn: 114874
2010-09-28 05:02:09 +08:00
} else {
if (TSK == TSK_ExplicitInstantiationDefinition &&
Record->getTemplateSpecializationKind() ==
TSK_ExplicitInstantiationDeclaration) {
Record->setTemplateSpecializationKind(TSK);
MarkVTableUsed(PointOfInstantiation, Record, true);
}
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
}
Make sure to set the template specialization kind of an explicit template instantiation of a member function of a class template. FIXME -= 2; llvm-svn: 83520
2009-10-08 09:19:17 +08:00
Eliminate the ASTContext parameter from RecordDecl::getDefinition() and CXXRecordDecl::getDefinition(); it's totally unnecessary. No functionality change. llvm-svn: 95836
2010-02-11 09:04:33 +08:00
Pattern = cast_or_null<CXXRecordDecl>(Record->getDefinition());
An explicit instantiation definition only instantiations those class members that have a definition. Also, use CheckSpecializationInstantiationRedecl as part of this instantiation to make sure that we diagnose the various kinds of problems that can occur with explicit instantiations. llvm-svn: 85270
2009-10-28 02:42:08 +08:00
if (Pattern)
InstantiateClassMembers(PointOfInstantiation, Pattern, TemplateArgs,
TSK);
Instantiating a class template should not instantiate the definition of any scoped enumeration members. Later uses of an enumeration temploid as a nested name specifier should cause its instantiation. Plus some groundwork for explicit specialization of member enumerations of class templates. llvm-svn: 152750
2012-03-15 07:13:10 +08:00
} else if (EnumDecl *Enum = dyn_cast<EnumDecl>(*D)) {
MemberSpecializationInfo *MSInfo = Enum->getMemberSpecializationInfo();
assert(MSInfo && "No member specialization information?");
if (MSInfo->getTemplateSpecializationKind()
== TSK_ExplicitSpecialization)
continue;
if (CheckSpecializationInstantiationRedecl(
PointOfInstantiation, TSK, Enum,
MSInfo->getTemplateSpecializationKind(),
MSInfo->getPointOfInstantiation(), SuppressNew) ||
SuppressNew)
continue;
if (Enum->getDefinition())
continue;
EnumDecl *Pattern = Enum->getInstantiatedFromMemberEnum();
assert(Pattern && "Missing instantiated-from-template information");
if (TSK == TSK_ExplicitInstantiationDefinition) {
if (!Pattern->getDefinition())
continue;
InstantiateEnum(PointOfInstantiation, Enum, Pattern, TemplateArgs, TSK);
} else {
MSInfo->setTemplateSpecializationKind(TSK);
MSInfo->setPointOfInstantiation(PointOfInstantiation);
}
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
}
}
}
/// \brief Instantiate the definitions of all of the members of the
/// given class template specialization, which was named as part of an
/// explicit instantiation.
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
void
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
Sema::InstantiateClassTemplateSpecializationMembers(
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
SourceLocation PointOfInstantiation,
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
ClassTemplateSpecializationDecl *ClassTemplateSpec,
TemplateSpecializationKind TSK) {
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
// C++0x [temp.explicit]p7:
// An explicit instantiation that names a class template
// specialization is an explicit instantion of the same kind
// (declaration or definition) of each of its members (not
// including members inherited from base classes) that has not
// been previously explicitly specialized in the translation unit
// containing the explicit instantiation, except as described
// below.
InstantiateClassMembers(PointOfInstantiation, ClassTemplateSpec,
Improve the AST representation and semantic analysis for extern templates. We now distinguish between an explicit instantiation declaration and an explicit instantiation definition, and know not to instantiate explicit instantiation declarations. Unfortunately, there is some remaining confusion w.r.t. instantiation of out-of-line member function definitions that causes trouble here. llvm-svn: 81053
2009-09-05 06:48:11 +08:00
getTemplateInstantiationArgs(ClassTemplateSpec),
TSK);
Explicit instantiations of templates now instantiate the definitions of class members (recursively). Only member classes are actually instantiated; the instantiation logic for member functions and variables are just stubs. llvm-svn: 71713
2009-05-14 04:28:22 +08:00
}
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
StmtResult
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
Sema::SubstStmt(Stmt *S, const MultiLevelTemplateArgumentList &TemplateArgs) {
Refactor the instantiation of statements into a generic tree transformation. llvm-svn: 79519
2009-08-20 15:17:43 +08:00
if (!S)
return Owned(S);
TemplateInstantiator Instantiator(*this, TemplateArgs,
SourceLocation(),
DeclarationName());
return Instantiator.TransformStmt(S);
}
OwningExprResult -> ExprResult. This patch brought to you by M-x query-replace-regexp \(Sema::\|Action::\|Parser::\|\)Owning\(Expr\|Stmt\)Result -> \2Result llvm-svn: 111903
2010-08-24 14:29:42 +08:00
ExprResult
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
Sema::SubstExpr(Expr *E, const MultiLevelTemplateArgumentList &TemplateArgs) {
Refactor the template-instantiation logic for expressions into a generic tree transformation (also used for recanonicalization) and a small amount of template-instantiation-specific logic. llvm-svn: 78645
2009-08-11 13:31:07 +08:00
if (!E)
return Owned(E);
Remove tabs, and whitespace cleanups. llvm-svn: 81346
2009-09-09 23:08:12 +08:00
Refactor the template-instantiation logic for expressions into a generic tree transformation (also used for recanonicalization) and a small amount of template-instantiation-specific logic. llvm-svn: 78645
2009-08-11 13:31:07 +08:00
TemplateInstantiator Instantiator(*this, TemplateArgs,
SourceLocation(),
DeclarationName());
return Instantiator.TransformExpr(E);
}
PR13470: Ensure that copy-list-initialization isntantiates as copy-list-initialization (and doesn't add an additional copy step): Fill in the ListInitialization bit when creating a CXXConstructExpr. Use it when instantiating initializers in order to correctly handle instantiation of copy-list-initialization. Teach TreeTransform that function arguments are initializations, and so need this special treatment too. Finally, remove some hacks which were working around SubstInitializer's shortcomings. llvm-svn: 170489
2012-12-19 09:39:02 +08:00
ExprResult Sema::SubstInitializer(Expr *Init,
const MultiLevelTemplateArgumentList &TemplateArgs,
bool CXXDirectInit) {
TemplateInstantiator Instantiator(*this, TemplateArgs,
SourceLocation(),
DeclarationName());
return Instantiator.TransformInitializer(Init, CXXDirectInit);
}
When instantiating the arguments to an initializer, use the TreeTransform version of TransformExprs() rather than explicit loop, so that we expand pack expansions properly. Test cast coming soon... llvm-svn: 123014
2011-01-08 03:35:17 +08:00
bool Sema::SubstExprs(Expr **Exprs, unsigned NumExprs, bool IsCall,
const MultiLevelTemplateArgumentList &TemplateArgs,
remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports them into the clang namespace. llvm-svn: 135852
2011-07-23 18:55:15 +08:00
SmallVectorImpl<Expr *> &Outputs) {
When instantiating the arguments to an initializer, use the TreeTransform version of TransformExprs() rather than explicit loop, so that we expand pack expansions properly. Test cast coming soon... llvm-svn: 123014
2011-01-08 03:35:17 +08:00
if (NumExprs == 0)
return false;
PR13470: Ensure that copy-list-initialization isntantiates as copy-list-initialization (and doesn't add an additional copy step): Fill in the ListInitialization bit when creating a CXXConstructExpr. Use it when instantiating initializers in order to correctly handle instantiation of copy-list-initialization. Teach TreeTransform that function arguments are initializations, and so need this special treatment too. Finally, remove some hacks which were working around SubstInitializer's shortcomings. llvm-svn: 170489
2012-12-19 09:39:02 +08:00
When instantiating the arguments to an initializer, use the TreeTransform version of TransformExprs() rather than explicit loop, so that we expand pack expansions properly. Test cast coming soon... llvm-svn: 123014
2011-01-08 03:35:17 +08:00
TemplateInstantiator Instantiator(*this, TemplateArgs,
SourceLocation(),
DeclarationName());
return Instantiator.TransformExprs(Exprs, NumExprs, IsCall, Outputs);
}
Use NestedNameSpecifierLoc within out-of-line variables, function, and tag definitions. Also, add support for template instantiation of NestedNameSpecifierLocs. llvm-svn: 126470
2011-02-25 10:25:35 +08:00
NestedNameSpecifierLoc
Sema::SubstNestedNameSpecifierLoc(NestedNameSpecifierLoc NNS,
const MultiLevelTemplateArgumentList &TemplateArgs) {
if (!NNS)
return NestedNameSpecifierLoc();
TemplateInstantiator Instantiator(*this, TemplateArgs, NNS.getBeginLoc(),
DeclarationName());
return Instantiator.TransformNestedNameSpecifierLoc(NNS);
}
Added locations and type source info for DeclarationName. llvm-svn: 110860
2010-08-12 06:01:17 +08:00
/// \brief Do template substitution on declaration name info.
DeclarationNameInfo
Sema::SubstDeclarationNameInfo(const DeclarationNameInfo &NameInfo,
const MultiLevelTemplateArgumentList &TemplateArgs) {
TemplateInstantiator Instantiator(*this, TemplateArgs, NameInfo.getLoc(),
NameInfo.getName());
return Instantiator.TransformDeclarationNameInfo(NameInfo);
}
Implement template instantiation for template names, including both template template parameters and dependent template names. For example, the oft-mentioned typename MetaFun::template apply<T1, T2>::type can now be instantiated, with the appropriate name lookup for "apply". llvm-svn: 68128
2009-04-01 02:38:02 +08:00
TemplateName
Kill off the TreeTransform::TransformTemplateName overload that has poor source-location information. llvm-svn: 126852
2011-03-03 02:46:51 +08:00
Sema::SubstTemplateName(NestedNameSpecifierLoc QualifierLoc,
TemplateName Name, SourceLocation Loc,
Implement template instantiation for member class templates. When performing template instantiation of the definitions of member templates (or members thereof), we build a data structure containing the template arguments from each "level" of template instantiation. During template instantiation, we substitute all levels of template arguments simultaneously. llvm-svn: 80389
2009-08-29 04:31:08 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs) {
Implement transformation of template names within the generic tree transform, then use the result for template instantiation. The generic transformation fixes a few issues: - It copes better with template template parameters and member templates (when they're implemented). - The logic used to replace template template parameters with their arguments is now centralized in TransformDecl, so that it will apply for other declaration-instantiation steps. - The error-recovery strategy is normalized now, so that any error results in a NULL TemplateName. llvm-svn: 78292
2009-08-06 14:41:21 +08:00
TemplateInstantiator Instantiator(*this, TemplateArgs, Loc,
DeclarationName());
Kill off the TreeTransform::TransformTemplateName overload that has poor source-location information. llvm-svn: 126852
2011-03-03 02:46:51 +08:00
CXXScopeSpec SS;
SS.Adopt(QualifierLoc);
return Instantiator.TransformTemplateName(SS, Name, Loc);
Implement template instantiation for template names, including both template template parameters and dependent template names. For example, the oft-mentioned typename MetaFun::template apply<T1, T2>::type can now be instantiated, with the appropriate name lookup for "apply". llvm-svn: 68128
2009-04-01 02:38:02 +08:00
}
Separate TemplateArgument instantiation logic into its own function. No functionality change. llvm-svn: 73176
2009-06-11 08:06:24 +08:00
Implement template argument deduction for pack expansions whose pattern is a template argument, which involves repeatedly deducing template arguments using the pattern of the pack expansion, then bundling the resulting deductions into an argument pack. We can now handle a variety of simple list-handling metaprograms using variadic templates. See, e.g., the new "count" metaprogram. llvm-svn: 122439
2010-12-23 05:19:48 +08:00
bool Sema::Subst(const TemplateArgumentLoc *Args, unsigned NumArgs,
TemplateArgumentListInfo &Result,
Track source information for template arguments and template specialization types. Preserve it through template instantiation. Preserve it through PCH, although TSTs themselves aren't serializable, so that's pretty much meaningless. llvm-svn: 85500
2009-10-29 16:12:44 +08:00
const MultiLevelTemplateArgumentList &TemplateArgs) {
Move the template instantiation logic for template arguments into the general tree transformation. Also, implement template instantiation for parameter packs. In addition, introduce logic to enter the appropriate context for subexpressions that are not potentially evaluated. llvm-svn: 78114
2009-08-05 06:27:00 +08:00
TemplateInstantiator Instantiator(*this, TemplateArgs, SourceLocation(),
DeclarationName());
Implement template argument deduction for pack expansions whose pattern is a template argument, which involves repeatedly deducing template arguments using the pattern of the pack expansion, then bundling the resulting deductions into an argument pack. We can now handle a variety of simple list-handling metaprograms using variadic templates. See, e.g., the new "count" metaprogram. llvm-svn: 122439
2010-12-23 05:19:48 +08:00
return Instantiator.TransformTemplateArguments(Args, NumArgs, Result);
Separate TemplateArgument instantiation logic into its own function. No functionality change. llvm-svn: 73176
2009-06-11 08:06:24 +08:00
}
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
Fix template instantiation of attributes. More specifically, fix the case where an attribute is attached to a forward declaration of a template function, and refers to parameters of that declaration, but is then inherited by the definition of that function. When the definition is instantiated, the parameter references need to be remapped. llvm-svn: 164710
2012-09-27 01:57:31 +08:00
static const Decl* getCanonicalParmVarDecl(const Decl *D) {
// When storing ParmVarDecls in the local instantiation scope, we always
// want to use the ParmVarDecl from the canonical function declaration,
// since the map is then valid for any redeclaration or definition of that
// function.
if (const ParmVarDecl *PV = dyn_cast<ParmVarDecl>(D)) {
if (const FunctionDecl *FD = dyn_cast<FunctionDecl>(PV->getDeclContext())) {
unsigned i = PV->getFunctionScopeIndex();
return FD->getCanonicalDecl()->getParamDecl(i);
}
}
return D;
}
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
llvm::PointerUnion<Decl *, LocalInstantiationScope::DeclArgumentPack *> *
LocalInstantiationScope::findInstantiationOf(const Decl *D) {
Fix template instantiation of attributes. More specifically, fix the case where an attribute is attached to a forward declaration of a template function, and refers to parameters of that declaration, but is then inherited by the definition of that function. When the definition is instantiated, the parameter references need to be remapped. llvm-svn: 164710
2012-09-27 01:57:31 +08:00
D = getCanonicalParmVarDecl(D);
Step #2/N of __label__ support: keep pushing LabelDecl forward, making them be template instantiated in a more normal way and make them handle attributes like other decls. This fixes the used/unused label handling stuff, making it use the same infrastructure as other decls. llvm-svn: 125771
2011-02-18 04:34:02 +08:00
for (LocalInstantiationScope *Current = this; Current;
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
Current = Current->Outer) {
Step #2/N of __label__ support: keep pushing LabelDecl forward, making them be template instantiated in a more normal way and make them handle attributes like other decls. This fixes the used/unused label handling stuff, making it use the same infrastructure as other decls. llvm-svn: 125771
2011-02-18 04:34:02 +08:00
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
// Check if we found something within this scope.
When searching for the instantiation of a locally-scoped tag declaration, also look for an instantiation of its previous declarations. Fixes PR8801. llvm-svn: 122361
2010-12-22 05:22:51 +08:00
const Decl *CheckD = D;
do {
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
LocalDeclsMap::iterator Found = Current->LocalDecls.find(CheckD);
When searching for the instantiation of a locally-scoped tag declaration, also look for an instantiation of its previous declarations. Fixes PR8801. llvm-svn: 122361
2010-12-22 05:22:51 +08:00
if (Found != Current->LocalDecls.end())
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
return &Found->second;
When searching for the instantiation of a locally-scoped tag declaration, also look for an instantiation of its previous declarations. Fixes PR8801. llvm-svn: 122361
2010-12-22 05:22:51 +08:00
// If this is a tag declaration, it's possible that we need to look for
// a previous declaration.
if (const TagDecl *Tag = dyn_cast<TagDecl>(CheckD))
De-virtualize getPreviousDecl() and getMostRecentDecl() when we know we have a redeclarable type, and only use the new virtual versions (getPreviousDeclImpl() and getMostRecentDeclImpl()) when we don't have that type information. This keeps us from penalizing users with strict type information (and is the moral equivalent of a "final" method). Plus, settle on the names getPreviousDecl() and getMostRecentDecl() throughout. llvm-svn: 148187
2012-01-15 00:38:05 +08:00
CheckD = Tag->getPreviousDecl();
When searching for the instantiation of a locally-scoped tag declaration, also look for an instantiation of its previous declarations. Fixes PR8801. llvm-svn: 122361
2010-12-22 05:22:51 +08:00
else
CheckD = 0;
} while (CheckD);
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
// If we aren't combined with our outer scope, we're done.
if (!Current->CombineWithOuterScope)
break;
}
Step #2/N of __label__ support: keep pushing LabelDecl forward, making them be template instantiated in a more normal way and make them handle attributes like other decls. This fixes the used/unused label handling stuff, making it use the same infrastructure as other decls. llvm-svn: 125771
2011-02-18 04:34:02 +08:00
Fix to PR12262 - assertion when substituting explicit template arguments does not substitute a sizeof-pack expression. The solution is proposed by Richard Smith. Differential Revision: http://llvm-reviews.chandlerc.com/D869 llvm-svn: 186306
2013-07-15 14:14:07 +08:00
// If we're performing a partial substitution during template argument
// deduction, we may not have values for template parameters yet.
if (isa<NonTypeTemplateParmDecl>(D) || isa<TemplateTypeParmDecl>(D) ||
isa<TemplateTemplateParmDecl>(D))
return 0;
Step #2/N of __label__ support: keep pushing LabelDecl forward, making them be template instantiated in a more normal way and make them handle attributes like other decls. This fixes the used/unused label handling stuff, making it use the same infrastructure as other decls. llvm-svn: 125771
2011-02-18 04:34:02 +08:00
// If we didn't find the decl, then we either have a sema bug, or we have a
// forward reference to a label declaration. Return null to indicate that
// we have an uninstantiated label.
assert(isa<LabelDecl>(D) && "declaration not instantiated in this scope");
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
return 0;
}
Move more stuff out of Sema.h. llvm-svn: 112026
2010-08-25 13:32:35 +08:00
void LocalInstantiationScope::InstantiatedLocal(const Decl *D, Decl *Inst) {
Fix template instantiation of attributes. More specifically, fix the case where an attribute is attached to a forward declaration of a template function, and refers to parameters of that declaration, but is then inherited by the definition of that function. When the definition is instantiated, the parameter references need to be remapped. llvm-svn: 164710
2012-09-27 01:57:31 +08:00
D = getCanonicalParmVarDecl(D);
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
llvm::PointerUnion<Decl *, DeclArgumentPack *> &Stored = LocalDecls[D];
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
if (Stored.isNull())
Stored = Inst;
Sema: Give a typically small DenseMap some inline capacity. Also reflow code a bit, no change in functionality. llvm-svn: 179382
2013-04-12 23:22:25 +08:00
else if (DeclArgumentPack *Pack = Stored.dyn_cast<DeclArgumentPack *>())
Pack->push_back(Inst);
else
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
assert(Stored.get<Decl *>() == Inst && "Already instantiated this local");
Clean up our handling of local instantiation scopes, which keep track of the mapping from local declarations to their instantiated counterparts during template instantiation. Previously, we tried to do some unholy merging of local instantiation scopes that involved storing a single hash table along with an "undo" list on the side... which was ugly, and never handled function parameters properly. Now, we just keep separate hash tables for each local instantiation scope, and "combining" two scopes means that we'll look in each of the combined hash tables. The combined scope stack is rarely deep, and this makes it easy to avoid the "undo" issues we were hitting. Also, I've simplified the logic for function parameters: if we're declaring a function and we need the function parameters to live longer, we just push them back into the local instantiation scope where we need them. Fixes PR6990. llvm-svn: 102732
2010-05-01 02:55:50 +08:00
}
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
void LocalInstantiationScope::InstantiatedLocalPackArg(const Decl *D,
Decl *Inst) {
Fix template instantiation of attributes. More specifically, fix the case where an attribute is attached to a forward declaration of a template function, and refers to parameters of that declaration, but is then inherited by the definition of that function. When the definition is instantiated, the parameter references need to be remapped. llvm-svn: 164710
2012-09-27 01:57:31 +08:00
D = getCanonicalParmVarDecl(D);
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
DeclArgumentPack *Pack = LocalDecls[D].get<DeclArgumentPack *>();
Pack->push_back(Inst);
}
void LocalInstantiationScope::MakeInstantiatedLocalArgPack(const Decl *D) {
Fix template instantiation of attributes. More specifically, fix the case where an attribute is attached to a forward declaration of a template function, and refers to parameters of that declaration, but is then inherited by the definition of that function. When the definition is instantiated, the parameter references need to be remapped. llvm-svn: 164710
2012-09-27 01:57:31 +08:00
D = getCanonicalParmVarDecl(D);
Implement substitution of a function parameter pack for its set of instantiated function parameters, enabling instantiation of arbitrary pack expansions involving function parameter packs. At this point, we can now correctly compile a simple, variadic print() example: #include <iostream> #include <string> void print() {} template<typename Head, typename ...Tail> void print(const Head &head, const Tail &...tail) { std::cout << head; print(tail...); } int main() { std::string hello = "Hello"; print(hello, ", world!", " ", 2011, '\n'); } llvm-svn: 123000
2011-01-08 00:43:16 +08:00
llvm::PointerUnion<Decl *, DeclArgumentPack *> &Stored = LocalDecls[D];
assert(Stored.isNull() && "Already instantiated this local");
DeclArgumentPack *Pack = new DeclArgumentPack;
Stored = Pack;
ArgumentPacks.push_back(Pack);
}
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which allows an argument pack determines via explicit specification of function template arguments to be extended by further, deduced arguments. For example: template<class ... Types> void f(Types ... values); void g() { f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int } There are a number of FIXMEs in here that indicate places where we need to implement + test retained expansions, plus a number of other places in deduction where we need to correctly cope with the explicitly-specified arguments when deducing an argument pack. Furthermore, it appears that the RecursiveASTVisitor needs to be auditied; it's missing some traversals (especially w.r.t. template arguments) that cause it not to find unexpanded parameter packs when it should. The good news, however, is that the tr1::tuple implementation now works fully, and the tr1::bind example (both from N2080) is actually working now. llvm-svn: 123163
2011-01-10 15:32:04 +08:00
void LocalInstantiationScope::SetPartiallySubstitutedPack(NamedDecl *Pack,
const TemplateArgument *ExplicitArgs,
unsigned NumExplicitArgs) {
assert((!PartiallySubstitutedPack || PartiallySubstitutedPack == Pack) &&
"Already have a partially-substituted pack");
assert((!PartiallySubstitutedPack
|| NumArgsInPartiallySubstitutedPack == NumExplicitArgs) &&
"Wrong number of arguments in partially-substituted pack");
PartiallySubstitutedPack = Pack;
ArgsInPartiallySubstitutedPack = ExplicitArgs;
NumArgsInPartiallySubstitutedPack = NumExplicitArgs;
}
NamedDecl *LocalInstantiationScope::getPartiallySubstitutedPack(
const TemplateArgument **ExplicitArgs,
unsigned *NumExplicitArgs) const {
if (ExplicitArgs)
*ExplicitArgs = 0;
if (NumExplicitArgs)
*NumExplicitArgs = 0;
for (const LocalInstantiationScope *Current = this; Current;
Current = Current->Outer) {
if (Current->PartiallySubstitutedPack) {
if (ExplicitArgs)
*ExplicitArgs = Current->ArgsInPartiallySubstitutedPack;
if (NumExplicitArgs)
*NumExplicitArgs = Current->NumArgsInPartiallySubstitutedPack;
return Current->PartiallySubstitutedPack;
}
if (!Current->CombineWithOuterScope)
break;
}
return 0;
}