[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

The fix was committed, and we just reverted the revert, so default is back to 
`-frelaxed-template-template-args`.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] Revert "[clang] Revert default behavior change of P0522R0 implementation (#91811)" (PR #91837)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov closed 
https://github.com/llvm/llvm-project/pull/91837
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] Revert "[clang] Revert default behavior change of P0522R0 implementation (#91811)" (PR #91837)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov ready_for_review 
https://github.com/llvm/llvm-project/pull/91837
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] Revert "[clang] Revert default behavior change of P0522R0 implementation (#91811)" (PR #91837)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/91837

>From 1a5b4761ba804a0998faf009370d74fa486014d9 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Sat, 11 May 2024 00:42:27 -0300
Subject: [PATCH] Revert "[clang] Revert default behavior change of P0522R0
 implementation (#91811)"

With blocking issues fixed, re-enable relaxed template template argument 
matching
by reverting these commits.

This reverts commit 4198aebc96cb0236fc63e29a92d886e6a2e3fedb.
This reverts commit 2d5634a4b39d8e5497b6a67caa54049b3cfade8e.
---
 clang/lib/Driver/ToolChains/Clang.cpp  | 14 +++---
 .../Driver/frelaxed-template-template-args.cpp |  6 +++---
 clang/test/Driver/rewrite-legacy-objc.m|  6 +++---
 clang/test/Driver/rewrite-objc.m   |  2 +-
 4 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/clang/lib/Driver/ToolChains/Clang.cpp 
b/clang/lib/Driver/ToolChains/Clang.cpp
index f0cc018b6668f..42feb1650574e 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -7249,15 +7249,15 @@ void Clang::ConstructJob(Compilation , const 
JobAction ,
   Args.addOptOutFlag(CmdArgs, options::OPT_fassume_unique_vtables,
  options::OPT_fno_assume_unique_vtables);
 
-  // -fno-relaxed-template-template-args is deprecated.
-  if (Arg *A = Args.getLastArg(options::OPT_frelaxed_template_template_args,
-   
options::OPT_fno_relaxed_template_template_args);
-  A &&
-  A->getOption().matches(options::OPT_fno_relaxed_template_template_args))
+  // -frelaxed-template-template-args is deprecated.
+  if (Arg *A =
+  Args.getLastArg(options::OPT_frelaxed_template_template_args,
+  options::OPT_fno_relaxed_template_template_args)) {
 D.Diag(diag::warn_drv_deprecated_arg)
 << A->getAsString(Args) << /*hasReplacement=*/false;
-  else
-CmdArgs.push_back("-fno-relaxed-template-template-args");
+if 
(A->getOption().matches(options::OPT_fno_relaxed_template_template_args))
+  CmdArgs.push_back("-fno-relaxed-template-template-args");
+  }
 
   // -fsized-deallocation is off by default, as it is an ABI-breaking change 
for
   // most platforms.
diff --git a/clang/test/Driver/frelaxed-template-template-args.cpp 
b/clang/test/Driver/frelaxed-template-template-args.cpp
index 136c360276a15..57fc4e3da6e5d 100644
--- a/clang/test/Driver/frelaxed-template-template-args.cpp
+++ b/clang/test/Driver/frelaxed-template-template-args.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang -fsyntax-only -### %s 2>&1 | FileCheck --check-prefix=CHECK-DEF 
%s
-// RUN: %clang -fsyntax-only -frelaxed-template-template-args %s 2>&1 | 
FileCheck --check-prefix=CHECK-ON --allow-empty %s
+// RUN: %clang -fsyntax-only -frelaxed-template-template-args %s 2>&1 | 
FileCheck --check-prefix=CHECK-ON %s
 // RUN: %clang -fsyntax-only -fno-relaxed-template-template-args %s 2>&1 | 
FileCheck --check-prefix=CHECK-OFF %s
 
-// CHECK-DEF: "-cc1"{{.*}} "-fno-relaxed-template-template-args"
-// CHECK-ON-NOT: warning: argument '-frelaxed-template-template-args' is 
deprecated [-Wdeprecated]
+// CHECK-DEF-NOT: "-cc1"{{.*}} "-fno-relaxed-template-template-args"
+// CHECK-ON:  warning: argument '-frelaxed-template-template-args' is 
deprecated [-Wdeprecated]
 // CHECK-OFF: warning: argument '-fno-relaxed-template-template-args' is 
deprecated [-Wdeprecated]
diff --git a/clang/test/Driver/rewrite-legacy-objc.m 
b/clang/test/Driver/rewrite-legacy-objc.m
index d45fb8c405c52..413a7a7a61f05 100644
--- a/clang/test/Driver/rewrite-legacy-objc.m
+++ b/clang/test/Driver/rewrite-legacy-objc.m
@@ -3,11 +3,11 @@
 // TEST0: "-cc1"
 // TEST0: "-rewrite-objc"
 // FIXME: CHECK-NOT is broken somehow, it doesn't work here. Check adjacency 
instead.
-// TEST0: "-stack-protector" "1" "-fblocks" 
"-fencode-extended-block-signature" "-fregister-global-dtors-with-atexit" 
"-fgnuc-version=4.2.1"{{.*}} "-fobjc-runtime=macosx-fragile" 
"-fno-objc-infer-related-result-type" "-fobjc-exceptions" "-fexceptions" 
"-fno-relaxed-template-template-args" "-fmax-type-align=16"
+// TEST0: "-stack-protector" "1" "-fblocks" 
"-fencode-extended-block-signature" "-fregister-global-dtors-with-atexit" 
"-fgnuc-version=4.2.1"{{.*}} "-fobjc-runtime=macosx-fragile" 
"-fno-objc-infer-related-result-type" "-fobjc-exceptions" "-fexceptions" 
"-fmax-type-align=16"
 // TEST0: rewrite-legacy-objc.m"
 // RUN: %clang --target=i386-apple-macosx10.9.0 -rewrite-legacy-objc %s -o - 
-### 2>&1 | \
 // RUN:   FileCheck -check-prefix=TEST1 %s
 // RUN: %clang --target=i386-apple-macosx10.6.0 -rewrite-legacy-objc %s -o - 
-### 2>&1 | \
 // RUN:   FileCheck -check-prefix=TEST2 %s
-// TEST1: "-stack-protector" "1" "-fblocks" 
"-fencode-extended-block-signature" "-fregister-global-dtors-with-atexit" 
"-fgnuc-version=4.2.1"{{.*}} "-fobjc-runtime=macosx-fragile" 
"-fobjc-subscripting-legacy-runtime" "-fno-objc-infer-related-result-type" 

[clang] Revert "[clang] Revert default behavior change of P0522R0 implementation (#91811)" (PR #91837)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/91837
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Allow pack expansions when partial ordering against template template parameters (PR #91833)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov closed 
https://github.com/llvm/llvm-project/pull/91833
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Allow pack expansions when partial ordering against template template parameters (PR #91833)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/91833

>From 06b9c19a1d194240be3199d50819090b10d697b6 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Fri, 10 May 2024 23:21:22 -0300
Subject: [PATCH] [clang] Allow pack expansions when partial ordering against
 template template parameters

When partial ordering alias templates against template template parameters,
allow pack expansions when the alias has a fixed-size parameter list.

These expansions were generally disallowed by proposed resolution for CWG1430.

By previously diagnosing these when checking template template parameters, we
would be too strict in trying to prevent any potential invalid use.

This flows against the more general idea that template template parameters are
weakly typed, that we would rather allow an argument that might be possibly
misused, and only diagnose the actual misuses during instantiation.

Since this interaction between P0522R0 and CWG1430 is also a backwards-compat
breaking change, we implement provisional wording to allow these.

Fixes https://github.com/llvm/llvm-project/issues/62529
---
 clang/docs/ReleaseNotes.rst|  2 ++
 clang/include/clang/Sema/Sema.h|  8 +++-
 clang/lib/Sema/SemaTemplate.cpp| 14 ++
 clang/lib/Sema/SemaTemplateDeduction.cpp   |  4 +++-
 ...plate_cxx1z.cpp => temp_arg_template_p0522.cpp} | 12 ++--
 5 files changed, 32 insertions(+), 8 deletions(-)
 rename clang/test/SemaTemplate/{temp_arg_template_cxx1z.cpp => 
temp_arg_template_p0522.cpp} (91%)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 4702b8c10cdbb..28ac54127383a 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -708,6 +708,8 @@ Bug Fixes to C++ Support
   expression.
 - Fix a bug in access control checking due to dealyed checking of friend 
declaration. Fixes (#GH12361).
 - Correctly treat the compound statement of an ``if consteval`` as an 
immediate context. Fixes (#GH91509).
+- When partial ordering alias templates against template template parameters,
+  allow pack expansions when the alias has a fixed-size parameter list. Fixes 
(#GH62529).
 
 Bug Fixes to AST Handling
 ^
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index 4efd3878e861b..869769f95fd7f 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -9216,6 +9216,12 @@ class Sema final : public SemaBase {
   /// receive true if the cause for the error is the associated constraints of
   /// the template not being satisfied by the template arguments.
   ///
+  /// \param PartialOrderingTTP If true, assume these template arguments are
+  /// the injected template arguments for a template template parameter.
+  /// This will relax the requirement that all its possible uses are valid:
+  /// TTP checking is loose, and assumes that invalid uses will be diagnosed
+  /// during instantiation.
+  ///
   /// \returns true if an error occurred, false otherwise.
   bool CheckTemplateArgumentList(
   TemplateDecl *Template, SourceLocation TemplateLoc,
@@ -9223,7 +9229,7 @@ class Sema final : public SemaBase {
   SmallVectorImpl ,
   SmallVectorImpl ,
   bool UpdateArgsWithConversions = true,
-  bool *ConstraintsNotSatisfied = nullptr);
+  bool *ConstraintsNotSatisfied = nullptr, bool PartialOrderingTTP = 
false);
 
   bool CheckTemplateTypeArgument(
   TemplateTypeParmDecl *Param, TemplateArgumentLoc ,
diff --git a/clang/lib/Sema/SemaTemplate.cpp b/clang/lib/Sema/SemaTemplate.cpp
index bae00c6292703..8219d5eed8db7 100644
--- a/clang/lib/Sema/SemaTemplate.cpp
+++ b/clang/lib/Sema/SemaTemplate.cpp
@@ -6556,7 +6556,8 @@ bool Sema::CheckTemplateArgumentList(
 TemplateArgumentListInfo , bool PartialTemplateArgs,
 SmallVectorImpl ,
 SmallVectorImpl ,
-bool UpdateArgsWithConversions, bool *ConstraintsNotSatisfied) {
+bool UpdateArgsWithConversions, bool *ConstraintsNotSatisfied,
+bool PartialOrderingTTP) {
 
   if (ConstraintsNotSatisfied)
 *ConstraintsNotSatisfied = false;
@@ -6627,9 +6628,14 @@ bool Sema::CheckTemplateArgumentList(
   bool PackExpansionIntoNonPack =
   NewArgs[ArgIdx].getArgument().isPackExpansion() &&
   (!(*Param)->isTemplateParameterPack() || 
getExpandedPackSize(*Param));
-  if (PackExpansionIntoNonPack && (isa(Template) ||
-   isa(Template))) {
-// Core issue 1430: we have a pack expansion as an argument to an
+  // CWG1430: Don't diagnose this pack expansion when partial
+  // ordering template template parameters. Some uses of the template could
+  // be valid, and invalid uses will be diagnosed later during
+  // instantiation.
+  if (PackExpansionIntoNonPack && !PartialOrderingTTP &&
+  (isa(Template) ||
+   isa(Template))) {
+// 

[clang] [clang] Allow pack expansions when partial ordering against template template parameters (PR #91833)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/91833

>From f882dca5a53a6da8ad92492f28f9eacffb34a780 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Fri, 10 May 2024 23:21:22 -0300
Subject: [PATCH] [clang] Allow pack expansions when partial ordering against
 template template parameters

When partial ordering alias templates against template template parameters,
allow pack expansions when the alias has a fixed-size parameter list.

These expansions were generally disallowed by proposed resolution for CWG1430.

By previously diagnosing these when checking template template parameters, we
would be too strict in trying to prevent any potential invalid use.

This flows against the more general idea that template template parameters are
weakly typed, that we would rather allow an argument that might be possibly
misused, and only diagnose the actual misuses during instantiation.

Since this interaction between P0522R0 and CWG1430 is also a backwards-compat
breaking change, we implement provisional wording to allow these.

Fixes https://github.com/llvm/llvm-project/issues/62529
---
 clang/docs/ReleaseNotes.rst|  2 ++
 clang/include/clang/Sema/Sema.h|  8 +++-
 clang/lib/Sema/SemaTemplate.cpp| 14 ++
 clang/lib/Sema/SemaTemplateDeduction.cpp   |  4 +++-
 ...plate_cxx1z.cpp => temp_arg_template_p0522.cpp} | 12 ++--
 5 files changed, 32 insertions(+), 8 deletions(-)
 rename clang/test/SemaTemplate/{temp_arg_template_cxx1z.cpp => 
temp_arg_template_p0522.cpp} (91%)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 7c5dcc59c7016..b9380357d04e6 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -707,6 +707,8 @@ Bug Fixes to C++ Support
   initialized, rather than evaluating them as a part of the larger manifestly 
constant evaluated
   expression.
 - Fix a bug in access control checking due to dealyed checking of friend 
declaration. Fixes (#GH12361).
+- When partial ordering alias templates against template template parameters,
+  allow pack expansions when the alias has a fixed-size parameter list. Fixes 
(#GH62529).
 
 Bug Fixes to AST Handling
 ^
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index 4efd3878e861b..869769f95fd7f 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -9216,6 +9216,12 @@ class Sema final : public SemaBase {
   /// receive true if the cause for the error is the associated constraints of
   /// the template not being satisfied by the template arguments.
   ///
+  /// \param PartialOrderingTTP If true, assume these template arguments are
+  /// the injected template arguments for a template template parameter.
+  /// This will relax the requirement that all its possible uses are valid:
+  /// TTP checking is loose, and assumes that invalid uses will be diagnosed
+  /// during instantiation.
+  ///
   /// \returns true if an error occurred, false otherwise.
   bool CheckTemplateArgumentList(
   TemplateDecl *Template, SourceLocation TemplateLoc,
@@ -9223,7 +9229,7 @@ class Sema final : public SemaBase {
   SmallVectorImpl ,
   SmallVectorImpl ,
   bool UpdateArgsWithConversions = true,
-  bool *ConstraintsNotSatisfied = nullptr);
+  bool *ConstraintsNotSatisfied = nullptr, bool PartialOrderingTTP = 
false);
 
   bool CheckTemplateTypeArgument(
   TemplateTypeParmDecl *Param, TemplateArgumentLoc ,
diff --git a/clang/lib/Sema/SemaTemplate.cpp b/clang/lib/Sema/SemaTemplate.cpp
index 480c0103ae335..954363259f6f2 100644
--- a/clang/lib/Sema/SemaTemplate.cpp
+++ b/clang/lib/Sema/SemaTemplate.cpp
@@ -6436,7 +6436,8 @@ bool Sema::CheckTemplateArgumentList(
 TemplateArgumentListInfo , bool PartialTemplateArgs,
 SmallVectorImpl ,
 SmallVectorImpl ,
-bool UpdateArgsWithConversions, bool *ConstraintsNotSatisfied) {
+bool UpdateArgsWithConversions, bool *ConstraintsNotSatisfied,
+bool PartialOrderingTTP) {
 
   if (ConstraintsNotSatisfied)
 *ConstraintsNotSatisfied = false;
@@ -6507,9 +6508,14 @@ bool Sema::CheckTemplateArgumentList(
   bool PackExpansionIntoNonPack =
   NewArgs[ArgIdx].getArgument().isPackExpansion() &&
   (!(*Param)->isTemplateParameterPack() || 
getExpandedPackSize(*Param));
-  if (PackExpansionIntoNonPack && (isa(Template) ||
-   isa(Template))) {
-// Core issue 1430: we have a pack expansion as an argument to an
+  // CWG1430: Don't diagnose this pack expansion when partial
+  // ordering template template parameters. Some uses of the template could
+  // be valid, and invalid uses will be diagnosed later during
+  // instantiation.
+  if (PackExpansionIntoNonPack && !PartialOrderingTTP &&
+  (isa(Template) ||
+   isa(Template))) {
+// CWG1430: 

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> @mizvekov Thank you! With that patch, clang not only doesn't crash on stdexec 
> with `-frelaxed-template-template-args`, but in fact accepts the code.

Thanks! The crash is still there and is pre-existing, but it's a 
'crash-on-invalid' issue, which is lower priority.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

@sam-mccall @bgra8 @ericniebler I believe this MR should fix your issues: 
https://github.com/llvm/llvm-project/pull/91833

Can you double check?

You might consider applying https://github.com/llvm/llvm-project/pull/91837,
since that is stacked on that and will revert the default back to enabling 
relaxed matching by default. Otherwise, you have to compile with 
`-frelaxed-template-template-args`.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Allow pack expansions when partial ordering against template template parameters (PR #91833)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov created 
https://github.com/llvm/llvm-project/pull/91833

When partial ordering alias templates against template template parameters, 
allow pack expansions when the alias has a fixed-size parameter list.

These expansions were generally disallowed by proposed resolution for CWG1430.

By previously diagnosing these when checking template template parameters, we 
would be too strict in trying to prevent any potential invalid use.

This flows against the more general idea that template template parameters are 
weakly typed, that we would rather allow an argument that might be possibly 
misused, and only diagnose the actual misuses during instantiation.

Since this interaction between P0522R0 and CWG1430 is also a backwards-compat 
breaking change, we implement provisional wording to allow these.

Fixes https://github.com/llvm/llvm-project/issues/62529

>From 98fdfb900582cb5f69c56b5c00106dbeff238338 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Fri, 10 May 2024 23:21:22 -0300
Subject: [PATCH] [clang] Allow pack expansions when partial ordering against
 template template parameters

When partial ordering alias templates against template template parameters,
allow pack expansions when the alias has a fixed-size parameter list.

These expansions were generally disallowed by proposed resolution for CWG1430.

By previously diagnosing these when checking template template parameters, we
would be too strict in trying to prevent any potential invalid use.

This flows against the more general idea that template template parameters are
weakly typed, that we would rather allow an argument that might be possibly
misused, and only diagnose the actual misuses during instantiation.

Since this interaction between P0522R0 and CWG1430 is also a backwards-compat
breaking change, we implement provisional wording to allow these.

Fixes https://github.com/llvm/llvm-project/issues/62529
---
 clang/docs/ReleaseNotes.rst  |  1 +
 clang/include/clang/Sema/Sema.h  |  8 +++-
 clang/lib/Sema/SemaTemplate.cpp  | 12 +---
 clang/lib/Sema/SemaTemplateDeduction.cpp |  4 +++-
 ...emplate_cxx1z.cpp => temp_arg_template_p0522.cpp} | 12 ++--
 5 files changed, 30 insertions(+), 7 deletions(-)
 rename clang/test/SemaTemplate/{temp_arg_template_cxx1z.cpp => 
temp_arg_template_p0522.cpp} (91%)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 7c5dcc59c7016..ddc5d5ff16a34 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -707,6 +707,7 @@ Bug Fixes to C++ Support
   initialized, rather than evaluating them as a part of the larger manifestly 
constant evaluated
   expression.
 - Fix a bug in access control checking due to dealyed checking of friend 
declaration. Fixes (#GH12361).
+- When partial ordering alias templates against template template parameters, 
allow pack expansions when the alias has a fixed-size parameter list. Fixes 
(#GH62529).
 
 Bug Fixes to AST Handling
 ^
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index 4efd3878e861b..3fc59488243e5 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -9216,6 +9216,12 @@ class Sema final : public SemaBase {
   /// receive true if the cause for the error is the associated constraints of
   /// the template not being satisfied by the template arguments.
   ///
+  /// \param PartialOrderTTP If true, assume these template arguments are
+  /// the injected template arguments for a template template parameter.
+  /// This will relax the requirement that all it's possible uses are valid.
+  /// TTP checking is loose, and assumes that invalid uses will be diagnosed
+  /// during instantiation.
+  ///
   /// \returns true if an error occurred, false otherwise.
   bool CheckTemplateArgumentList(
   TemplateDecl *Template, SourceLocation TemplateLoc,
@@ -9223,7 +9229,7 @@ class Sema final : public SemaBase {
   SmallVectorImpl ,
   SmallVectorImpl ,
   bool UpdateArgsWithConversions = true,
-  bool *ConstraintsNotSatisfied = nullptr);
+  bool *ConstraintsNotSatisfied = nullptr, bool PartialOrderTTP = false);
 
   bool CheckTemplateTypeArgument(
   TemplateTypeParmDecl *Param, TemplateArgumentLoc ,
diff --git a/clang/lib/Sema/SemaTemplate.cpp b/clang/lib/Sema/SemaTemplate.cpp
index 480c0103ae335..88d0af191e6c8 100644
--- a/clang/lib/Sema/SemaTemplate.cpp
+++ b/clang/lib/Sema/SemaTemplate.cpp
@@ -6436,7 +6436,8 @@ bool Sema::CheckTemplateArgumentList(
 TemplateArgumentListInfo , bool PartialTemplateArgs,
 SmallVectorImpl ,
 SmallVectorImpl ,
-bool UpdateArgsWithConversions, bool *ConstraintsNotSatisfied) {
+bool UpdateArgsWithConversions, bool *ConstraintsNotSatisfied,
+bool PartialOrderTTP) {
 
   if (ConstraintsNotSatisfied)
 *ConstraintsNotSatisfied = false;
@@ -6507,8 

[clang] [Clang] Fix tests broken by #91811 (PR #91822)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.


https://github.com/llvm/llvm-project/pull/91822
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Revert default behavior change of P0522R0 implementation (PR #91811)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov created 
https://github.com/llvm/llvm-project/pull/91811

This partially reverts b86e0992bfa6c58be077d82d824016f590ac5d90.

Just the default is changed back, on the Driver side.
No Frontend changes.
The positive spelling of the flag is undeprecated.

No documentation changes or changelog entries because we plan to revert this 
revert as soon as https://github.com/llvm/llvm-project/issues/62529 is fixed.

>From 553dcf394229a67371ad84e7eb477691346136e9 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Fri, 10 May 2024 18:00:41 -0300
Subject: [PATCH] [clang] Revert default behavior change of P0522R0
 implementation

This partially reverts b86e0992bfa6c58be077d82d824016f590ac5d90
Just the default is changed back, on the Driver side.
No Frontend changes.
The positive spelling of the flag is undeprecated.

No documentation changes or changelog entries because we plan to revert
this revert as soon as https://github.com/llvm/llvm-project/issues/62529
is fixed.
---
 clang/lib/Driver/ToolChains/Clang.cpp  | 14 +++---
 .../Driver/frelaxed-template-template-args.cpp |  6 --
 2 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/clang/lib/Driver/ToolChains/Clang.cpp 
b/clang/lib/Driver/ToolChains/Clang.cpp
index 42feb1650574e..f0cc018b6668f 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -7249,15 +7249,15 @@ void Clang::ConstructJob(Compilation , const 
JobAction ,
   Args.addOptOutFlag(CmdArgs, options::OPT_fassume_unique_vtables,
  options::OPT_fno_assume_unique_vtables);
 
-  // -frelaxed-template-template-args is deprecated.
-  if (Arg *A =
-  Args.getLastArg(options::OPT_frelaxed_template_template_args,
-  options::OPT_fno_relaxed_template_template_args)) {
+  // -fno-relaxed-template-template-args is deprecated.
+  if (Arg *A = Args.getLastArg(options::OPT_frelaxed_template_template_args,
+   
options::OPT_fno_relaxed_template_template_args);
+  A &&
+  A->getOption().matches(options::OPT_fno_relaxed_template_template_args))
 D.Diag(diag::warn_drv_deprecated_arg)
 << A->getAsString(Args) << /*hasReplacement=*/false;
-if 
(A->getOption().matches(options::OPT_fno_relaxed_template_template_args))
-  CmdArgs.push_back("-fno-relaxed-template-template-args");
-  }
+  else
+CmdArgs.push_back("-fno-relaxed-template-template-args");
 
   // -fsized-deallocation is off by default, as it is an ABI-breaking change 
for
   // most platforms.
diff --git a/clang/test/Driver/frelaxed-template-template-args.cpp 
b/clang/test/Driver/frelaxed-template-template-args.cpp
index dd6265ba8375e..136c360276a15 100644
--- a/clang/test/Driver/frelaxed-template-template-args.cpp
+++ b/clang/test/Driver/frelaxed-template-template-args.cpp
@@ -1,5 +1,7 @@
-// RUN: %clang -fsyntax-only -frelaxed-template-template-args %s 2>&1 | 
FileCheck --check-prefix=CHECK-ON %s
+// RUN: %clang -fsyntax-only -### %s 2>&1 | FileCheck --check-prefix=CHECK-DEF 
%s
+// RUN: %clang -fsyntax-only -frelaxed-template-template-args %s 2>&1 | 
FileCheck --check-prefix=CHECK-ON --allow-empty %s
 // RUN: %clang -fsyntax-only -fno-relaxed-template-template-args %s 2>&1 | 
FileCheck --check-prefix=CHECK-OFF %s
 
-// CHECK-ON:  warning: argument '-frelaxed-template-template-args' is 
deprecated [-Wdeprecated]
+// CHECK-DEF: "-cc1"{{.*}} "-fno-relaxed-template-template-args"
+// CHECK-ON-NOT: warning: argument '-frelaxed-template-template-args' is 
deprecated [-Wdeprecated]
 // CHECK-OFF: warning: argument '-fno-relaxed-template-template-args' is 
deprecated [-Wdeprecated]

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

By the way, creduce/cvise won't help much here unless the interestingness test 
accounts for 'works on GCC'.

Otherwise, It'd be trivial to conjure some snippet of code that works before 
P0522, but breaks afterward as intended.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

I am repeating myself here, but the crash happens after a bunch of errors: it's 
not significant, we have evidence this sort of crash is associated with error 
recovery.

This patch implements a standard mandated breaking change, and this 'stdexec' 
is user code.
Without evidence that stdexec does something that should be accepted within 
reason, I don't think that's grounds for revert.

The simplest thing to look for here: Does this project already workaround the 
changes as they were implemented in GCC?
If so, this could be just a matter of changing a preprocessor test.

Here is what I think would be a good evidence that something is wrong with this 
patch: A snippet of code, extracted from 'stdexec', that works on GCC, works on 
official clang, but doesn't after this patch.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> @mizvekov I have a [reduced test 
> case](https://github.com/llvm/llvm-project/files/15261978/repro.zip) for the 
> crash @sam-mccall reported.
> 
> Clang does not crash before but crashes at this revision.

Thanks. I confirm it crashes, but it crashes on clang 18.1.4 as well, if one 
just passes `-frelaxed-template-template-args`.
So most likely this is the same problem, we exposed it by default, but wasn't 
caused by changes in this patch per se.

The program became invalid with the change. The crash is most likely secondary, 
caused by bad error recovery.

The former is the most interesting thing here, we can try to see if it's one of 
the cases where we can solve with additional wording, but the reduction process 
might have mangled this.

Is it possible the original code already had workaround for 
template-template-args with GCC, but the workaround was otherwise GCC specific, 
not using the feature testing macro?


https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> Here's a preprocessed file: 
> [repro.zip](https://github.com/llvm/llvm-project/files/15250584/repro.zip)
> 
> I tried to reduce, and got rid of most of the test code and some of the 
> stdexec code, but there's still a lot left. I hit the end of my timebox on 
> that. Maybe creduce can do better.

That crashes with released clang 18.1.4 as well, same stack trace it seems.
```
Homebrew clang version 18.1.4
Target: arm64-apple-darwin23.4.0
```

Is it possible you may have reduced into a different bug?
It's always helpful in that case to keep testing with known good compiler while 
reducing.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> Thanks, guarding the second specialization with the feature test macro works.
> 
> 
> 
> I will try to reduce the test case tomorrow, if you still need this.
> 
> 

Thanks. If it's not too much work for you, that would be great. Otherwise, I 
think a pretty good guess can be made from reading the source code, but I may 
not be able to try until next week. If that happens and I still don't manage to 
reproduce it, I will ask again.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

Oh I see the code already includes workaround for GCC vs non-GCC. It's possible 
in this case you may replace the workaround with a check for the feature 
testing macro.

But if this is a new ambiguity not covered by any of the cases I am tracking, 
it could still be worthwhile to isolate it. It could be we can cover it with 
new provisional wording.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> Hi, is there a way to make a compile-time check for this feature?

Yes, this is exposed by a standard feature testing macro: 
https://en.cppreference.com/w/cpp/feature_test#cpp_template_template_args

> 
> Looking at
> 

Thanks for reporting this. A few questions:
* Does https://github.com/llvm/llvm-project/pull/90820 help here?
* Does it reproduce with GCC, using `-std=c++17` and above?
* Are you able to reduce this test case?


https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] CTAD: fix the aggregate deduction guide for alias templates. (PR #90894)

[Matheus Izvekov via cfe-commits]

@@ -261,6 +261,13 @@ AG ag = {1};
 // CHECK:   | `-BuiltinType {{.*}} 'int'
 // CHECK:   `-ParmVarDecl {{.*}} 'int'
 
+template 
+using BG = G;
+BG bg(1.0);
+// CHECK-LABEL: Dumping 
+// CHECK:  FunctionTemplateDecl {{.*}} implicit 
+// CHECK: |-CXXDeductionGuideDecl {{.*}} 'auto (int) -> G' aggregate

mizvekov wrote:

But that it is a child of something is trivial, because all declarations are 
children of something, except a translation unit declaration.

But the fact that this line test that this thing is neither the first nor the 
last child of that thing, means that it is quite likely to break with unrelated 
test churn.

https://github.com/llvm/llvm-project/pull/90894
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] CTAD: fix the aggregate deduction guide for alias templates. (PR #90894)

[Matheus Izvekov via cfe-commits]

@@ -261,6 +261,13 @@ AG ag = {1};
 // CHECK:   | `-BuiltinType {{.*}} 'int'
 // CHECK:   `-ParmVarDecl {{.*}} 'int'
 
+template 
+using BG = G;
+BG bg(1.0);
+// CHECK-LABEL: Dumping 
+// CHECK:  FunctionTemplateDecl {{.*}} implicit 
+// CHECK: |-CXXDeductionGuideDecl {{.*}} 'auto (int) -> G' aggregate

mizvekov wrote:

The line is a CHECK, not a CHECK-NEXT, so it could appear any number of lines 
afterwards.
So this means it could be a child of something else that appears after it.

The CHECK matcher matches any number of characters before the first character 
specified,
So it would match something like `| | |-CXXDeductionGuideDecl {{.*}} 'auto 
(int) -> G' aggregate`
Ie, it could be deeply nested into something else.

Instead, I think it just demonstrates:
* It is a child of something.
* It's neither the first nor the last child of that thing.

https://github.com/llvm/llvm-project/pull/90894
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][Sema] Explicit template arguments are not substituted into the exception specification of a function (PR #90760)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.

LGTM

https://github.com/llvm/llvm-project/pull/90760
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] CTAD: fix the aggregate deduction guide for alias templates. (PR #90894)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.

This looks like a straightforward fix to me as well

LGTM

https://github.com/llvm/llvm-project/pull/90894
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] CTAD: fix the aggregate deduction guide for alias templates. (PR #90894)

[Matheus Izvekov via cfe-commits]

@@ -261,6 +261,13 @@ AG ag = {1};
 // CHECK:   | `-BuiltinType {{.*}} 'int'
 // CHECK:   `-ParmVarDecl {{.*}} 'int'
 
+template 
+using BG = G;
+BG bg(1.0);
+// CHECK-LABEL: Dumping 
+// CHECK:  FunctionTemplateDecl {{.*}} implicit 
+// CHECK: |-CXXDeductionGuideDecl {{.*}} 'auto (int) -> G' aggregate

mizvekov wrote:

Nit: Since this isn't matching the structure of the AST anyway, it would 
probably be a good idea to avoid the possibility of unrelated test churn.

```suggestion
// CHECK: CXXDeductionGuideDecl {{.*}} 'auto (int) -> G' aggregate
```

https://github.com/llvm/llvm-project/pull/90894
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] CTAD: fix the aggregate deduction guide for alias templates. (PR #90894)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/90894
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] 62c2959 - [clang] NFC: cxx_status mark P0522R0 as unreleased

[Matheus Izvekov via cfe-commits]

Author: Matheus Izvekov
Date: 2024-05-02T13:55:33-03:00
New Revision: 62c29593be317f6cfaed8ffbcc016bd2c94c35d4

URL: 
https://github.com/llvm/llvm-project/commit/62c29593be317f6cfaed8ffbcc016bd2c94c35d4
DIFF: 
https://github.com/llvm/llvm-project/commit/62c29593be317f6cfaed8ffbcc016bd2c94c35d4.diff

LOG: [clang] NFC: cxx_status mark P0522R0 as unreleased

Addressing post-commit review on #89807

Added: 


Modified: 
clang/www/cxx_status.html

Removed: 




diff  --git a/clang/www/cxx_status.html b/clang/www/cxx_status.html
index b61399a26b1c34..b39cfbd76df253 100755
--- a/clang/www/cxx_status.html
+++ b/clang/www/cxx_status.html
@@ -466,7 +466,7 @@ C++23 implementation status
 
   Clang 19 (Partial)
 The lifetime extension of temporaries bound to member references
-by default member initializers in aggregate initialization was 
+by default member initializers in aggregate initialization was
 not supported now.
 
   
@@ -1141,7 +1141,7 @@ C++17 implementation status
 
   Matching template template parameters to compatible arguments
   https://wg21.link/p0522r0;>P0522R0 (DR)
-  Clang 19 (10)
+  Clang 19 (10)
 
 
   Removing deprecated dynamic exception specifications



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> Sounds like should perhaps note that we are implementing our own resolution, 
> until there's an update to the cwg issue that can be referred to?

That could be. Is there another similar issue we could use as a reference on 
the format for this?

https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

It would not be splitting though, it would be wholesale duplication in that 
case.
This file is the only part of the suite still testing the old non-conformant 
mode, and I fail to see a test case we wouldn't be wanting to test on both.

https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

Note we are not implementing the solution Jason posted on the core mailing 
list, neither on the previous patch, as we have a better solution than current 
GCC on this, nor on this MR, as GCC implements no such workaround and still 
fails this test.

https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

I had some discussion about that with @Endilll on the previous MR regarding 
this core issue: https://github.com/llvm/llvm-project/pull/89807

Since there is no posting at all in core about any possible solutions, I wanted 
to get feedback from them before claiming to be implementing a solution in the 
public documentation.

Also, this needs to test a non-confirming mode until we finally remove the 
corresponding flag, and that is against the rules for the DR suite.

https://github.com/llvm/llvm-project/pull/90820
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Implement provisional wording for CWG2398 regarding packs (PR #90820)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov created 
https://github.com/llvm/llvm-project/pull/90820

This solves some ambuguity introduced in P0522 regarding how template template 
parameters are partially ordered, and should reduce the negative impact of 
enabling `-frelaxed-template-template-args` by default.

A template template parameter containing no packs should be more specialized 
than one that does.

Given the following example:
```C++
template struct A;
template class TT1, class T4> struct A>; #1
template class TT2, class T6> struct A>; #2

template struct B;
template struct A>;
```

Prior to P0522, #2 would be the only viable candidate. After P0522, #1 is also 
viable, and neither is considered more specialized, so this becomes ambiguous.
With this change, #2 is considered more specialized, so that is what is picked 
and we remain compatible with pre-P0522 implementations.

The problem we solve here is that the specialization rules in 
`[temp.arg.template]/4` are defined in terms of a rewrite to function 
templates, but in the case of partial ordering, the rules on how P and A lists 
are matched to each other is swapped regarding how specialization makes sense 
conceptually.

---

Since this changes provisional implementation of CWG2398 which has not been 
released yet, and already contains a changelog entry, we don't provide a 
changelog entry here.

>From b7df9a342724479e129af252f2ff37229f30d41b Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Wed, 1 May 2024 22:29:45 -0300
Subject: [PATCH] [clang] Implement provisional wording for CWG2398 regarding
 packs

This solves some ambuguity introduced in P0522 regarding how
template template parameters are partially ordered, and should reduce
the negative impact of enabling `-frelaxed-template-template-args`
by default.

A template template parameter containing no packs should be more
specialized than one that does.

Given the following example:
```C++
template struct A;
template class TT1, class T4> struct A>; #1
template class TT2, class T6> struct A>; #2

template struct B;
template struct A>;
```

Prior to P0522, #2 would be the only viable candidate.
After P0522, #1 is also viable, and neither is considered more specialized,
so this becomes ambiguous.
With this change, #2 is considered more specialized, so that is what is picked
and we remain compatible with pre-P0522 implementations.

The problem we solve here is that the specialization rules in
`[temp.arg.template]/4` are defined in terms of a rewrite to
function templates, but in the case of partial ordering, the
rules on how P and A lists are matched to each other
is swapped regarding how specialization makes sense conceptually.

---

Since this changes provisional implementation of CWG2398 which has
not been released yet, and already contains a changelog entry,
we don't provide a changelog entry here.
---
 clang/include/clang/Sema/Sema.h  |  5 +-
 clang/lib/Sema/SemaTemplate.cpp  | 10 +--
 clang/lib/Sema/SemaTemplateDeduction.cpp | 79 ++--
 clang/test/SemaTemplate/cwg2398.cpp  |  6 --
 4 files changed, 68 insertions(+), 32 deletions(-)

diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index a80ac6dbc76137..2bea8732f61fcd 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -9238,7 +9238,7 @@ class Sema final : public SemaBase {
CheckTemplateArgumentKind CTAK);
   bool CheckTemplateTemplateArgument(TemplateTemplateParmDecl *Param,
  TemplateParameterList *Params,
- TemplateArgumentLoc );
+ TemplateArgumentLoc , bool IsDeduced);
 
   void NoteTemplateLocation(const NamedDecl ,
 std::optional ParamRange = {});
@@ -9708,7 +9708,8 @@ class Sema final : public SemaBase {
 sema::TemplateDeductionInfo );
 
   bool isTemplateTemplateParameterAtLeastAsSpecializedAs(
-  TemplateParameterList *PParam, TemplateDecl *AArg, SourceLocation Loc);
+  TemplateParameterList *PParam, TemplateDecl *AArg, SourceLocation Loc,
+  bool IsDeduced);
 
   void MarkUsedTemplateParameters(const Expr *E, bool OnlyDeduced,
   unsigned Depth, llvm::SmallBitVector );
diff --git a/clang/lib/Sema/SemaTemplate.cpp b/clang/lib/Sema/SemaTemplate.cpp
index 989f3995ca5991..2c921d4365db0a 100644
--- a/clang/lib/Sema/SemaTemplate.cpp
+++ b/clang/lib/Sema/SemaTemplate.cpp
@@ -6384,7 +6384,8 @@ bool Sema::CheckTemplateArgument(
 
   case TemplateArgument::Template:
   case TemplateArgument::TemplateExpansion:
-if (CheckTemplateTemplateArgument(TempParm, Params, Arg))
+if (CheckTemplateTemplateArgument(TempParm, Params, Arg,
+  /*IsDeduced=*/CTAK != CTAK_Specified))
   return true;
 
 SugaredConverted.push_back(Arg.getArgument());
@@ -8296,7 

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov closed 
https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][Sema] fix a bug on constraint check with template friend function (PR #90646)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

I agree with @sdkrystian, even though the test crashes for maybe yet another 
reason, it demonstrates you can friend a function from a different template 
context, so comparing the depths from different branches is not helpful.

https://github.com/llvm/llvm-project/pull/90646
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/89807

>From 1756044e71d756f7102f962d0298627ede27871c Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Tue, 9 Apr 2024 01:14:28 -0300
Subject: [PATCH] [clang] Enable C++17 relaxed template template argument
 matching by default

In order to implement this as a DR and avoid breaking reasonable code
that worked before P0522, this patch implements a provisional resolution
for CWG2398: When deducing template template parameters against each other,
and the argument side names a template specialization, instead of just
deducing A, we instead deduce a synthesized template template parameter based
on A, but with it's parameters using the template specialization's arguments
as defaults.

The driver flag is deprecated with a warning.

With this patch, we finally mark C++17 support in clang as complete.

Fixes https://github.com/llvm/llvm-project/issues/36505
---
 clang/docs/ReleaseNotes.rst   |  19 +++
 .../clang/Basic/DiagnosticDriverKinds.td  |   2 +-
 clang/include/clang/Basic/LangOptions.def |   2 +-
 clang/include/clang/Driver/Options.td |   8 +-
 clang/lib/Driver/SanitizerArgs.cpp|   9 +-
 clang/lib/Driver/ToolChains/Clang.cpp |  16 +-
 clang/lib/Sema/SemaTemplate.cpp   |   3 -
 clang/lib/Sema/SemaTemplateDeduction.cpp  | 107 +-
 .../temp/temp.arg/temp.arg.template/p3-2a.cpp |   2 +-
 clang/test/CodeGenCXX/mangle-concept.cpp  |   4 +-
 .../frelaxed-template-template-args.cpp   |   5 +
 clang/test/Lexer/cxx-features.cpp |   6 +-
 clang/test/SemaTemplate/cwg2398.cpp   | 139 ++
 clang/test/SemaTemplate/default-arguments.cpp |   7 +-
 .../instantiate-template-template-parm.cpp|  17 +--
 clang/test/SemaTemplate/nested-template.cpp   |   8 +-
 clang/test/SemaTemplate/temp_arg_template.cpp |   6 +-
 .../SemaTemplate/temp_arg_template_cxx1z.cpp  |   2 +-
 clang/www/cxx_status.html |  18 +--
 19 files changed, 317 insertions(+), 63 deletions(-)
 create mode 100644 clang/test/Driver/frelaxed-template-template-args.cpp
 create mode 100644 clang/test/SemaTemplate/cwg2398.cpp

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index a1390d6536b28c..bd7f96246fd407 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -48,6 +48,11 @@ C++ Specific Potentially Breaking Changes
 - Clang now diagnoses function/variable templates that shadow their own 
template parameters, e.g. ``template void T();``.
   This error can be disabled via `-Wno-strict-primary-template-shadow` for 
compatibility with previous versions of clang.
 
+- The behavior controlled by the `-frelaxed-template-template-args` flag is now
+  on by default, and the flag is deprecated. Until the flag is finally removed,
+  it's negative spelling can be used to obtain compatibility with previous
+  versions of clang.
+
 ABI Changes in This Version
 ---
 - Fixed Microsoft name mangling of implicitly defined variables used for thread
@@ -88,6 +93,17 @@ sections with improvements to Clang's support for those 
languages.
 
 C++ Language Changes
 
+- C++17 support is now completed, with the enablement of the
+  relaxed temlate template argument matching rules introduced in P0522,
+  which was retroactively applied as a defect report.
+  While the implementation already existed since Clang 4, it was turned off by
+  default, and was controlled with the `-frelaxed-template-template-args` flag.
+  In this release, we implement provisional wording for a core defect on
+  P0522 (CWG2398), which avoids the most serious compatibility issues caused
+  by it, allowing us to enable it by default in this release.
+  The flag is now deprecated, and will be removed in the next release, but can
+  still be used to turn it off and regain compatibility with previous versions
+  (#GH36505).
 - Implemented ``_BitInt`` literal suffixes ``__wb`` or ``__WB`` as a Clang 
extension with ``unsigned`` modifiers also allowed. (#GH85223).
 
 C++17 Feature Support
@@ -164,6 +180,9 @@ Resolutions to C++ Defect Reports
 - Clang now diagnoses declarative nested-name-specifiers with 
pack-index-specifiers.
   (`CWG2858: Declarative nested-name-specifiers and pack-index-specifiers 
`_).
 
+- P0522 implementation is enabled by default in all language versions, and
+  provisional wording for CWG2398 is implemented.
+
 C Language Changes
 --
 
diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td 
b/clang/include/clang/Basic/DiagnosticDriverKinds.td
index ed3fd9b1c4a55b..9781fcaa4ff5e9 100644
--- a/clang/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td
@@ -435,7 +435,7 @@ def warn_drv_diagnostics_misexpect_requires_pgo : Warning<
 def warn_drv_clang_unsupported : 

[clang] Fix a crash introduced by 3d5e9ab by adding a nullptr check. (PR #90301)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.


https://github.com/llvm/llvm-project/pull/90301
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> @mizvekov We have a bunch of related issues, could you look at them 
> https://github.com/llvm/llvm-project/issues?q=is%3Aissue+is%3Aopen+%22-frelaxed-template-template-args%22
>  ?

Removed a bunch of duplicates. 4 issues remaining:

1) #36505 Is the issue we are fixing in this patch.
2) #65843 Is unrelated.
3) #63281 Is not a bug
4) #62529 The bug is accepting it, although perhaps we could seek clarification 
if we should exclude template template parameters from 
[CWG1430](https://cplusplus.github.io/CWG/issues/1430.html)


https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/89807

>From 4ee58efa0f154b531dcc674b6f4fe084182aa803 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Tue, 9 Apr 2024 01:14:28 -0300
Subject: [PATCH] [clang] Enable C++17 relaxed template template argument
 matching by default

In order to implement this as a DR and avoid breaking reasonable code
that worked before P0522, this patch implements a provisional resolution
for CWG2398: When deducing template template parameters against each other,
and the argument side names a template specialization, instead of just
deducing A, we instead deduce a synthesized template template parameter based
on A, but with it's parameters using the template specialization's arguments
as defaults.

The driver flag is deprecated with a warning.

With this patch, we finally mark C++17 support in clang as complete.

Fixes https://github.com/llvm/llvm-project/issues/36505
---
 clang/docs/ReleaseNotes.rst   |  19 +++
 .../clang/Basic/DiagnosticDriverKinds.td  |   2 +-
 clang/include/clang/Basic/LangOptions.def |   2 +-
 clang/include/clang/Driver/Options.td |   8 +-
 clang/lib/Driver/SanitizerArgs.cpp|   9 +-
 clang/lib/Driver/ToolChains/Clang.cpp |  16 ++-
 clang/lib/Sema/SemaTemplate.cpp   |   3 -
 clang/lib/Sema/SemaTemplateDeduction.cpp  | 107 +++-
 .../temp/temp.arg/temp.arg.template/p3-2a.cpp |   2 +-
 clang/test/CodeGenCXX/mangle-concept.cpp  |   4 +-
 .../frelaxed-template-template-args.cpp   |   5 +
 clang/test/Lexer/cxx-features.cpp |   6 +-
 clang/test/SemaTemplate/cwg2398.cpp   | 115 ++
 clang/test/SemaTemplate/default-arguments.cpp |   7 +-
 .../instantiate-template-template-parm.cpp|  17 ++-
 clang/test/SemaTemplate/nested-template.cpp   |   8 +-
 clang/test/SemaTemplate/temp_arg_template.cpp |   6 +-
 .../SemaTemplate/temp_arg_template_cxx1z.cpp  |   2 +-
 clang/www/cxx_status.html |  18 ++-
 19 files changed, 293 insertions(+), 63 deletions(-)
 create mode 100644 clang/test/Driver/frelaxed-template-template-args.cpp
 create mode 100644 clang/test/SemaTemplate/cwg2398.cpp

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index a1390d6536b28c..bd7f96246fd407 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -48,6 +48,11 @@ C++ Specific Potentially Breaking Changes
 - Clang now diagnoses function/variable templates that shadow their own 
template parameters, e.g. ``template void T();``.
   This error can be disabled via `-Wno-strict-primary-template-shadow` for 
compatibility with previous versions of clang.
 
+- The behavior controlled by the `-frelaxed-template-template-args` flag is now
+  on by default, and the flag is deprecated. Until the flag is finally removed,
+  it's negative spelling can be used to obtain compatibility with previous
+  versions of clang.
+
 ABI Changes in This Version
 ---
 - Fixed Microsoft name mangling of implicitly defined variables used for thread
@@ -88,6 +93,17 @@ sections with improvements to Clang's support for those 
languages.
 
 C++ Language Changes
 
+- C++17 support is now completed, with the enablement of the
+  relaxed temlate template argument matching rules introduced in P0522,
+  which was retroactively applied as a defect report.
+  While the implementation already existed since Clang 4, it was turned off by
+  default, and was controlled with the `-frelaxed-template-template-args` flag.
+  In this release, we implement provisional wording for a core defect on
+  P0522 (CWG2398), which avoids the most serious compatibility issues caused
+  by it, allowing us to enable it by default in this release.
+  The flag is now deprecated, and will be removed in the next release, but can
+  still be used to turn it off and regain compatibility with previous versions
+  (#GH36505).
 - Implemented ``_BitInt`` literal suffixes ``__wb`` or ``__WB`` as a Clang 
extension with ``unsigned`` modifiers also allowed. (#GH85223).
 
 C++17 Feature Support
@@ -164,6 +180,9 @@ Resolutions to C++ Defect Reports
 - Clang now diagnoses declarative nested-name-specifiers with 
pack-index-specifiers.
   (`CWG2858: Declarative nested-name-specifiers and pack-index-specifiers 
`_).
 
+- P0522 implementation is enabled by default in all language versions, and
+  provisional wording for CWG2398 is implemented.
+
 C Language Changes
 --
 
diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td 
b/clang/include/clang/Basic/DiagnosticDriverKinds.td
index ed3fd9b1c4a55b..9781fcaa4ff5e9 100644
--- a/clang/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td
@@ -435,7 +435,7 @@ def warn_drv_diagnostics_misexpect_requires_pgo : Warning<
 def warn_drv_clang_unsupported 

[clang] Fix a crash introduced by 3d5e9ab by adding a nullptr check. (PR #90301)

[Matheus Izvekov via cfe-commits]

@@ -54,7 +54,7 @@ class UncountedCallArgsChecker
   bool shouldVisitImplicitCode() const { return false; }
 
   bool TraverseDecl(Decl *D) {
-if (isa(D) && isRefType(safeGetName(D)))
+if (D && isa(D) && isRefType(safeGetName(D)))
   return true;
 return RecursiveASTVisitor::TraverseDecl(D);
   }

mizvekov wrote:

Shouldn't this be overriding `TraverseTemplateInstantiations(ClassTemplateDecl 
*D)` instead?

 `RecursiveASTVisitor::TraverseDecl`  will ensure you won't be 
visiting null nodes.

https://github.com/llvm/llvm-project/pull/90301
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -507,10 +507,62 @@ static TemplateDeductionResult 
DeduceNonTypeTemplateArgument(
   S, TemplateParams, NTTP, DeducedTemplateArgument(New), T, Info, Deduced);
 }
 
+static NamedDecl *DeduceTemplateArguments(Sema , NamedDecl *A,
+  TemplateArgument Default) {
+  switch (A->getKind()) {
+  case Decl::TemplateTypeParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())

mizvekov wrote:

I see.

The code I posted was an attempt to construct such an example, where a pack 
doesn't need to be defaulted as-written for the proposed changes in the rules 
to have an effect. Except that example hits the wall at something else first, 
and I haven't fully investigated that. I haven't excluded the possibility that 
such an example could be constructed yet.

So when I proposed that provisional wording, I didn't exclude packs from the 
equation. I don't see a reason to, as without excluding them, the rules stay 
simpler and more general and should still work.

If we want to exclude them for implementation efficiency reasons, we could, as 
presumably it could still work as-if.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -507,10 +507,62 @@ static TemplateDeductionResult 
DeduceNonTypeTemplateArgument(
   S, TemplateParams, NTTP, DeducedTemplateArgument(New), T, Info, Deduced);
 }
 
+static NamedDecl *DeduceTemplateArguments(Sema , NamedDecl *A,
+  TemplateArgument Default) {
+  switch (A->getKind()) {
+  case Decl::TemplateTypeParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())
+  return A;
+auto *R = TemplateTypeParmDecl::Create(
+S.Context, A->getDeclContext(), SourceLocation(), SourceLocation(),
+T->getDepth(), T->getIndex(), T->getIdentifier(),
+T->wasDeclaredWithTypename(), /*ParameterPack=*/false,
+T->hasTypeConstraint());
+R->setDefaultArgument(
+S.Context.getTrivialTypeSourceInfo(Default.getAsType()));
+if (R->hasTypeConstraint()) {
+  auto *C = R->getTypeConstraint();
+  R->setTypeConstraint(C->getConceptReference(),
+   C->getImmediatelyDeclaredConstraint());

mizvekov wrote:

It's the same tests I linked in the other thread: 
https://github.com/llvm/llvm-project/blob/15f02723d49be9a828fbf072966a225babd60457/clang/test/CXX/temp/temp.decls/temp.fct/temp.func.order/p6.cpp#L48

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -0,0 +1,115 @@
+// RUN: %clang_cc1 %s -fsyntax-only -std=c++23 
-verify=expected,new
+// RUN: %clang_cc1 %s -fsyntax-only -std=c++23 
-fno-relaxed-template-template-args -verify=expected,old

mizvekov wrote:

Thanks.

Since the core issue is not resolved by WG21, I don't want to claim to be 
resolving it here with this patch.

And while we don't remove the non-conforming flag, we should still test it, and 
it's just much more economical and error proof that we reuse the same tests.


https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov dismissed 
https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/89807

>From 43f813d0a1a87b6cad9b859237489778f4f2945f Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Tue, 9 Apr 2024 01:14:28 -0300
Subject: [PATCH] [clang] Enable C++17 relaxed template template argument
 matching by default

In order to implement this as a DR and avoid breaking reasonable code
that worked before P0522, this patch implements a provisional resolution
for CWG2398: When deducing template template parameters against each other,
and the argument side names a template specialization, instead of just
deducing A, we instead deduce a synthesized template template parameter based
on A, but with it's parameters using the template specialization's arguments
as defaults.

The driver flag is deprecated with a warning.

With this patch, we finally mark C++17 support in clang as complete.

Closes #55894
---
 clang/docs/ReleaseNotes.rst   |  18 +++
 .../clang/Basic/DiagnosticDriverKinds.td  |   2 +-
 clang/include/clang/Basic/LangOptions.def |   2 +-
 clang/include/clang/Driver/Options.td |   8 +-
 clang/lib/Driver/SanitizerArgs.cpp|   9 +-
 clang/lib/Driver/ToolChains/Clang.cpp |  16 ++-
 clang/lib/Sema/SemaTemplate.cpp   |   3 -
 clang/lib/Sema/SemaTemplateDeduction.cpp  | 107 +++-
 .../temp/temp.arg/temp.arg.template/p3-2a.cpp |   2 +-
 clang/test/CodeGenCXX/mangle-concept.cpp  |   4 +-
 .../frelaxed-template-template-args.cpp   |   5 +
 clang/test/Lexer/cxx-features.cpp |   6 +-
 clang/test/SemaTemplate/cwg2398.cpp   | 115 ++
 clang/test/SemaTemplate/default-arguments.cpp |   7 +-
 .../instantiate-template-template-parm.cpp|  17 ++-
 clang/test/SemaTemplate/nested-template.cpp   |   8 +-
 clang/test/SemaTemplate/temp_arg_template.cpp |   6 +-
 .../SemaTemplate/temp_arg_template_cxx1z.cpp  |   2 +-
 clang/www/cxx_status.html |  18 ++-
 19 files changed, 292 insertions(+), 63 deletions(-)
 create mode 100644 clang/test/Driver/frelaxed-template-template-args.cpp
 create mode 100644 clang/test/SemaTemplate/cwg2398.cpp

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index f5e5d3a2e6ea36..f525bdd73010cb 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -48,6 +48,11 @@ C++ Specific Potentially Breaking Changes
 - Clang now diagnoses function/variable templates that shadow their own 
template parameters, e.g. ``template void T();``.
   This error can be disabled via `-Wno-strict-primary-template-shadow` for 
compatibility with previous versions of clang.
 
+- The behavior controlled by the `-frelaxed-template-template-args` flag is now
+  on by default, and the flag is deprecated. Until the flag is finally removed,
+  it's negative spelling can be used to obtain compatibility with previous
+  versions of clang.
+
 ABI Changes in This Version
 ---
 - Fixed Microsoft name mangling of implicitly defined variables used for thread
@@ -88,6 +93,16 @@ sections with improvements to Clang's support for those 
languages.
 
 C++ Language Changes
 
+- C++17 support is now completed, with the enablement of the
+  relaxed temlate template argument matching rules introduced in P0522,
+  which was retroactively applied as a defect report.
+  While the implementation already existed since Clang 4, it was turned off by
+  default, and was controlled with the `-frelaxed-template-template-args` flag.
+  In this release, we implement provisional wording for a core defect on
+  P0522 (CWG2398), which avoids the most serious compatibility issues caused
+  by it, allowing us to enable it by default in this release.
+  The flag is now deprecated, and will be removed in the next release, but can
+  still be used to turn it off and regain compatibility with previous versions.
 - Implemented ``_BitInt`` literal suffixes ``__wb`` or ``__WB`` as a Clang 
extension with ``unsigned`` modifiers also allowed. (#GH85223).
 
 C++20 Feature Support
@@ -152,6 +167,9 @@ Resolutions to C++ Defect Reports
 - Clang now diagnoses declarative nested-name-specifiers with 
pack-index-specifiers.
   (`CWG2858: Declarative nested-name-specifiers and pack-index-specifiers 
`_).
 
+- P0522 implementation is enabled by default in all language versions, and
+  provisional wording for CWG2398 is implemented.
+
 C Language Changes
 --
 
diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td 
b/clang/include/clang/Basic/DiagnosticDriverKinds.td
index ed3fd9b1c4a55b..9781fcaa4ff5e9 100644
--- a/clang/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td
@@ -435,7 +435,7 @@ def warn_drv_diagnostics_misexpect_requires_pgo : Warning<
 def warn_drv_clang_unsupported : Warning<
   "the clang compiler does not support 

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -0,0 +1,115 @@
+// RUN: %clang_cc1 %s -fsyntax-only -std=c++23 
-verify=expected,new

mizvekov wrote:

While it's true this is a DR, I just don't think for this particular case it's 
worth the cost of testing every single mode, as there are no particularly 
interesting interactions there.
Moot point as I will be moving it away anyway.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -0,0 +1,115 @@
+// RUN: %clang_cc1 %s -fsyntax-only -std=c++23 
-verify=expected,new
+// RUN: %clang_cc1 %s -fsyntax-only -std=c++23 
-fno-relaxed-template-template-args -verify=expected,old

mizvekov wrote:

Okay, I think I misunderstood then what `CXX/drs` was about, I should probably 
move it elsewhere.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Enable C++17 relaxed template template argument matching by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov updated 
https://github.com/llvm/llvm-project/pull/89807

>From 3f6e50edc7b4d4bf4781c71bd29f48224b62822d Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Tue, 9 Apr 2024 01:14:28 -0300
Subject: [PATCH] [clang] Enable C++17 relaxed template template argument
 matching by default

In order to implement this as a DR and avoid breaking reasonable code
that worked before P0522, this patch implements a provisional resolution
for CWG2398: When deducing template template parameters against each other,
and the argument side names a template specialization, instead of just
deducing A, we instead deduce a synthesized template template parameter based
on A, but with it's parameters using the template specialization's arguments
as defaults.

The driver flag is deprecated with a warning.

With this patch, we finally mark C++17 support in clang as complete.

Closes #55894
---
 clang/docs/ReleaseNotes.rst   |  18 +++
 .../clang/Basic/DiagnosticDriverKinds.td  |   2 +-
 clang/include/clang/Basic/LangOptions.def |   2 +-
 clang/include/clang/Driver/Options.td |   8 +-
 clang/lib/Driver/SanitizerArgs.cpp|   9 +-
 clang/lib/Driver/ToolChains/Clang.cpp |  16 ++-
 clang/lib/Sema/SemaTemplate.cpp   |   3 -
 clang/lib/Sema/SemaTemplateDeduction.cpp  | 107 +++-
 clang/test/CXX/drs/cwg2398.cpp| 115 ++
 .../temp/temp.arg/temp.arg.template/p3-2a.cpp |   2 +-
 clang/test/CodeGenCXX/mangle-concept.cpp  |   4 +-
 .../frelaxed-template-template-args.cpp   |   5 +
 clang/test/Lexer/cxx-features.cpp |   6 +-
 clang/test/SemaTemplate/default-arguments.cpp |   7 +-
 .../instantiate-template-template-parm.cpp|  17 ++-
 clang/test/SemaTemplate/nested-template.cpp   |   8 +-
 clang/test/SemaTemplate/temp_arg_template.cpp |   6 +-
 .../SemaTemplate/temp_arg_template_cxx1z.cpp  |   2 +-
 clang/www/cxx_status.html |  18 ++-
 19 files changed, 292 insertions(+), 63 deletions(-)
 create mode 100644 clang/test/CXX/drs/cwg2398.cpp
 create mode 100644 clang/test/Driver/frelaxed-template-template-args.cpp

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index f5e5d3a2e6ea36..f525bdd73010cb 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -48,6 +48,11 @@ C++ Specific Potentially Breaking Changes
 - Clang now diagnoses function/variable templates that shadow their own 
template parameters, e.g. ``template void T();``.
   This error can be disabled via `-Wno-strict-primary-template-shadow` for 
compatibility with previous versions of clang.
 
+- The behavior controlled by the `-frelaxed-template-template-args` flag is now
+  on by default, and the flag is deprecated. Until the flag is finally removed,
+  it's negative spelling can be used to obtain compatibility with previous
+  versions of clang.
+
 ABI Changes in This Version
 ---
 - Fixed Microsoft name mangling of implicitly defined variables used for thread
@@ -88,6 +93,16 @@ sections with improvements to Clang's support for those 
languages.
 
 C++ Language Changes
 
+- C++17 support is now completed, with the enablement of the
+  relaxed temlate template argument matching rules introduced in P0522,
+  which was retroactively applied as a defect report.
+  While the implementation already existed since Clang 4, it was turned off by
+  default, and was controlled with the `-frelaxed-template-template-args` flag.
+  In this release, we implement provisional wording for a core defect on
+  P0522 (CWG2398), which avoids the most serious compatibility issues caused
+  by it, allowing us to enable it by default in this release.
+  The flag is now deprecated, and will be removed in the next release, but can
+  still be used to turn it off and regain compatibility with previous versions.
 - Implemented ``_BitInt`` literal suffixes ``__wb`` or ``__WB`` as a Clang 
extension with ``unsigned`` modifiers also allowed. (#GH85223).
 
 C++20 Feature Support
@@ -152,6 +167,9 @@ Resolutions to C++ Defect Reports
 - Clang now diagnoses declarative nested-name-specifiers with 
pack-index-specifiers.
   (`CWG2858: Declarative nested-name-specifiers and pack-index-specifiers 
`_).
 
+- P0522 implementation is enabled by default in all language versions, and
+  provisional wording for CWG2398 is implemented.
+
 C Language Changes
 --
 
diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td 
b/clang/include/clang/Basic/DiagnosticDriverKinds.td
index ed3fd9b1c4a55b..9781fcaa4ff5e9 100644
--- a/clang/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td
@@ -435,7 +435,7 @@ def warn_drv_diagnostics_misexpect_requires_pgo : Warning<
 def warn_drv_clang_unsupported : Warning<
   "the clang compiler does not support '%0'">;
 

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

So Jason pointed out that GCC's provisional wording for CWG2398 picks a dubious 
candidate for this example:
```C++
template struct match2;

template class t1,typename T>
struct match2, typename t1::type > { typedef int type; }; // #5
 

template class t2,typename T0,typename T1>
struct match2, typename t2::type > { typedef int type; }; // 
#6 

template  struct Q { typedef int type; };
match2, int> m;
```

They pick #6, where with this PR we stay with ambiguous.
According to [this GCC 
bug](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114841), he suggests changing 
GCC to adopt the wording proposed here.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -507,10 +507,62 @@ static TemplateDeductionResult 
DeduceNonTypeTemplateArgument(
   S, TemplateParams, NTTP, DeducedTemplateArgument(New), T, Info, Deduced);
 }
 
+static NamedDecl *DeduceTemplateArguments(Sema , NamedDecl *A,
+  TemplateArgument Default) {
+  switch (A->getKind()) {
+  case Decl::TemplateTypeParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())

mizvekov wrote:

The test `foo` in 
`clang/test/CXX/temp/temp.decls/temp.fct/temp.func.order/p6.cpp` exercises 
this, but it doesn't involve any default arguments, so any effects are 
incidental / harmless.

I have actually just come up with a relevant test case that shows something 
interesting that I wasn't aware before, this could possibly be new info for the 
core issue.

https://godbolt.org/z/rnbzhnqG7

This shows a change of behavior introduced by P05522, only instead of 
introducing ambiguity, it causes different candidates to be picked. Also 
affects GCC.

---

Regarding the FIXME:

So the (Type | NonType | Template) template parameter declarations store the 
default argument for themselves. They have a `setDefaultArgument` method for 
setting that. While for template template parameter declarations the 
`setDefaultArgument` takes a `TemplateArgument`, and that means it can use any 
kind of TemplateArgument as a default argument, including packs.

The same is not true for (Type | NonType) template parameter declarations, 
their `setDefaultArgument` only takes a (Type / Expression), which means a pack 
can't be passed.

This is done presumably to save space in those declarations, while a Type and 
an Expression are a single pointer, a `TemplateArgument` is considerably 
larger, and you would need at least a pointer and an integer to represent a 
pack.

To 'fix' this FIXME, we would have to change that, and allow those declarations 
to be larger in at least some circumstances.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -519,13 +571,45 @@ DeduceTemplateArguments(Sema , TemplateParameterList 
*TemplateParams,
 return TemplateDeductionResult::Success;
   }
 
-  if (TemplateTemplateParmDecl *TempParam
-= dyn_cast(ParamDecl)) {
+  if (auto *TempParam = dyn_cast(ParamDecl)) {
 // If we're not deducing at this depth, there's nothing to deduce.
 if (TempParam->getDepth() != Info.getDeducedDepth())
   return TemplateDeductionResult::Success;
 
-DeducedTemplateArgument 
NewDeduced(S.Context.getCanonicalTemplateName(Arg));
+auto NewDeduced = DeducedTemplateArgument(Arg);
+// Provisional resolution for CWG2398: If Arg is also a template template
+// param, and it names a template specialization, then we deduce a
+// synthesized template template parameter based on A, but using the TS's
+// arguments as defaults.
+if (auto *TempArg = dyn_cast_or_null(
+Arg.getAsTemplateDecl())) {
+  assert(Arg.getKind() == TemplateName::Template);
+  assert(!TempArg->isExpandedParameterPack());
+
+  TemplateParameterList *As = TempArg->getTemplateParameters();
+  if (DefaultArguments.size() != 0) {
+assert(DefaultArguments.size() <= As->size());
+SmallVector Params(As->size());
+for (unsigned I = 0; I < DefaultArguments.size(); ++I)
+  Params[I] =
+  DeduceTemplateArguments(S, As->getParam(I), DefaultArguments[I]);
+for (unsigned I = DefaultArguments.size(); I < As->size(); ++I)
+  Params[I] = As->getParam(I);
+// FIXME: We could unique these, and also the parameters, but we don't
+// expect programs to contain a large enough amount of these deductions
+// for that to be worthwhile.

mizvekov wrote:

Yeah, doesn't affect identity. The results from deduction are assumed to be 
sugared, this was one of the changes I made to the deduction machinery back in 
2021 or so.

Where identity is concerned, the types should be canonicalized, and we have a 
special step for canonicalizing template template parameters.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -1133,8 +1133,8 @@ C++17 implementation status
 
 
   Matching template template parameters to compatible arguments
-  https://wg21.link/p0522r0;>P0522R0
-  Partial (10)
+  https://wg21.link/p0522r0;>P0522R0 (DR)
+  Clang 4 (10)

mizvekov wrote:

This entry was one of the review requests from @zygoloid in the original phab 
PR: https://reviews.llvm.org/D109496?id=384873#3031693

I think saying we implemented in Clang 4 is more accurate, we even explain in 
the note that it was behind a switch. But still users have known about it and 
have been using it for a long time.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -507,10 +507,62 @@ static TemplateDeductionResult 
DeduceNonTypeTemplateArgument(
   S, TemplateParams, NTTP, DeducedTemplateArgument(New), T, Info, Deduced);
 }
 
+static NamedDecl *DeduceTemplateArguments(Sema , NamedDecl *A,
+  TemplateArgument Default) {
+  switch (A->getKind()) {
+  case Decl::TemplateTypeParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())
+  return A;
+auto *R = TemplateTypeParmDecl::Create(
+S.Context, A->getDeclContext(), SourceLocation(), SourceLocation(),
+T->getDepth(), T->getIndex(), T->getIdentifier(),
+T->wasDeclaredWithTypename(), /*ParameterPack=*/false,
+T->hasTypeConstraint());
+R->setDefaultArgument(
+S.Context.getTrivialTypeSourceInfo(Default.getAsType()));
+if (R->hasTypeConstraint()) {
+  auto *C = R->getTypeConstraint();
+  R->setTypeConstraint(C->getConceptReference(),
+   C->getImmediatelyDeclaredConstraint());
+}
+return R;
+  }
+  case Decl::NonTypeTemplateParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())
+  return A;
+auto *R = NonTypeTemplateParmDecl::Create(
+S.Context, A->getDeclContext(), SourceLocation(), SourceLocation(),
+T->getDepth(), T->getIndex(), T->getIdentifier(), T->getType(),
+/*ParameterPack=*/false, T->getTypeSourceInfo());
+R->setDefaultArgument(Default.getAsExpr());
+if (auto *PTC = T->getPlaceholderTypeConstraint())
+  R->setPlaceholderTypeConstraint(PTC);
+return R;
+  }
+  case Decl::TemplateTemplateParm: {

mizvekov wrote:

A template template parameter can have a template template parameter as a 
parameter. These parameters can have default arguments too.

I thought I saw this test case covered in the clang test suite, but I was 
mistaken.

Here is one example test that will hit this case:
```C++
template  struct X {};

template class T2 = X> struct A {};

template struct B {};

template class TT1, class T4> struct B>; // #1
template class> class TT2, class T5, 
template  class T6> struct B> {}; // #2

template struct B>;
```

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -8343,58 +8343,52 @@ bool 
Sema::CheckTemplateTemplateArgument(TemplateTemplateParmDecl *Param,
   // C++1z [temp.arg.template]p3: (DR 150)
   //   A template-argument matches a template template-parameter P when P
   //   is at least as specialized as the template-argument A.
-  // FIXME: We should enable RelaxedTemplateTemplateArgs by default as it is a
-  //  defect report resolution from C++17 and shouldn't be introduced by
-  //  concepts.
-  if (getLangOpts().RelaxedTemplateTemplateArgs) {

mizvekov wrote:

Yes, I think if there are problems in existing C++03 code, we should know about 
it, I can't see how that same problem wouldn't necessarily apply to newer 
standards.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -507,10 +507,62 @@ static TemplateDeductionResult 
DeduceNonTypeTemplateArgument(
   S, TemplateParams, NTTP, DeducedTemplateArgument(New), T, Info, Deduced);
 }
 
+static NamedDecl *DeduceTemplateArguments(Sema , NamedDecl *A,
+  TemplateArgument Default) {
+  switch (A->getKind()) {
+  case Decl::TemplateTypeParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())
+  return A;
+auto *R = TemplateTypeParmDecl::Create(
+S.Context, A->getDeclContext(), SourceLocation(), SourceLocation(),
+T->getDepth(), T->getIndex(), T->getIdentifier(),
+T->wasDeclaredWithTypename(), /*ParameterPack=*/false,
+T->hasTypeConstraint());
+R->setDefaultArgument(
+S.Context.getTrivialTypeSourceInfo(Default.getAsType()));
+if (R->hasTypeConstraint()) {
+  auto *C = R->getTypeConstraint();
+  R->setTypeConstraint(C->getConceptReference(),
+   C->getImmediatelyDeclaredConstraint());

mizvekov wrote:

Note we don't do anything interesting about concepts in this patch, we are just 
doing a member-wise copy of the declarations, except for the DefaultArgument. 
Examples where concepts are being copied along here already exist in the clang 
test suite.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -507,10 +507,62 @@ static TemplateDeductionResult 
DeduceNonTypeTemplateArgument(
   S, TemplateParams, NTTP, DeducedTemplateArgument(New), T, Info, Deduced);
 }
 
+static NamedDecl *DeduceTemplateArguments(Sema , NamedDecl *A,
+  TemplateArgument Default) {
+  switch (A->getKind()) {
+  case Decl::TemplateTypeParm: {
+auto *T = cast(A);
+// FIXME: DefaultArgument can't represent a pack.
+if (T->isParameterPack())

mizvekov wrote:

It's possible to write code that hits this condition, in fact such code is 
already in the clang test suite.

Currently the Decls for Type and Expr parameters don't store a whole 
TemplateArgument for the default argument, in order to save a few bytes, they 
store only a TSI and Expr* respectively, which doesn't cover the Pack case.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -8343,58 +8343,52 @@ bool 
Sema::CheckTemplateTemplateArgument(TemplateTemplateParmDecl *Param,
   // C++1z [temp.arg.template]p3: (DR 150)
   //   A template-argument matches a template template-parameter P when P
   //   is at least as specialized as the template-argument A.
-  // FIXME: We should enable RelaxedTemplateTemplateArgs by default as it is a
-  //  defect report resolution from C++17 and shouldn't be introduced by
-  //  concepts.
-  if (getLangOpts().RelaxedTemplateTemplateArgs) {

mizvekov wrote:

It's a removal of the flag's behavior, with the addition of provisional wording 
changes, in order to moot the original reason the flag existed.

I see where the confusion comes from. The original patch did not remove the 
flag completely, just changed the default, but still deprecated it. After some 
offline discussion with @zygoloid, we decided to remove the flag as well, but I 
forgot to update the title.

@zygoloid thoughts on that?

How about for starters we just change the default, but still deprecate the flag 
both ways anyway?
If the transition is a concern, I think it would be a problem that we don't 
start to warn the users of the flags as soon as possible. Otherwise, I think 
this will be specially problematic if there are current users of the negative 
spelling `-fno-relaxed-template-template-args`.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -8343,58 +8343,52 @@ bool 
Sema::CheckTemplateTemplateArgument(TemplateTemplateParmDecl *Param,
   // C++1z [temp.arg.template]p3: (DR 150)
   //   A template-argument matches a template template-parameter P when P
   //   is at least as specialized as the template-argument A.
-  // FIXME: We should enable RelaxedTemplateTemplateArgs by default as it is a
-  //  defect report resolution from C++17 and shouldn't be introduced by
-  //  concepts.
-  if (getLangOpts().RelaxedTemplateTemplateArgs) {

mizvekov wrote:

The code following this removal contains the exact same code, but outside the 
if-block and de-indented.
It's a shame the GitHub diff viewer is not very good at showing this, phab was 
much better.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

@@ -8343,58 +8343,52 @@ bool 
Sema::CheckTemplateTemplateArgument(TemplateTemplateParmDecl *Param,
   // C++1z [temp.arg.template]p3: (DR 150)
   //   A template-argument matches a template template-parameter P when P
   //   is at least as specialized as the template-argument A.
-  // FIXME: We should enable RelaxedTemplateTemplateArgs by default as it is a
-  //  defect report resolution from C++17 and shouldn't be introduced by
-  //  concepts.
-  if (getLangOpts().RelaxedTemplateTemplateArgs) {

mizvekov wrote:

Only the flag is deprecated, we want the new 'on' behavior to stay.

https://github.com/llvm/llvm-project/pull/89807
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] deprecate frelaxed-template-template-args, make it on by default (PR #89807)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov created 
https://github.com/llvm/llvm-project/pull/89807

This patch will finally allow us to mark C++17 support in clang as complete.

This is a continuation of the review process from an [old PR in 
phab](https://reviews.llvm.org/D109496).

Recap: The original patch from phab initially contained no workarounds / 
provisional resolutions for core defects,
Though testing by @yxsamliu [here](https://reviews.llvm.org/D109496#3101839) 
showed problems with some important libraries used in GPU applications, and we 
ended up reverting it.

In order to implement this as a DR and avoid breaking reasonable code that 
worked before P0522, this patch implements a provisional resolution for 
CWG2398: When deducing template template parameters against each other, and the 
argument side names a template specialization, instead of just deducing A, we 
instead deduce a synthesized template template parameter based on A, but with 
it's parameters using the template specialization's arguments as defaults.

This does not fix all possible regressions introduced by P0522, but it does 
seem to match in effect an undocumented workaround implemented by GCC. Though 
it's unconfirmed how closely we match, as I have not received official word 
regarding this yet.
Fixing other regressions will require more extensive changes, some of them 
would require possibly controversial wording changes, and so is left for future 
work.

The driver flag is deprecated with a warning, and it will not have any effect.

@yxsamliu Can you confirm this version avoids any breakages in your setup?

CC: @yuanfang-chen @MaskRay @chandlerc @jicama @lichray @AaronBallman 

>From 873043f6ea08b8828a5e061cf67c70e39b66b8b8 Mon Sep 17 00:00:00 2001
From: Matheus Izvekov 
Date: Tue, 9 Apr 2024 01:14:28 -0300
Subject: [PATCH] [clang] deprecate frelaxed-template-template-args, make it on
 by default

In order to implement this as a DR and avoid breaking reasonable code
that worked before P0522, this patch implements a provisional resolution
for CWG2398: When deducing template template parameters against each other,
and the argument side names a template specialization, instead of just
deducing A, we instead deduce a synthesized template template parameter based
on A, but with it's parameters using the template specialization's arguments
as defaults.

The driver flag is deprecated with a warning, and it will not have any effect.
---
 .../clang/Basic/DiagnosticDriverKinds.td  |  2 +-
 clang/include/clang/Basic/LangOptions.def |  1 -
 clang/include/clang/Driver/Options.td |  7 +-
 clang/lib/Driver/SanitizerArgs.cpp|  9 +-
 clang/lib/Driver/ToolChains/Clang.cpp | 11 +--
 clang/lib/Frontend/InitPreprocessor.cpp   |  4 +-
 clang/lib/Sema/SemaTemplate.cpp   | 88 -
 clang/lib/Sema/SemaTemplateDeduction.cpp  | 99 +--
 .../temp/temp.arg/temp.arg.template/p3-2a.cpp |  2 +-
 clang/test/CodeGenCXX/mangle-concept.cpp  |  4 +-
 .../frelaxed-template-template-args.cpp   |  5 +
 clang/test/Lexer/cxx-features.cpp |  5 +-
 clang/test/SemaTemplate/default-arguments.cpp |  7 +-
 .../instantiate-template-template-parm.cpp| 17 ++--
 clang/test/SemaTemplate/nested-template.cpp   |  8 +-
 clang/test/SemaTemplate/temp_arg_template.cpp |  6 +-
 .../SemaTemplate/temp_arg_template_cxx1z.cpp  |  2 +-
 clang/www/cxx_status.html | 17 ++--
 18 files changed, 183 insertions(+), 111 deletions(-)
 create mode 100644 clang/test/Driver/frelaxed-template-template-args.cpp

diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td 
b/clang/include/clang/Basic/DiagnosticDriverKinds.td
index ed3fd9b1c4a55b..9781fcaa4ff5e9 100644
--- a/clang/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td
@@ -435,7 +435,7 @@ def warn_drv_diagnostics_misexpect_requires_pgo : Warning<
 def warn_drv_clang_unsupported : Warning<
   "the clang compiler does not support '%0'">;
 def warn_drv_deprecated_arg : Warning<
-  "argument '%0' is deprecated, use '%1' instead">, InGroup;
+  "argument '%0' is deprecated%select{|, use '%2' instead}1">, 
InGroup;
 def warn_drv_deprecated_custom : Warning<
   "argument '%0' is deprecated, %1">, InGroup;
 def warn_drv_assuming_mfloat_abi_is : Warning<
diff --git a/clang/include/clang/Basic/LangOptions.def 
b/clang/include/clang/Basic/LangOptions.def
index 8ef6700ecdc78e..2dfe70f29d7c85 100644
--- a/clang/include/clang/Basic/LangOptions.def
+++ b/clang/include/clang/Basic/LangOptions.def
@@ -158,7 +158,6 @@ LANGOPT(GNUAsm, 1, 1, "GNU-style inline 
assembly")
 LANGOPT(Coroutines, 1, 0, "C++20 coroutines")
 LANGOPT(CoroAlignedAllocation, 1, 0, "prefer Aligned Allocation according to 
P2014 Option 2")
 LANGOPT(DllExportInlines  , 1, 1, "dllexported classes dllexport inline 
methods")
-LANGOPT(RelaxedTemplateTemplateArgs, 1, 0, "C++17 relaxed matching of 

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

@@ -13456,6 +13458,15 @@ void Sema::AddInitializerToDecl(Decl *RealDecl, Expr 
*Init, bool DirectInit) {
 return;
   }
 
+  if (VDecl->isInvalidDecl()) {
+CorrectDelayedTyposInExpr(Init, VDecl);
+ExprResult Recovery =
+CreateRecoveryExpr(Init->getBeginLoc(), Init->getEndLoc(), {Init});
+if (Expr *E = Recovery.get())

mizvekov wrote:

This is ok, I see that action result's get method will return a nullptr in 
unset/failed case anyway, and we don't need to make the distinction here.

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.


https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov deleted 
https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

@@ -13456,6 +13458,15 @@ void Sema::AddInitializerToDecl(Decl *RealDecl, Expr 
*Init, bool DirectInit) {
 return;
   }
 
+  if (VDecl->isInvalidDecl()) {
+CorrectDelayedTyposInExpr(Init, VDecl);
+ExprResult Recovery =
+CreateRecoveryExpr(Init->getBeginLoc(), Init->getEndLoc(), {Init});
+if (Expr *E = Recovery.get())

mizvekov wrote:

Oops. I think this is still UB, please double check:
```suggestion
if (ExprResult Recovery =
 CreateRecoveryExpr(Init->getBeginLoc(), Init->getEndLoc(), {Init}))
 VDecl->setInit(Recovery.get());
```

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.

LGTM, Thanks!

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

@@ -13435,16 +13435,18 @@ void Sema::checkNonTrivialCUnion(QualType QT, 
SourceLocation Loc,
 void Sema::AddInitializerToDecl(Decl *RealDecl, Expr *Init, bool DirectInit) {
   // If there is no declaration, there was an error parsing it.  Just ignore
   // the initializer.
-  if (!RealDecl || RealDecl->isInvalidDecl()) {
+  if (!RealDecl) {
 CorrectDelayedTyposInExpr(Init, dyn_cast_or_null(RealDecl));
 return;
   }
 
   if (CXXMethodDecl *Method = dyn_cast(RealDecl)) {

mizvekov wrote:

Minor nit: We can use separate init-statement since we adopted C++17; You call.
```suggestion
  if (auto *Method = dyn_cast(RealDecl); 
!Method->isInvalidDecl()) {
```

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Fix high memory consumption during pack deduction (PR #88637)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

@term-est It would be helpful if you could post the surrounding code context 
where this crashed, This is pointed at the top of the stacktrace:
```
0.  
/home/est/Working-Directory/DS/wetend-emulator-new/WetendEmulator.hpp:85:128: 
current parser token ')'
```

You said, and I confirm from the stack trace, it looks like this happened in a 
clangd invocation.
Does it also happen during compilation with clang?

Are you able to share prints and dumps from values surrounding the crash site?
Some examples:
```
NumNamedPacks
Info.getNumExplicitArgs()
Info.getDeducedDepth()
S.CurrentInstantiationScope != nullptr
PartialPackDepthIndex
Packs[...].Index
```



https://github.com/llvm/llvm-project/pull/88637
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

@@ -3647,15 +3647,28 @@ class IsTypeDeclaredInsideVisitor
 };
 } // namespace
 
-/// This function checks if the function has 'auto' return type that contains
+/// This function checks if the given function has a return type that contains
 /// a reference (in any way) to a declaration inside the same function.
-bool ASTNodeImporter::hasAutoReturnTypeDeclaredInside(FunctionDecl *D) {
+bool ASTNodeImporter::hasReturnTypeDeclaredInside(FunctionDecl *D) {
   QualType FromTy = D->getType();
   const auto *FromFPT = FromTy->getAs();
   assert(FromFPT && "Must be called on FunctionProtoType");
 
+  auto IsCXX11LambdaWithouTrailingReturn = [&]() {
+if (!Importer.FromContext.getLangOpts().CPlusPlus11)

mizvekov wrote:

I think CPlusPlus11 is set for C++11 *and later*, but we want to return early 
here for C++14 onwards, as in that case lambda with deduced return type will 
have AutoType.
```suggestion
if (Importer.FromContext.getLangOpts().CPlusPlus14)
```

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.

LGTM, Thanks!

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

@@ -3647,15 +3647,19 @@ class IsTypeDeclaredInsideVisitor
 };
 } // namespace
 
-/// This function checks if the function has 'auto' return type that contains
+/// This function checks if the given function has a return type that contains
 /// a reference (in any way) to a declaration inside the same function.
-bool ASTNodeImporter::hasAutoReturnTypeDeclaredInside(FunctionDecl *D) {
+bool ASTNodeImporter::hasReturnTypeDeclaredInside(FunctionDecl *D) {
   QualType FromTy = D->getType();
   const auto *FromFPT = FromTy->getAs();
   assert(FromFPT && "Must be called on FunctionProtoType");
 
+  bool IsLambdaDefinition = false;
+  if (const auto *MD = dyn_cast(D))
+IsLambdaDefinition = cast(MD->getDeclContext())->isLambda();

mizvekov wrote:

Also, this problem of the return type not being wrapped on an AutoType only 
happens on lambdas in *C++11*.

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

@@ -3647,15 +3647,19 @@ class IsTypeDeclaredInsideVisitor
 };
 } // namespace
 
-/// This function checks if the function has 'auto' return type that contains
+/// This function checks if the given function has a return type that contains
 /// a reference (in any way) to a declaration inside the same function.
-bool ASTNodeImporter::hasAutoReturnTypeDeclaredInside(FunctionDecl *D) {
+bool ASTNodeImporter::hasReturnTypeDeclaredInside(FunctionDecl *D) {
   QualType FromTy = D->getType();
   const auto *FromFPT = FromTy->getAs();
   assert(FromFPT && "Must be called on FunctionProtoType");
 
+  bool IsLambdaDefinition = false;
+  if (const auto *MD = dyn_cast(D))
+IsLambdaDefinition = cast(MD->getDeclContext())->isLambda();

mizvekov wrote:

No, you don't need to hunt for return statements.

You can just check on the CXXMethodDecl that it's FunctionProtoType has a 
trailing return (`hasTrailingReturn` method).
I think that should cover it for all cases.

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

@@ -3647,15 +3647,19 @@ class IsTypeDeclaredInsideVisitor
 };
 } // namespace
 
-/// This function checks if the function has 'auto' return type that contains
+/// This function checks if the given function has a return type that contains
 /// a reference (in any way) to a declaration inside the same function.
-bool ASTNodeImporter::hasAutoReturnTypeDeclaredInside(FunctionDecl *D) {
+bool ASTNodeImporter::hasReturnTypeDeclaredInside(FunctionDecl *D) {
   QualType FromTy = D->getType();
   const auto *FromFPT = FromTy->getAs();
   assert(FromFPT && "Must be called on FunctionProtoType");
 
+  bool IsLambdaDefinition = false;
+  if (const auto *MD = dyn_cast(D))
+IsLambdaDefinition = cast(MD->getDeclContext())->isLambda();

mizvekov wrote:

Thanks, this works, but if we are going this way, why not also check the lambda 
was defined without a return type?

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

> I tested this idea, might be like this:

I meant something else. This patch makes it so the delayed function type 
importation happens unconditionally.
I was aware that change would cause breakages.

What I meant is that, we have a special sugar type node that marks deduction, 
this AutoType, and we apply it everywhere else, except this lambda case you 
exposed. I was going to suggest we do the same for these lambdas. That would 
make things more consistent.



https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

@@ -3649,19 +3649,15 @@ class IsTypeDeclaredInsideVisitor
 
 /// This function checks if the function has 'auto' return type that contains
 /// a reference (in any way) to a declaration inside the same function.

mizvekov wrote:

```suggestion
/// This function checks if the given function has a return type that contains
/// a reference (in any way) to a declaration inside the same function.
```

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov commented:

I think it would be a good idea to double check this for performance 
regressions, since this case will recurse into the function every time with 
this patch.
Though I don't know if there is a better way to test it than to merge it and 
wait for it to be picked up by google.

Another idea to avoid any potential regressions, is to mark the return type as 
deduced anyway, with the distinction that 'auto' was not written.

https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [ASTImporter] Fix infinite recurse on return type declared inside body (#68775) (PR #89096)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/89096
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] Improve stack usage to increase recursive initialization depth (PR #88546)

[Matheus Izvekov via cfe-commits]

@@ -1070,57 +1077,24 @@ class Sema;
 };
 
   private:
-SmallVector Candidates;
-llvm::SmallPtrSet Functions;
-
-// Allocator for ConversionSequenceLists. We store the first few of these
-// inline to avoid allocation for small sets.
-llvm::BumpPtrAllocator SlabAllocator;
+ASTContext 
+SmallVector Candidates;
+llvm::SmallPtrSet Functions;

mizvekov wrote:

I see, it's fine either way  

https://github.com/llvm/llvm-project/pull/88546
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] Improve stack usage to increase recursive initialization depth (PR #88546)

[Matheus Izvekov via cfe-commits]

@@ -1070,57 +1077,24 @@ class Sema;
 };
 
   private:
-SmallVector Candidates;
-llvm::SmallPtrSet Functions;
-
-// Allocator for ConversionSequenceLists. We store the first few of these
-// inline to avoid allocation for small sets.
-llvm::BumpPtrAllocator SlabAllocator;
+ASTContext 
+SmallVector Candidates;
+llvm::SmallPtrSet Functions;

mizvekov wrote:

pre-existing small nit, for consistency:
```suggestion
SmallVector Candidates;
SmallPtrSet Functions;
```

https://github.com/llvm/llvm-project/pull/88546
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Improve error recovery for id-expressions referencing invalid decls (PR #81662)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/81662
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Improve error recovery for id-expressions referencing invalid decls (PR #81662)

[Matheus Izvekov via cfe-commits]

@@ -3453,6 +3453,10 @@ ExprResult Sema::BuildDeclarationNameExpr(const 
CXXScopeSpec ,
NeedsADL, R.isOverloadedResult(),
R.begin(), R.end());
 
+  if (ULE && R.isSingleResult() && R.getFoundDecl()->isInvalidDecl()) {

mizvekov wrote:

```suggestion
  if (R.isSingleResult() && R.getFoundDecl()->isInvalidDecl()) {
```
AST Node `Create*` functions normally can't fail.

https://github.com/llvm/llvm-project/pull/81662
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Improve error recovery for id-expressions referencing invalid decls (PR #81662)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/81662
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Improve error recovery for id-expressions referencing invalid decls (PR #81662)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.

Minor nit, otherwise LGTM.

https://github.com/llvm/llvm-project/pull/81662
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [C++20] [Modules] Introduce -fexperimental-modules-reduced-bmi (PR #85050)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov approved this pull request.

I share the objections that it may be too soon to introduce a driver flag for 
this. Only a frontend flag is ok for now, but it's non blocking on my side 
because it doesn't look like it will be particularly hard to deprecate it later.

https://github.com/llvm/llvm-project/pull/85050
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [C++20] [Modules] Introduce -fexperimental-modules-reduced-bmi (PR #85050)

[Matheus Izvekov via cfe-commits]

@@ -4045,6 +4045,24 @@ static bool RenderModulesOptions(Compilation , const 
Driver ,
   // module fragment.
   CmdArgs.push_back("-fskip-odr-check-in-gmf");
 
+  if (Args.hasArg(options::OPT_modules_reduced_bmi) &&
+  (Input.getType() == driver::types::TY_CXXModule ||
+   Input.getType() == driver::types::TY_PP_CXXModule)) {
+CmdArgs.push_back("-fexperimental-modules-reduced-bmi");
+
+if (Args.hasArg(options::OPT_fmodule_output_EQ))
+  Args.AddLastArg(CmdArgs, options::OPT_fmodule_output_EQ);
+else
+  CmdArgs.push_back(Args.MakeArgString(
+  "-fmodule-output=" +
+  getCXX20NamedModuleOutputPath(Args, Input.getBaseInput(;

mizvekov wrote:

Would it be reasonable to push two separate arguments, instead of concatenating 
them?

https://github.com/llvm/llvm-project/pull/85050
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [C++20] [Modules] Introduce -fexperimental-modules-reduced-bmi (PR #85050)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/85050
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [C++20] [Modules] Introduce -fexperimental-modules-reduced-bmi (PR #85050)

[Matheus Izvekov via cfe-commits]

@@ -1061,6 +1070,16 @@ CodeGenAction::CreateASTConsumer(CompilerInstance , 
StringRef InFile) {
 CI.getPreprocessor().addPPCallbacks(std::move(Callbacks));
   }
 
+  if (CI.getFrontendOpts().GenReducedBMI &&
+  !CI.getFrontendOpts().ModuleOutputPath.empty()) {
+std::vector> Consumers{2};

mizvekov wrote:

Minor nit: https://llvm.org/docs/CodingStandards.html#id28
```suggestion
std::vector> Consumers(2);
```

https://github.com/llvm/llvm-project/pull/85050
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov requested changes to this pull request.

Thanks for the improvement!

Can you also add a test case?

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

@@ -13456,6 +13455,14 @@ void Sema::AddInitializerToDecl(Decl *RealDecl, Expr 
*Init, bool DirectInit) {
 return;
   }
 
+  if (VDecl->isInvalidDecl()) {
+CorrectDelayedTyposInExpr(Init, VDecl);
+VDecl->setInit(
+CreateRecoveryExpr(Init->getBeginLoc(), Init->getEndLoc(), {Init})
+.get());

mizvekov wrote:

`CreateRecoveryExpr` returns an ExprResult, which means it not only returns an 
expression on success, but it can also return an error, and using `get()` in 
that case is UB, you have to check it first.

It can return an error in case we are in an SFINAE context, and also when using 
the frontend flag `-fno-recovery-ast`.

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

@@ -13435,8 +13435,7 @@ void Sema::checkNonTrivialCUnion(QualType QT, 
SourceLocation Loc,
 void Sema::AddInitializerToDecl(Decl *RealDecl, Expr *Init, bool DirectInit) {
   // If there is no declaration, there was an error parsing it.  Just ignore
   // the initializer.
-  if (!RealDecl || RealDecl->isInvalidDecl()) {
-CorrectDelayedTyposInExpr(Init, dyn_cast_or_null(RealDecl));
+  if (!RealDecl) {
 return;
   }

mizvekov wrote:

There is an odd change in behavior here, is that intentional?
If there is no declaration, or the declaration is not a VarDecl, then we don't 
try to correct typos in `Init` anymore.

https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Sema] Preserve the initializer of invalid VarDecls (PR #88645)

[Matheus Izvekov via cfe-commits]

https://github.com/mizvekov edited 
https://github.com/llvm/llvm-project/pull/88645
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang][Index] Use canonical function parameter types in USRs (PR #68222)

[Matheus Izvekov via cfe-commits]

mizvekov wrote:

@sdkrystian ping, do you still intend to continue this? Just adding your 
example as a test case would be fine.

https://github.com/llvm/llvm-project/pull/68222
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [clang] Fix high memory consumption during pack deduction (PR #88637)

[Matheus Izvekov via cfe-commits]

@@ -831,7 +831,7 @@ class PackDeductionScope {
 if (IsPartiallyExpanded)
   PackElements += NumPartialPackArgs;
 else if (IsExpanded)
-  PackElements += *FixedNumExpansions;
+  PackElements += FixedNumExpansions.value_or(1);

mizvekov wrote:

The assert is not necessary, unless it added explanation.
The `*` operator on the optional here should already assert in that case.

I would advise you run this test case on an llvm build with runtime checks 
enabled. The compiler is going off the rails before this point, so maybe there 
is an earlier assert that could narrow it down.

Otherwise, the change lacks explanation, about what is the problem, and what 
the fix does.

https://github.com/llvm/llvm-project/pull/88637
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits