[clang] d214bec - [clang] Improve deduction of reference typed NTTP (#110393)
via cfe-commits
cfe-commits at lists.llvm.org
Tue Oct 1 16:57:49 PDT 2024
Author: Matheus Izvekov
Date: 2024-10-01T20:57:47-03:00
New Revision: d214bec5161e9e6fd9e4c024f6068597822b1d4e
URL: https://github.com/llvm/llvm-project/commit/d214bec5161e9e6fd9e4c024f6068597822b1d4e
DIFF: https://github.com/llvm/llvm-project/commit/d214bec5161e9e6fd9e4c024f6068597822b1d4e.diff
LOG: [clang] Improve deduction of reference typed NTTP (#110393)
This improves the existing workaround for a core issue introduced in
CWG1770.
When performing template argument deduction for an NTTP which the
parameter side is a reference, instead of dropping the references for
both sides, just make the argument be same reference typed as the
parameter, in case the argument is not already a reference type.
Fixes #73460
Added:
Modified:
clang/docs/ReleaseNotes.rst
clang/lib/Sema/SemaTemplateDeduction.cpp
clang/test/SemaTemplate/temp_arg_nontype_cxx1z.cpp
Removed:
################################################################################
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index c351ccff0b6ca0..6e7a5fb76b602b 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -465,6 +465,7 @@ Bug Fixes to C++ Support
- Fixed an assertion failure in debug mode, and potential crashes in release mode, when
diagnosing a failed cast caused indirectly by a failed implicit conversion to the type of the constructor parameter.
- Fixed an assertion failure by adjusting integral to boolean vector conversions (#GH108326)
+- Fixed an issue deducing non-type template arguments of reference type. (#GH73460)
- Mangle friend function templates with a constraint that depends on a template parameter from an enclosing template as members of the enclosing class. (#GH110247)
- Fixed an issue in constraint evaluation, where type constraints on the lambda expression
containing outer unexpanded parameters were not correctly expanded. (#GH101754)
diff --git a/clang/lib/Sema/SemaTemplateDeduction.cpp b/clang/lib/Sema/SemaTemplateDeduction.cpp
index 0a973a3a0ce036..03ff8145e3b4ac 100644
--- a/clang/lib/Sema/SemaTemplateDeduction.cpp
+++ b/clang/lib/Sema/SemaTemplateDeduction.cpp
@@ -442,18 +442,18 @@ DeduceNonTypeTemplateArgument(Sema &S, TemplateParameterList *TemplateParams,
// FIXME: It's not clear how deduction of a parameter of reference
// type from an argument (of non-reference type) should be performed.
- // For now, we just remove reference types from both sides and let
- // the final check for matching types sort out the mess.
- ValueType = ValueType.getNonReferenceType();
- if (ParamType->isReferenceType())
- ParamType = ParamType.getNonReferenceType();
- else
- // Top-level cv-qualifiers are irrelevant for a non-reference type.
- ValueType = ValueType.getUnqualifiedType();
+ // For now, we just make the argument have same reference type as the
+ // parameter.
+ if (ParamType->isReferenceType() && !ValueType->isReferenceType()) {
+ if (ParamType->isRValueReferenceType())
+ ValueType = S.Context.getRValueReferenceType(ValueType);
+ else
+ ValueType = S.Context.getLValueReferenceType(ValueType);
+ }
return DeduceTemplateArgumentsByTypeMatch(
S, TemplateParams, ParamType, ValueType, Info, Deduced,
- TDF_SkipNonDependent,
+ TDF_SkipNonDependent | TDF_IgnoreQualifiers,
PartialOrdering ? PartialOrderingKind::NonCall
: PartialOrderingKind::None,
/*ArrayBound=*/NewDeduced.wasDeducedFromArrayBound(), HasDeducedAnyParam);
diff --git a/clang/test/SemaTemplate/temp_arg_nontype_cxx1z.cpp b/clang/test/SemaTemplate/temp_arg_nontype_cxx1z.cpp
index ae06055bf52651..8d9356bb3201cd 100644
--- a/clang/test/SemaTemplate/temp_arg_nontype_cxx1z.cpp
+++ b/clang/test/SemaTemplate/temp_arg_nontype_cxx1z.cpp
@@ -613,3 +613,11 @@ struct {
template<typename T>
using a = s<f(T::x)>;
}
+
+namespace GH73460 {
+ template <class T, T, T> struct A;
+ template <class T, T n> struct A<T, n, n> {};
+
+ int j;
+ template struct A<int&, j, j>;
+} // namespace GH73460
More information about the cfe-commits
mailing list