lime added a comment.
Well, Something happened after rebasing this patch on D126907
<https://reviews.llvm.org/D126907>. `s41` below was rejected as the constrain
generated from `template <C>` was no longer considered to subsume the constrain
generated from `template <typename T> requires C` in the template template
argument, which is not the case <https://godbolt.org/z/Gq7h58hGb> in both GCC
and MSVC. However, GCC and MSVC also accept the redeclaration for `S`, which
might be ill-formed because of this rule
<https://eel.is/c++draft/temp.decls#temp.over.link-7>. If this kind of
redeclaration happens on `X`, GCC and MSVC will reject it. Rebasing this patch
on D126907 <https://reviews.llvm.org/D126907> will also not make the both
redeclaration valid.
Personally, I decided to make `s41` valid for Clang, a clue might be making the
`QualType`s the same in the parameters of two generated constrains.
template <typename T> concept C = T::f();
template <typename T> concept C1 = T::f();
template <C> struct X {};
template <typename T> requires C<T> struct X; // ill-formed for sure
template <C1> struct Y {};
template <template <typename T> requires C<T> class> struct S {};
template <template <C> class> struct S; // GCC and MSVC accept this
S<X> sx; // my target
S<Y> sy; // ill-formed for sure
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D134128/new/
https://reviews.llvm.org/D134128
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
