erichkeane added a comment. In D80743#2074121 <https://reviews.llvm.org/D80743#2074121>, @erichkeane wrote:
> @rsmith I think this implements what you've suggested? I'm struggling a > little with the template instantiations here, I'm not terribly sure I > understand them as well as I'd hope. > > This seems to 'work' for a small subset of works, but it doesn't properly > register the typedef to the LocalInstantiationScope, so the normal template > instantiation (like here > https://github.com/llvm/llvm-project/blob/master/clang/test/SemaCXX/cxx1z-class-template-argument-deduction.cpp#L156) > ends up hitting the 'findInstantiationOf' assert here: > https://github.com/llvm/llvm-project/blob/master/clang/lib/Sema/SemaTemplateInstantiate.cpp#L3564 > > I can see the parameter itself with the correct type being registered here: > https://github.com/llvm/llvm-project/blob/master/clang/lib/Sema/SemaTemplate.cpp#L2249 > > But I don't have a good idea on how I'm supposed to do this with the typedef. > Since it is a materialized typedef, it doesn't seem like we have the 'old' > decl to register, nor access to the LocalInstantiationScope. > > Can you make a suggestion? Am I missing something simple, or did I just not > do what you suggested? If thats the case, can you rephrase your thought? > > Thanks! > -Erich Hmm... so I tried to assign the typedef to the CXXRecordDecl inside the template instead (giving it an immediate parent), but it this error: llvm-project/clang/test/SemaCXX/cxx1z-class-template-argument-deduction.cpp:149:24: error: type 'int' cannot be used prior to '::' because it has no members using U = typename T::type; ^ llvm-project/clang/test/SemaCXX/cxx1z-class-template-argument-deduction.cpp:150:10: note: in instantiation of template class 'look_into_current_instantiation::C<int>' requested here C(T, U); ^ llvm-project/clang/test/SemaCXX/cxx1z-class-template-argument-deduction.cpp:156:5: note: while substituting deduced template arguments into function template '<deduction guide for C>' [with T = int] C c = {1, 2}; Is that the intended error? Or am I causing an instantiation we shouldn't be (and breaking the reason for this TreeTransform-er)? This is the reproducer, including a comment that makes me think this is the intended behavior: // We should have a substitution failure in the immediate context of // deduction when using the C(T, U) constructor (probably; core wording // unclear). template<typename T> struct C { using U = typename T::type; C(T, U); }; struct R { R(int); typedef R type; }; C(...) -> C<R>; C c = {1, 2}; CHANGES SINCE LAST ACTION https://reviews.llvm.org/D80743/new/ https://reviews.llvm.org/D80743 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
