sammccall added a comment. Sorry about the delay here.
Making the whole AutoTypeLoc resolve to the concept doesn't seem right - the `auto` part does not refer to the concept, and in principle refers to another type entirely. Really I think there should be a child node representing just the concept part, probably a ConceptReference. Compared to the case we discussed where constrained auto is a template type param, this looks easier: all the information is there in AutoTypeLoc. I think we need RecursiveASTVisitor to assemble it into a ConceptReference, and (to use it with SelectionTree) we need the ability to store it in a DynTypedNode. ================ Comment at: clang-tools-extra/clangd/unittests/FindTargetTests.cpp:581 + + [[Fooable]] auto i = 42; + )cpp"; ---------------- this is going to have the same behavior on the `auto` token, right? This is my main practical concern, that go-to-definition, hover, find-refs, go-to-type etc on `auto` will now treat `Fooable` as their target. (That said, I'm not sure exactly how common it is for `auto` to be constrained in a non-dependent context...) Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D154853/new/ https://reviews.llvm.org/D154853 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits