rjmccall added inline comments. ================ Comment at: lib/Sema/TreeTransform.h:9135 @@ +9134,3 @@ + if (NamedDecl *D = R.getAsSingle<NamedDecl>()) { + D = D->getUnderlyingDecl(); + if (isa<FieldDecl>(D) || isa<IndirectFieldDecl>(D) || ---------------- getAsSingle already looks through to the underlying decl.
================ Comment at: lib/Sema/TreeTransform.h:9138 @@ +9137,3 @@ + isa<MSPropertyDecl>(D)) + return getSema().BuildPossibleImplicitMemberExpr( + SS, SourceLocation(), R, /*TemplateArgs=*/nullptr, ---------------- rnk wrote: > I believe this will always fail, so one alternative way to do this would be > to add a stripped down entry point into Sema just for template instantiation > that diagnoses UnresolvedLookupExprs finding instance members. At the very least, we should leave a comment explaining that we don't expect this to succeed, and that we're just doing this for diagnostic purposes. Might want to assert that as well. http://reviews.llvm.org/D6700 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits