randyli planned changes to this revision.
randyli added inline comments.
================
Comment at: clang/lib/Sema/SemaExprMember.cpp:690
+ // int n = a.B::m;
+ if (BaseExpr && isa<CXXRecordDecl>(DC) && isa<CXXRecordDecl>(RDecl)) {
+ CXXRecordDecl *SRecord = cast<CXXRecordDecl>(DC)->getCanonicalDecl();
----------------
rjmccall wrote:
> We don't generally cite bug numbers in the source code; we just cite the
> language rule. You can reference the bug number in your test, though.
>
> The standard says that this is an error even when the member access is
> implicit, so the check for `BaseExpr` is incorrect.
>
> Conversely, the standard says this is not an error when the declaration
> referenced is not a non-static data member or member function, so you cannot
> do this before the lookup; in fact, you cannot do it until we've fully
> resolved the lookup, which may require contextual information.
>
> All of the code paths that build a non-static member eventually funnel into
> `BuildMemberExpr`, so maybe that's the right place for the check? But note
> that you have to check explicitly for a non-static member, because
> `MemberExpr` is used for all of them.
>
> Technically, this rule wasn't in C++98, but since it does appear in C++03, I
> agree we should just enforce it unconditionally, as we usually treat C++03 as
> an errata release rather than its own language mode.
ok, thank you for the suggestion, I'll look into BuildMemeberExpr and try
again
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D122460/new/
https://reviews.llvm.org/D122460
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
