================
@@ -738,16 +738,47 @@ Parser::DeclGroupPtrTy Parser::ParseUsingDeclaration(
return nullptr;
}
- if (!Tok.is(tok::identifier)) {
+ Decl *UED = nullptr;
+
+ if (Tok.is(tok::identifier)) {
----------------
Endilll wrote:
1) You're correct that `ParseOptionalCXXScopeSpecifier` is going to provide us
a `TemplateIdAnnotation` every time there is an identifier followed by a "less
than" sign.
2) On top of what I do here, `ParseBaseTypeSpecifier` handles
`computed-type-specifier` (which is `decltype` and pack indexing) and dependent
types. If you take a look at which tokens that function expects, they are
`tok::kw_decltype`, `tok::annot_decltype`, `tok::annot_pack_indexing_type`,
`tok::annot_template_id`, and `tok::identifier`. If we remove tokens that are
coming from `computed-type-specifier`, we'll be left with just
`tok::annot_template_id` and `tok::identifier`, which are exactly the tokens I
expect.
> Line 723, ParseOptionalCXXScopeSpecifier, there are options to make it a type
> only context for error recovery
I'd be wary to mark this as a type-only context, unless the Standard says so. I
believe it doesn't for `using enum`, but does for base specifier, because the
latter uses `class-or-decltype` grammar.
https://github.com/llvm/llvm-project/pull/95399
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
