[PATCH] D111175: [Clang] Extend init-statement to allow alias-declaration

[Aaron Ballman via Phabricator via cfe-commits]

aaron.ballman added a comment.

Just nits from me, but otherwise LGTM!



================
Comment at: clang/include/clang/Basic/DiagnosticParseKinds.td:556
+def warn_cxx20_alias_in_init_statement  : Warning<
+  "alias declaration in this context is incompatible with C++ standards before C++23">,
+  DefaultIgnore, InGroup<CXXPre2bCompat>;
----------------



================
Comment at: clang/include/clang/Parse/Parser.h:1982
+  ParseAliasDeclarationInInitStatement(DeclaratorContext Context,
+                                       ParsedAttributesWithRange &attrs);
 
----------------



================
Comment at: clang/lib/Parse/ParseExprCXX.cpp:1915
+Parser::ParseAliasDeclarationInInitStatement(DeclaratorContext Context,
+                                             ParsedAttributesWithRange &attrs) {
+  assert(Tok.is(tok::kw_using) && "Expected using");
----------------



================
Comment at: clang/test/Parser/cxx2b-init-statement.cpp:19
+
+    if (using enum ns::e; false){}  // expected-error {{expected '='}}
+
----------------
erichkeane wrote:
> Is there any way to have this elaborate more?  The 'expected 'TOKEN'' is my least favorite type of error.   Could we change this to be something like, 'expected '=' in type alias' or something?
FWIW, I'm fine with the diagnostic as-is because I don't expect anyone to actually *use* this functionality, so I don't expect a ton of users to hit this. I'd say it's fine to address when a user files a bug showing some code where the diagnostic is unclear.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D111175/new/

https://reviews.llvm.org/D111175