[PATCH] D64914: Implement P1771

Richard Smith - zygoloid via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Sun Jul 21 04:45:38 PDT 2019

rsmith added a comment.

Please update cxx_status.html to mark P1771R1 as implemented in SVN.

Comment at: clang/test/CXX/dcl.dcl/dcl.attr/dcl.attr.nodiscard/p2.cpp:87-88
 // expected-warning at 28 {{use of the 'nodiscard' attribute is a C++17 extension}}
+// expected-warning at 66 {{use of the 'nodiscard' attribute is a C++17 extension}}
+// expected-warning at 71 {{use of the 'nodiscard' attribute is a C++17 extension}}
aaron.ballman wrote:
> erichkeane wrote:
> > aaron.ballman wrote:
> > > Is Core treating this paper as a DR? I don't have a strong opinion here, but for the nodiscard with a message version, I made it a C++2a extension. I don't have a strong opinion, but I sort of prefer doing whatever Core decides.
> > I am unfamiliar with what Core is treating it as, but my understanding is that EWG encouraged implementers to treat it as such.  
> We expose the attribute in all its glory in all language modes, so these changes already do what we want in effect. The only real question is whether we want to claim it's a C++17 extension or a C++2a extension. If a user turns on extension warnings, we should probably tell them when the feature was added, which is C++2a. It would be a bit weird to claim this is a C++17 when the feature test for it is `__has_attribute(nodiscard) == 201907L` (due to the normative wording changes).
> But if Core moves it as a DR, then C++17 is fine, though I suppose SD-6 would need to make clear what is required for each given attribute feature test value to give us the answer.
We moved this change as a DR, so this feature should be treated as if it were always part of the spec.



More information about the cfe-commits mailing list