aaron.ballman added a comment. In https://reviews.llvm.org/D41056#951083, @alexfh wrote:
> In https://reviews.llvm.org/D41056#950605, @khuttun wrote: > > > In https://reviews.llvm.org/D41056#950570, @Eugene.Zelenko wrote: > > > > > May be //bugprone// is better module then //misc//? > > > > > > Maybe. I can move it if all the reviewers think that it would be better > > suited there. > > > Yup, bugprone- should be a better category for this, IMO. > > I wonder whether libc++ folks are interested in marking unique_ptr::release() > with `__attribute__ ((warn_unused_result))`. A compiler warning (with > -Werror) would be a better protection against this kind of a bug. There's a push in WG21 to mark more of the library with `[[nodiscard]]`: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2017/p0600r1.pdf If we have a check for this, I do not think it should be specific to `unique_ptr::release()`, but instead be more broadly applicable to APIs that should be marked `[[nodiscard]]` but are not (currently). P0600R1 is a good place to start, but I'm guessing there are POSIX APIs (among others) that would also qualify. Repository: rL LLVM https://reviews.llvm.org/D41056 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
