erichkeane added a comment. In D51650#3127097 <https://reviews.llvm.org/D51650#3127097>, @aaron.ballman wrote:
> In D51650#3126569 <https://reviews.llvm.org/D51650#3126569>, @akuegel wrote: > >> Since it is not clear whether the semantic change was intended, I think it >> makes sense to revert the patch for now. If it is intended, it might be good >> to mention it in the change description, so that people are warned. > > That looks like an unintended change to me, likely due to the new mutual > exclusion checks. Thanks for letting us know! It was SORTA intended, I was going for 'conservative fix' here, but the 'target' change was more of a 'while I was there' bit. I'm away for the next two weeks, but can re-land this without that plus a test to validate a winner (as that wasn't really the case before IIRC). That said, the above example gives me some willies with regards to setting up multiversioning.... The interaction of two declarations here, 1 which adds a 'target' to the pragma-push seems fishy/confusing. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D51650/new/ https://reviews.llvm.org/D51650 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
