https://github.com/ojhunt requested changes to this pull request.
This change is not correct. We allow specializations that are not more specialized as an extension, and this PR makes them errors in all cases. This should have been understood from the use of an extension warning in checkMoreSpecializedThanPrimary rather than an error. Similarly the test changes were blindly changing existing accepted code into invalid code. That indicates a failure of the author to review the LLM slop prior to posting the review, and a failure to understand what it was doing. This only change that seems plausibly valid is the `if (Invalid) Specialization->setInvalidDecl()` change and honestly it seems more like a bug in setInvalidDecl that a call can make a previously invalid decl lose the invalid flag (@mizvekov what are your thoughts on making such a change?) The LLVM project policy requires that LLM generated code is vetted by the person posting it before putting the burden on project reviewers to verify it. If you're unfamiliar with the codebase you should be at least making an attempt and this PR does not seem to demonstrate that any such work happened. https://github.com/llvm/llvm-project/pull/181561 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
