https://github.com/tblah commented:
I would imagine that a language frontend could have more precise alias information to work with. For example, in flang we do our best to hack fortran language rules into TBAA (plus some attributes) but it isn't a perfect fit for the language semantics. That said, if this provides enough precision for your use case then nothing about this is incorrect. Alias analysis is done in the middle end for a lot of cases. https://github.com/llvm/llvm-project/pull/218453 _______________________________________________ llvm-branch-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
