MuellerMP wrote: @GkvJwa Regarding the failing test: I have to apologize. It seems there is some Borland C++ compatibility switch that I did not know about that allows these constructs to be parsed at least. Regarding compilation this test does seem to fail very quickly though: https://godbolt.org/z/6eGfrr4Kn . I assume the Borland C++ switch does not have any maintainer and just has some parser tests.
I would suggest emitting the diagnostic only for non Borland C++. Regarding the Borland behavior: it might make even sense to deprecate Borland C++ support since its in such a broken state? But that really is not for me to decide. @efriedma-quic Can you confirm that we should move ahead with this diagnostic for non Borland C++? An alternative would be, like I said earlier, to support this SEH & C++ mixing in the SEH state numbering algo. In the latter case it might make sense to investigate what the Borland C++ compiler emits for the test code (especially which personality routine is chosen - since that is not an obvious decision IMO). https://github.com/llvm/llvm-project/pull/172287 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
