steakhal wrote: Before I'd continue my review, I'd like a bit more context about the crash. So we ended up with different "sorts", but I feel like that is because we don't capture implicit conversions that would play the role of widening or matching the operands type like standard arithmetic conversions and promotion rules would dictate.
I also looked at the test cases and it wasn't apparent where exactly it crashed, and why we need a stack address escape diag to trigger it. My best bet would be that we need some diag to trigger refutation, but then we could use a `clang_analyzer_warnIfReached` I think to convey this. Could you tell us a bit more about this crash? And I we have not touched any related parts for ages now, how did you discover this now? Did you improve your fuzzer to cover more patterns? https://github.com/llvm/llvm-project/pull/143310 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
