AaronBallman wrote: > > If you see failures locally, it's best to understand where they coming > > from: even if it passes on bots it just means we have a hole in our test > > coverage. > > Sure, but it wasn't the first time I've seen local test failures that doesn't > reproduce anywhere else (`Clang :: LibClang/symbols.test`). By the time I > merged this PR, me and Aaron have spent hours staring at the changes without > any progress. So we wanted an additional confirmation that this is not in the > same as `Clang :: LibClang/symbols.test` failure, especially looking at > pre-commit CI that reported just a several seemingly unrelated clang-tidy > failures. I hope it is acceptable that we decided not to spend even more time > before trying this PR out on wider range of buildbots.
I think everyone is correct here and we're all on the same page. Typically, local failures mean the code isn't correct so it's not ready to commit. However, one-off circumstances do happen where tests will fail locally but pass everywhere else (for example, people working on slightly out-of-norm configurations will sometimes have persistent local failures that are unrelated to changes in the patch). When precommit CI comes back green or with only false positives and you've got such local failures, speculative commits to see whether an issue "is real" or not do happen on occasion and are appropriate so long as necessary follow-up actions are timely (which they were in this case). https://github.com/llvm/llvm-project/pull/71322 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
