jh7370 wrote: > Hello! Sorry, new here. Why is it preferable to have that comment to > `LIT_XFAIL` deep in test code where someone new to the repo may not be able > to find it (at least not without AI help) instead of just invalidate/skip the > test as this PR would have appeared to do? Does this issue not have a 100% > repro rate on Darwin architecture? I'm running into the same failure and I > don't have Crowdstrike on my machine (or at least I assume I don't, followed > https://www.help.brown.edu/wiki/spaces/kb/pages/4446847031/How+to+Confirm+that+your+CrowdStrike+installation+was+successful#Launching-the-Application > and nothing shows up)
It sounds to me like you have a different setup to what others have had. I'm fairly confident that some LLVM build bots run the tests on Darwin as standard, which means that we'd have noticed by now if there was a 100% failure rate (or even an occasional flakiness). That means that if you're seeing a 100% failure rate locally, something on your local system is causing that. It could be a different AV, it could simply be your OS configuration is somehow different to the build bots etc. As for why the comment is "deep", it isn't really - it's right there in the test file that has the problem. Normal test debugging procedure would see you go to the test that is failing so you can inspect what that test is doing. The comment is then easily visible and explains what to do about the issue. Expanding that comment to other known failure cases is welcome, if you can pinpoint what is causing the behaviour, of course. The LIT_XFAIL variable is also documented in the lit documentation (see https://llvm.org/docs/CommandGuide/lit.html#cmdoption-lit-xfail). https://github.com/llvm/llvm-project/pull/192521 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
