Understand why tests sometimes fail on windows

[paulmillr]

Sometimes it passes, sometimes it doesn't. Feels random.

How can we make tests pass more often? Other platforms have higher pass rate.

https://github.com/paulmillr/chokidar/actions/runs/12972801273/job/36180667608

[43081j]

i had another look at this recently and it seems to be the same test that fails at least

removing a directory recursively sometimes never emits the subdir's unlink event (i.e. the test expects an event for the dir being removed and the subdir below it)

but if you re-run, it usually passes. so it feels like a timing problem rather than the event just not being emitted. that, or windows just sometimes doesn't emit the events

[paulmillr]

Were you able to reproduce the issue locally instead of only github ci?

[43081j]

i haven't. last i checked, it worked locally

i'll give it another go next time im on my windows machine though 👍