On Sat, Jul 05, 2025 at 04:55:59PM +0200, Marcin Owsiany wrote: > Looks like `st_mtime` is a lie, and Linux filesystems don't work the way I > thought they would :-D > > https://github.com/Rudd-O/ledgerhelpers/issues/28#issuecomment-3039053708 > > Not sure what's the best way to avoid these failures without a sleep()....
I've seen quite a number of race conditions in tests being solved by a simple sleep, so using a sleep() looks completely ok to me. But there is more, I see you wrote this in the PR: > As for the sleep duration, I tried 2 milliseconds at first, but then > it failed after a few hundred runs. Without a sleep, it failed after a > dozen or so runs. With 10ms it didn't fail even after a few dozen > thousand runs. It is awesome that you took the time to do that to ensure that the build always happens reliably. I wish everybody here had those high standards. Thanks a lot!
