-
-
Notifications
You must be signed in to change notification settings - Fork 2.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix more test failures #28900
Fix more test failures #28900
Conversation
https://github.com/ppy/osu/actions/runs/9985890747/job/27597501295 In this case, the settings overlay is taking a very long time to load (on a background thread), and pops in when it finishes loading because it's been requested to open. The opens the settings overlay, closes it (by pressing escape, this does not actually close it because it's not loaded yet), and then enters song select by pressing 'P' 3 times. The settings overlay finishes loading at just the right opportune moment to eat one of the 'P' key presses.
https://github.com/smoogipoo/osu/actions/runs/9986761756/job/27599851263 This is a bit of a workaround, likely timing related. I don't foresee an until step in this case to cause false-passes.
https://github.com/smoogipoo/osu/actions/runs/9990112749/job/27610257309 Comments are loaded asynchronously, both from the initial request and the following message-post request. By sheer timing luck, these could be out of order and the assertion on the posted message could fail.
Re: 3f4e56b The test failure here can be reproed 100% on both macOS-arm64 and linux-x64 (on my beefy PC) under smoogipoo@a522c0e It's very weird so I'm investigating further. It's like our We have seen it on master: https://github.com/ppy/osu/actions/runs/9981737481/job/27587894137 It could be something super dumb, like JIT startup time. |
https://github.com/ppy/osu/actions/runs/10002179087/job/27648253709 The editor could be pushed before the exit actually occurs.
3838a1f
to
d7ae950
Compare
https://github.com/ppy/osu/pull/28900/checks?check_run_id=27652166871 This is an attempt. Going frame-by-frame I noticed that there's one frame in which the text is loaded but the FillFlowContainer/GridContainer haven't properly validated so the text is not positioned correctly (it's overflowing the panel to the left). If the cursor is moved at this exact time, then it may not be properly positioned for the following assertion, even though it is _somewhere_ on the panel. If the above is the case, then this is a known o!f issue, but not a simple one to solve. I haven't reproed this locally.
Makes it easy to compare this line versus the one in OsuGame.PresentBeatmap(). At the moment it's just GUID which is... not useful!
I'm going to bring this out of draft now with a few fixes, going into the weekend. It's still not where I want it to be but the remaining test failures are weird - some are timing out because the BDL threads are not running fast enough. |
I'm going to open this as a WIP PR for easier tracking. It'll be finished when I'm happier with the test results.