Add more asserts to try to track down RenderPadding intrinsics bug #32933
Conversation
Piinks
added
a: tests
|
Should #32817 be reverted along with this PR? That would provide one additional instance of this flake occurring and perhaps help with debugging. But, maybe it would happen too frequently that it would hurt the team's overall productivity from having to re-run the flaky tests. |
|
Has anyone been able to reproduce the failures locally on Windows? |
|
I'm not sure, I've only seen the failures in the context of the Windows Cirrus tests flaking. |
|
It's crazy, but the tests keep timing out on Cirrus on all three platforms. I wonder what is going on... |
|
(PR triage): @Hixie Do you still have plans to follow-up on this PR? |
|
I need to track down the timeouts. Haven't had much time recently. The timeouts are very suspicious. If someone wants to take it over that's totally fine. |
|
Dealing with timeouts in #33932 |
|
Turns out these asserts are just really really REALLY expensive (one test on my machine ends up taking about 18 hours to run instead of about 5 seconds). I'll have to try a different approach. |
See #19696