testdata: attempt to deflake goroutines test#5384
Conversation
|
I also find those failures very annoying. The tests are not so precise as to preclude these changes. Anyone have any objections for me to squash/merge? |
|
This kinda does defeat the original purpose of these tests. I wrote them a long time ago to show that TinyGo worked concurrently. What about only running this test on systems without thread support? I don't think they're flaky there. |
|
I think perhaps now that I have increased the runner sizes this should be less of a problem for the CI. |
|
I hit this frequently when doing local testing, so I don't think increasing a builder size really fixes it... |
b1deac8 to
a6bcc8e
Compare
Pushed that, at least |
|
@jakebailey can you please rebase against the latest |
a6bcc8e to
a2bb953
Compare
aykevl
left a comment
There was a problem hiding this comment.
Looks good to me. @deadprogram any thoughts?
|
Do we actually still need this? I think some recent changes such as #5487 have addressed the core problem. |
|
If that worked, then we can close this and the two other issues. |
|
Er, well, the part that works on #4965 is still relevant, but the timer stuff might not be? |
|
Going to close this for now, I have not seen a race in a while; maybe the goroutine one is still around but I haven't seen it either. Maybe both issues can be closed? |
|
Just hit goroutines again, sadly: |
a2bb953 to
dd5ecf4
Compare
dd5ecf4 to
3d18efc
Compare
|
Thanks @jakebailey now merging. |
Fixes #4965
I got tired of hitting this flake locally and in CI. This change seems to make it pass, but it's a little unclear without more eyes whether or not I have defeated the purpose of this test somehow as I of course arrived here after this was committed...