-
-
Notifications
You must be signed in to change notification settings - Fork 672
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
Correct flakiness in macOS CI #2637
Conversation
.github/workflows/ci.yml
Outdated
# Github runners seem to have intermittent connectivity issues. | ||
# See https://github.com/beeware/toga/issues/2632 | ||
- name: Tune GitHub-hosted runner network | ||
uses: smorimoto/tune-github-hosted-runner-network@v1 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This user providing this Action doesn't seem like they are likely to turn nefarious at some point...but pinning just v1
means we'll automatically start running any changes they merge. That said, the README says they plan to make no changes, so even pinning it to a specific commit hash should be fine. Throwing this out there as, at least, an observation since this is all pretty small risk.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fair; I'll update the pin to 1.0.0.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
i'll definitely be happy if this resolves the issues because half to three-fourths of my CI runs were failing on one of these 🤞🏼
testbed linux...
:( |
Dammit CI... can you hold it together for 30 minutes? I guess there's another problem to resolve... that's the first time I've seen this one, though. For the purposes of this PR, I'll rerun and pretend it never happened... :-) |
Corrects the intermittent failure seen in macOS CI with the
test_secondary_window_with_content
test.The underlying cause of the crash appears to be an issue with removal of constraints; since the test involves the addition of content in a window and then rapidly deleting that window, it's at least conceptually consistent that the issue is related to ObjC reference counting on the very-short-lived window.
The problem can't be reproduced with 100% reliability, so it's difficult to confirm with 100% reliability that the problem has been resolved; but I've re-run the macOS x86-64 test multiple times with this fix in place, and it doesn't seem to be occurring any more.
This isn't a very satisfying fix - this sort of cleanup shouldn't be required; but until beeware/rubicon-objc#256 is addressed, I think it's about as good as we're going to get.
Also includes the "network fix" suggested in the discussion for #2632.
Fixes #2631.
Fixes #2632.
PR Checklist: