(TEST) Apply Florian's patch to 2.30.x #4204
Conversation
TomAugspurger
changed the title
|
The test failure on windows 3.7 was a know flaky test that's now skipped in #4166. It can be ignored. |
|
I opened a separate PR over in #4212 to see what test failures we currently have in the main branch and saw there are many failures on both Windows and Travis CI builds. Lots of the test failures (maybe all?) look the same as the ones we're seeing in this PR, but it's difficult to directly compare the builds when there are > 20 failures. I ended up not releasing 2.30.1 last Friday as, I think, the Travis builds weren't completed late in the evening and then on Saturday I saw more tests were failing than expected. Given the current state of our test suite, I'm inclined to not release right now and instead focus on fixing tests. Thoughts @TomAugspurger? Also, I can cross-post this over in the release issue on the community tracker |
Agreed. |
(cherry picked from commit debe1b8)
|
Woo, tests pass 🎉 I'll carry on with releasing 2.30.1 |
|
Great, thanks! |
2 participants
Trying out #4176 against 2.30 to check if the failures on that PR are only present on master.
If this passes, I don't think we merge this PR. Instead, we should merge #4176 despite the failing tests and then just backport the merged PR to 2.30.x. That should play more nicely with versioneer.