-
-
Notifications
You must be signed in to change notification settings - Fork 7.9k
Test with Python 3.12 #24711
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
Test with Python 3.12 #24711
Conversation
b1b212d
to
64c9805
Compare
Status:
Should be replaced by |
The |
The strange this is the following:
which I interpret as webp being available. I tested installing webp before building pillow, but no change. |
The remaining failing test is apparently a bit tricky as it already have a few skip markers...
I guess the thing to do is to add yet another marker, probably for Python version. |
We should not be skipping that; it's a real bug:
|
Sorry, I've been seeing (and ignoring) the event loop failure locally. |
I have no idea how to proceed here. Spent some time trying to read up, but cannot really say that I found a workaround (simple enough for me to get it). |
I can try to put some cycles into fixing the event loop issues after we get 3.7 mostly out the door. |
on a bit of testing with cpython main this does not fail but it does fail with 3.12.0a3 and 3.12.0a4. I will finish bisecting this to sort out if this is intended to be fixed on cpython main or if they need a bug report that they un-broke us. |
Seems like it passes with 3.12.alpha-7. |
I'm seeing ~54 failures locally:
Some of these are the utcdatetime issue (being dealt with), some of these are Python segfaulting on exit (which I need to sort out a way to get a better reproducer of!), and some of these are things that look like:
that appear to be pyparsing or pytest doing something a bit wild and not escaping |
Also FYI, on Fedora, which has 3.12 in Rawhide, I don't have any of these failures with 3.8.0rc1. I think this just needs a rebase. |
330ebca
to
db5f2d7
Compare
b95b30c
to
436713a
Compare
I had to change to no-build-isolation to not pull in numpy 1.25 during installation. This means that the scheduled tests are not installing Matplotlib using the pre-release NumPy. I guess that is not really the purpose? |
Not sure if I somehow broke the Mac-test or if it is intermittent? Apart from that, I think this is good to go. |
The macOS failure is intermittent; skipped in #24597. |
Should this possibly be backported to 3.8? |
@meeseeksdev backport to v3.8.x |
…711-on-v3.8.x Backport PR #24711 on branch v3.8.x (Test with Python 3.12)
PR Summary
PR Checklist
Documentation and Tests
pytest
passes)Release Notes
.. versionadded::
directive in the docstring and documented indoc/users/next_whats_new/
.. versionchanged::
directive in the docstring and documented indoc/api/next_api_changes/
next_whats_new/README.rst
ornext_api_changes/README.rst