Debug GraalPy/uv test failures on Windows#2741
Conversation
|
That passed! So it seems the bug is limited to only gp311, when combined with uv/build[uv]. My guess is on some kind of Windows path bug with GraalPy311, does that sound plausible @timfel ? (see #2734 (comment) for the error we've been debugging, I'm curious about the use of D: vs C: drives as a possible trigger). My inclination is to skip the test, document known issue and move on. This workaround in the tests is ugly though (global state). How long would you like to continue supporting gp311 in cibuildwheel Tim? I remember spring 26 was a possible answer. In that case I'd be happy for this workaround to live in the tests for a few months. |
|
A test skip & mention in docs is probably fine. |
|
sorry, I have been out sick and had not seen this.
it does, we've found quite a few windows issues during the upgrade to 3.12
We released the latest graalpy in September, at some point we said giving people 6 months overlap should be fine, so I wouldn't mind dropping gp311 by the end of March, if that's ok with you? |
Sounds good. We can pull out this test hackery then. |
9c518a0 to
6c49819
Compare
|
@copilot add a small mention to docs about the incompatibility of GraalPy311 with |
* Initial plan * docs: note GraalPy311 incompatibility with uv/build[uv] on Windows Co-authored-by: joerick <1244307+joerick@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: joerick <1244307+joerick@users.noreply.github.com>
|
Gah! It's now decided to fail on gp312 as well! Here's the error-
Looks to be the same issue. I don't suppose you have any ideas @timfel ? |
8de21e9 to
b4ab257
Compare
|
My priority is to get the CI green, so I'm gonna skip that for now. #2754 tracks the bug. |
Debugging windows failures in #2734. Perhaps the error only occurs on GP311...