TST: fix compatibility pytest 8.0#15809
Conversation
|
Thank you for your contribution to Astropy! 🌌 This checklist is meant to remind the package maintainers who will review this pull request of some common things to look for.
|
|
👋 Thank you for your draft pull request! Do you know that you can use |
2bfed6d to
f142b5e
Compare
f142b5e to
6b068da
Compare
|
Thanks! Failures only show up in the predeps job in daily cron, so I just kicked off the cron jobs. Maybe I need to add remote data directive to the devinfra job below because these were obviously overlooked. Not sure why I didn't catch these in #15054 ; that PR also had cron jobs run. astropy/.github/workflows/ci_cron_weekly.yml Lines 51 to 54 in 8737067 |
|
Huh why is devinfra even failing here? I have restarted a previously passing build on main from 3 days ago to see if this is even related or not to this PR: https://github.com/astropy/astropy/actions/runs/7375363390 UPDATE: Previously successful build now fails on rerun, so we can ignore devinfra failure in this PR. |
|
This failure is related: https://github.com/astropy/astropy/actions/runs/7412818159/job/20170416343?pr=15809 |
|
Oh good catch. You actually reported it already but I didn't see it locally so I assumed it was gone. I can look this one first thing tomorrow but feel free to push to this branch in the mean time if it's considered urgent. |
Since pytest 8.0 is still in RC stage and not released, not super urgent. I can wait. Thanks! |
pllim
left a comment
There was a problem hiding this comment.
The checks are probably too detailed but does not hurt. If I were to fix this from scratch, I would probably only focused on the original message and let the rest pass through, to avoid so many extra checks. But given that the work is already done and CI is okay with it, let's keep this and simplify in the future if we have to.
Thanks!
Not sure about the remaining ignore_sigint failure as I also cannot reproduce it locally. Looks like it is used during buffer flushing in hdulist.py. Though who are we to deny people from corrupting their own data if they choose to send a sigint when they flush buffer? 👹
|
I'll worry about the remaining failures later. One fire at a time. Thanks! |
|
Note for my future self: I was thinking about adding UPDATE: I think I confused myself. |
…809-on-v6.0.x Backport PR #15809 on branch v6.0.x (TST: fix compatibility pytest 8.0)
Description
This pull request is to address issues seen with pytest 8.0.0rc1, which makes previously hidden warnings visible in tests that use
pytest.warnsFixes #15807
I went with the simplest approach (besides straightforwardly ignoring newly discovered warnings) of blessing each warning as "expected", and just reject anything new that would appear in the future, but maybe in some cases these "new" warnings should be regarded as bugs, in which case it's not just a matter of adjusting tests, but I'll leave that up to be discussed by subpackages maintainers.