Repository navigation
Pin pyerfa to fix test failures with version 1.7.2 - #11308
Conversation
| install_requires = | ||
| numpy>=1.17 | ||
| pyerfa>=1.7 | ||
| pyerfa>=1.7,<1.7.2 |
There was a problem hiding this comment.
Btw, we also have a requires = block above which I think is useless. But will keep that for another PR.
https://setuptools.readthedocs.io/en/latest/references/keywords.html
requires is superseded by install_requires and should not be used anymore.
| install_requires = | ||
| numpy>=1.17 | ||
| pyerfa>=1.7 | ||
| pyerfa>=1.7,<1.7.2 |
There was a problem hiding this comment.
Maybe this is safer in case we decided that the bug is really upstream and we can never use astropy with this particular bug fix version? (Not sure about the syntax itself.)
| pyerfa>=1.7,<1.7.2 | |
| pyerfa>=1.7,!=1.7.2 |
There was a problem hiding this comment.
Not sure which one is best, it could also be that tests on Astropy need to be updated. So we can update this when @mhvk can have a look.
There was a problem hiding this comment.
Yes, definitely some tests in astropy need to be updated. But I don't understand the test that you first found to fail... Anyway, good to not have to worry about breaking unrelated builds!!!!
There was a problem hiding this comment.
I think the one in PR is better, as we can't yet guarantee that we would be compatible with any of the future versions. In the past, there was only one version bundled, so it's probably not as a big deal to stay very restrictive within astropy.
(Ideally, we should be stating upper bounds on versions, too, hopefully, the package resolvers will get better support for that soonish)
There was a problem hiding this comment.
Yes, agree that for now to just effectively pick a given version is best. I do hope to get around to fixing the astropy tests soon, and will try to make it version-dependent, as we do for numpy.
pllim
left a comment
pllim
left a comment
There was a problem hiding this comment.
Okay, let's get this in to remove block... Thanks!
pllim
added
zzz 💤 merge-when-ci-passes
|
Remote data job failure unrelated, though I would think the numpy nightly wheel would have been updated by now and the errors would disappear but that is another issue. 32-bit job failure also unrelated. |
Ref #11305