Repository navigation
MNT: Pin numpy<2 for v5.3.x - #15234
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.
|
| if isinstance(fitter, fitting.TRFLSQFitter): | ||
| ctx = np.errstate(invalid="ignore", divide="ignore") | ||
| if not NUMPY_LT_2_0 or not SCIPY_LT_1_11_2: | ||
| if not SCIPY_LT_1_11_2: |
There was a problem hiding this comment.
Hmm... Looks like this warning somehow went away again in dev scipy?
that now fails in devdeps with scipy dev. Since this is backport branch and the warning may or may not appear, let's just blanket ignore the unsuccessful fit because test case is not well defined to begin with.
|
I think this makes a lot of sense! |
|
@mhvk , wanna approve? 😉 |
| "setuptools_scm>=6.2", | ||
| "cython==0.29.34", | ||
| "oldest-supported-numpy", | ||
| "numpy>=1.25,<2", |
There was a problem hiding this comment.
Is there a reason to have the minimum build version higher than the minimum runtime version? That’s generally the thing that has caused problems in the past and is the reason for oldest-supported-numpy.
There was a problem hiding this comment.
Numpy 1.25 is the first version that supports backward compatible builds. The release notes explicitly mention that the mechanism makes oldest-supported-numpy not needed
|
@astrofrog and @saimn gave verbal approval on Slack, so merging. Thanks, everyone! |
|
What about v5.0.x? Should we make the same change there? |
|
I already did that much earlier due to something else. So now both of them are pinned. |
Description
I think it is reasonable to say that for v5.3.x, we will not support NumPy 2.0 or later? The support for v5.3.x will end anyway when v6.0.x is cut, which is schedule to be by the end of 2023.
Motivations:
main. Backport is just more churn.