Fix scikit-sparse dependency#12
Conversation
matospiso
left a comment
matospiso
left a comment
There was a problem hiding this comment.
Thank you @mariansam for spotting this issue and for the fix.
My suggested changes are just cosmetics, apart from those I see no issue with merging this and releasing v1.1.1.
Regarding Python 3.9 - I don't think we should add support for it since it reached its end of life on October 31, 2025, meaning it no longer receives bug fixes, security patches, or enhancements.
Thanks again for your effort :)
| "scikit-sparse >= 0.4.8", | ||
| "scipy >= 1.7.3", | ||
|
|
||
| 'scikit-sparse < 0.5.0', |
There was a problem hiding this comment.
simply
scikit-sparse >=0.4.8, <0.5.0
should achieve the same,
the requires-python below will filter the incompatible ones
There was a problem hiding this comment.
That's not the case unfortunately. When trying to install scikit-sparse 0.4.8 under Python 3.11 (uv python pin 3.11 && uv pip install scikit-sparse==0.4.8) the installation fails with ModuleNotFoundError: No module named 'numpy' (at least when using uv) because that version of scikit-sparse does not specify the numpy dependency for that version of Python. Yikes.
There was a problem hiding this comment.
Oops, looks like I pushed wrong changes while testing the version requirements. Fixed, should be alright now.
this is a force-push as the original version commited changes that shouldn't have been commited
2 participants
Installing of the package broke a bit as scikit-sparse released 0.5.0 which isn't API-compatible. This PR fixes the versions required in pyproject.toml. Different scikit-sparse versions are compatible with different Python versions, the new setup supports Python >= 3.10. Perhaps we could add support for Python 3.9 to satisfy i-Hun's needs.
After merging, could you please make a release to PyPI so I can update the version in Cornac? See uv > Publishing your package.