Remove tests for Python 3.7 since it reached its end-of-life

[Andrew-S-Rosen]

Python 3.7 reached its end-of-life on 06-27-2023. As such, I removed it from the test matrix and from the classifiers list in setup.py. I didn't change the python_requires minimum version in setup.py though to allow for some flexibility for those that are slow to upgrade.

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: -0.09% ⚠️

Comparison is base (508c173) 88.14% compared to head (fb3c727) 88.06%.

❗ Current head fb3c727 differs from pull request most recent head 014b178. Consider uploading reports for the commit 014b178 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #851      +/-   ##
==========================================
- Coverage   88.14%   88.06%   -0.09%     
==========================================
  Files          44       44              
  Lines        3593     3593              
==========================================
- Hits         3167     3164       -3     
- Misses        426      429       +3     

see 1 file with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rkingsbury]

Thanks @arosen93 ! This has been on my mind for a while but I wasn't sure when to pull the trigger. You have convinced me. Can you please also update the docs to "3.8+"? There is this line:

Maggma is written in Python and supports Python 3.7+.

I didn't change the python_requires minimum version in setup.py though to allow for some flexibility for those that are slow to upgrade.

Tell me more about this. I suppose this means that a user could install maggma in a Python 3.7 environment, whereas if we change it to 3.8, they would be prevented from doing so? I'm on the fence about this approach; I kinda feel like as long as it is possible to install with Py3.7, we ought to be testing against it. But it is becoming a pain to support so I'm in favor of dropping it.

[Andrew-S-Rosen]

Happy to update the docs! Thanks for pointing that out.

And yes, you are interpreting the 'setup.py` details correctly. I'd personally vote to just update it to 3.8 but wasn't sure your opinion on it so took a more conservative approach. I suppose if a user really wants to use Maggma with Python 3.7, then they can just use a PyPI version that supports it.

[rkingsbury]

And yes, you are interpreting the 'setup.py` details correctly. I'd personally vote to just update it to 3.8 but wasn't sure your opinion on it so took a more conservative approach. I suppose if a user really wants to use Maggma with Python 3.7, then they can just use a PyPI version that supports it.

Thanks; yeah I think I'd prefer to keep everything consistent (e.g., we test until we truly drop support). @munrojm , any reservations about officially dropping Python 3.7 support?

[rkingsbury]

Side note, this question of how long to support 3.7 was a big part of what motivated my foundation PR on version support

emmet and atomate2 already set python_requires=3.8, and since those are probably the downstream packages that use maggma most, I think we're safe to drop 3.7

[Andrew-S-Rosen]

@rkingsbury: I agree. I think consistency is smart and there shouldn't be any issues here. I have now updated the PR to reflect this discussion, including both the update to the docs and the setup.py.

[rkingsbury]

Given that pymatgen is now at 3.9+, I'm gonna pull the trigger on this.

See materialsproject/pymatgen#3283