Bump Python requirements in setup.cfg and rmm_dev.yml

[shwina]

No description provided.

[leofang]

While you're at it, I think it'd be nice to also update this file? conda/environments/rmm_dev_cuda11.5.yml

[bdice]

Good catch @leofang. Other places referencing Python 3.7:

• conda/environments/*.yml
• README.md
• setup.py classifiers
• Several places in ci/local (which probably doesn't work at present due to other problems like outdated CUDA images -- maybe just leave this code alone)

[jakirkham]

Thanks Ashwin! 😀

Had a couple questions below

[jakirkham]

Do we need the upper bound here? Just wondering since we are dropping elsewhere

[shwina]

Yes -- these are the .yml files used to create dev environments. Since we don't typically want development with 3.10, we keep the upper bound here. This doesn't stop developers from manually building RMM with Python 3.10 if they desire.

[shwina]

Is it possible to do something like the following?

conda create -f rmm_dev_env.yml python=3.9

If so, we can (in theory) drop the upper bound here and make it the developer's responsibility to specify which Python they want to use.

[bdice]

That’s a good idea — I have seen other packages specify a yml with a user-provided python spec (edit: apparently this is not a thing, I was wrong). Then update the README/docs to include that specification in the commands to create a conda environment.

On that note, it might even make sense to remove the upper bound (use >=11.5) for cudatoolkit and recommend that users pick their version as needed, instead of maintaining two environment files for 11.5 and 11.6.

[shwina]

But is that a valid conda command, and does it do the expected thing here (override the python spec in the .yml?)

[bdice]

Darn. @shwina I could have sworn I had seen this before but the specifiers after -f stuff.yml are ignored, so you can't do what I was suggesting. Sorry about that. 😢

[shwina]

rerun tests

[github-actions]

This PR has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this PR if it is no longer required. Otherwise, please respond with a comment indicating any updates. This PR will be labeled inactive-90d if there is no activity in the next 60 days.

[vyasr]

@shwina what's holding this PR up? Is it just waiting on a resolution of the upper bound question above?

[shwina]

@vyasr I think that discussion can be resolved. This should be good to merge front m my perspective.

[vyasr]

@bdice @jakirkham let me know if either of you thinks that discussion is a blocker, otherwise I'll aim to merge this by EOD to get ahead of burndown. Trying to avoid surprises ahead of code freeze mid-GTC.

[jakirkham]

Nothing blocking from me

[vyasr]

Received approval from @bdice offline. Merging now

[vyasr]

@gpucibot merge