Maintenance status - Githubissues

scikit-learn-contrib / sklearn-ann

Integration with (approximate) nearest neighbors libraries for scikit-learn + clustering based on with kNN-graphs.

https://sklearn-ann.readthedocs.io/en/latest/

BSD 3-Clause "New" or "Revised" License

14 stars 4 forks source link

Maintenance status #8

Closed flying-sheep closed 9 months ago

flying-sheep commented 1 year ago

Hi, is this still maintained?

We’d be interested to include something like it into scanpy.

In case you’re no longer interested in maintenance, maybe we could transfer this project to scverse and take over, publish it to PyPI, …

frankier commented 1 year ago

Hi @flying-sheep ! Thanks for getting in contact about this.

It's not currently maintained, no. On the other hand I did start doing the work to try and polish this a bit more. I would therefore like to make a counter-proposal.

My original intention before getting distracted by other work was to include this scikit-learn-contrib. So, what if I try to go ahead with transferring it there? I would then be happy to add you onto the relevant scikit-learn-contrib team. Would this work? From my perspective it could be better to keep this in an org with a general emphasis if possible.

flying-sheep commented 1 year ago

That sounds like a great idea! What would be needed to transfer it there?

frankier commented 1 year ago

Hi! So I am a bit time poor at the moment -- I'm unplugging for a couple of weeks starting tomorrow. It looks like one of the requirements is to setup CI. I tried this but https://github.com/frankier/sklearn-ann/actions/runs/5412841941 .-- unfortunately FAISS has changed its interface. I will fix this and open an issue to move to sklearn contrib when I get back. I understand this might not fit your own timetable, but maybe we can still make it work? Let me know either way.

flying-sheep commented 1 year ago

We would prefer that to happen faster, but not at the expense of your free time of course.

I’m good at setting up CI and things, maybe I’ll just do that.

frankier commented 1 year ago

I recognise your handle, so I have added you if you want to fix up the CI to expedite things, I think I can open the issue on sklearn contrib before the 12th, otherwise I will fix ci and do it then. Phone only atm

flying-sheep commented 1 year ago

Thank you! And now I hope you don’t worry and enjoy your away time!

flying-sheep commented 1 year ago

OK, decisions. I like do do things the following way, in increasing order of opinion

[x] Ruff instead of flake8 for faster style checks and fixes
[x] pre-commit for formatting and style checks (no black or ruff in our dependencies)
- [ ] pre-commit.ci for running those checks (auto-update PRs from pre-commit.ci, and one can quickly see if a PR failure is tests or style because there’s two checks instead of one)
[x] Hatch instead of poetry for PEP 621 support (standardized pyproject.toml layout instead of poetry specific)
[x] tests outside of the source directory and package in src directory, as recommended by Pytest
[x] Squash & merge as only merge option on GitHub, branch protection

So yeah, I think Ruff is just a straight upgrade, so I’m just going to do that one.

frankier commented 1 year ago

All sounds fine to me!

flying-sheep commented 1 year ago

Great! Since I’m no admin, I can’t turn on pre-commit.ci integration, or “Squash & Merge”, …

Seems like the “don’t commit directly to the main branch” hook fails when run in GH actions. It should also work on pre-commit.ci:

grafik

flying-sheep commented 1 year ago

I also remembered another thing that would be great to enable:

[ ] https://docs.readthedocs.io/en/stable/guides/pull-requests.html

flying-sheep commented 1 year ago

Hi! I hope your time unplugging was nice! Do you think we can proceed with this?

frankier commented 1 year ago

Just putting the hatch PR together now. Hold on a moment.

frankier commented 1 year ago

I've tried to get the ball rolling over at https://github.com/scikit-learn-contrib/scikit-learn-contrib/issues/64

I think there might be some stuff to tighten up, but maybe now we can see how receptive the scikit-learn-contrib org is an what they think is most important to target.

frankier commented 1 year ago

Looks like no response. Would it be useful to upload this package to pypi (under its current name) in the interim?

flying-sheep commented 1 year ago

Maybe we can get in touch with them some other way. It says it’s maintained by one person, @egy1st, who seems to be active on GitHub, but hasn’t merged anything into that repo in 9 months.

@egy1st, is it possible that you ask the scientific python community for some help with reviewing things and getting them in?

egy1st commented 1 year ago

@flying-sheep @frankier

When you mentioned receiving a message that the repo is maintained by @egy1st, I needed clarification because I am not a member of the scikit-learn-contrib community. I have contributed an algorithm called denmune to their repo at https://github.com/scikit-learn-contrib/denmune-clustering-algorithm. I am just a contributor, a maintainer but not an admin/member. Since I didn't receive any private messages as egy1st, what message were you referring to, and when you received it? I don't have any inside information beyond the public GitHub thread you shared, so I'm unsure what additional messages you mentioned."

flying-sheep commented 1 year ago

Hmm, maybe I read the wrong readme then, sorry. I now see that you’re simply the maintainer of the last package that got added. Sorry about the ping!

@frankier the people who are actually part of that org are here: https://github.com/orgs/scikit-learn-contrib/people maybe we can get in touch

frankier commented 1 year ago

@adrinjalali Ping! Any info on what the next steps are to get this package added?

adrinjalali commented 1 year ago

We'll respond on https://github.com/scikit-learn-contrib/scikit-learn-contrib/issues/64, sorry the backlog is kinda long, but we'll get to it.