feature request: putting kernprof (and line-profiler) on conda-forge

[kratsg]

I went ahead and put together a v1 recipe in conda-forge/staged-recipes#32134 for putting kernprof and line-profiler on conda-forge. I am happy to drop my name off / swap it with other team members if you'd like to be added to maintain the recipe there. I am unsure if the team is familiar with conda-forge but I would recommend having your package shipped there as well as this should also make it easy for users to use - say - pixi and run pixi run kernprof or similar and this pulls in everything they need. You'd also get the benefits of the cross-compilation support provided across the various platforms that is available with conda-forge.

[Erotemic]

I'm happy to have someone else take care of conda-forge maintenance. I've got no issue with it being on there, and I'm glad it is, I just don't have bandwidth to maintain it.

[kratsg]

I can help maintain to some extent, but I don't know how line-profiler in general is built, so I can mostly only guess superificially from looking at the setup.py. Do you know why the current setup.py requires the cmake wrapper? @henryiii could comment more, but he mentioned the current one should be rewritten.

[Erotemic]

It doesn't need it. It is what I was familiar with at the time, and I was working with scikit-build. I'd be happy to simplify it to a more modern and declarative pyproject.toml.

[kratsg]

@Erotemic thanks -- I think whenever the line_profiler can move towards a more modern setup that would be awesome. I unfortunately don't have time to commit as I am oversubscribed, but I can help with specific questions if you ping. The current recommendation (@henryiii can correct me) would be to use scikit-build-core if possible: https://scikit-build-core.readthedocs.io/en/latest/ with some documentation/guidelines here: https://learn.scientific-python.org/development/guides/packaging-compiled/

Re: maintenance -- most of the time will be automated PRs like this one: conda-forge/kernprof-feedstock#2 -- where the conda-forge ecosystem will autoupdate or do migrations of underlying pinned packages (e.g. new python version, new clang, etc...) and it usually requires just making sure everything passes before merging things in.