Collaborate with Scientific Python?

[pllim]

During the Scientific Python Developers Summit 2023, there was interest to use GitHub Actions workflows from here, particularly for wheel building. Even though the immediate need at the summit was a simpler workflow than what is supported here, it is undesirable for Scientific Python to reinvent the wheel when OpenAstronomy workflow is generic enough to be upstream and can be beneficial to the wider scientific Python community. Are you interested to collaborate with them?

cc @bsipocz

[ConorMacBride]

Hi @pllim, this is a great idea — I'm always in favour of not reinventing the wheel (building)! I would be interested in collaborating to make the workflow(s) suitable for the wider community. Is there anything in particular that should be discussed? Any missing features/use cases? Are they hoping to move this repo to the Scientific Python org (this would break CI for existing users)?

[Cadair]

I think I did mention the existence of this repo on the Scientific Python discourse a while back, so hopefully people know about it. I am all in favour of making this work for everyone.

[pllim]

I had made Scientific Python aware of your interest in this issue yesterday in their Discord. So hopefully people from there will respond here if they are interested to collaborate. 🤞

[stefanv]

It would be interesting to see how we can use and contribute back to these!

On the one hand, it would simplify workflow configuration a lot. On the other, I note that we often have to tweak our workflows to work around compiler & platform issues. See, e.g., the skimage wheel building workflow.

[Cadair]

Figuring out how to nicely expose CIBW config while giving people the power to do complex things is certainly a balancing act. See #77 #126 #52 as a starter. Would certainly be good to have input from people who push CIBW harder than we do.