Without further comments specific to NumPy, I understand we should wait for resolution of the related scipy/scipy and scientific-python/spin PRs to be resolved before looking at this one. Should we mark it as "draft" until then?

scientific-python/spin#242 and scientific-python/spin#241 are also related. What I'd really like is if we could simply delegate to upstream, adding the one argument we want to add, without needing to update our spin build command every time something changes in spin.

Note that spin should not be used to generate the changelog and update the notes, those are both part of the release preparation, not the document generation.

This seems reasonable once the linked spin PR lands, but it doesn't do anything extra yet, it just adds the support code to be able to pass arguments to meson compile and meson install. So I'd put this on hold until it's actionable, and then add some new feature that uses that added flexibility.

Note that spin should not be used to generate the changelog and update the notes, those are both part of the release preparation, not the document generation.

@charris spin is a UX for development tasks (and that can/does include recurring release tasks), and it already has spin changelog and spin notes tasks, so I'm not sure what your comment is referring to. There's nothing in this PR that changes anything here.

Closing this PR. This change should be inside spin and not here. I will give some context shortly. Sorry for the noise.

Regarding context please see my explanation in scientific-python/spin#238 (comment)