Evgeni Burovski

Rebased on main to resolve conflicts this PR picked up, and to replace `dev.py` with `spin` in the docs.

Okay, I think I addressed all review comments modulo two higher-level questions, https://github.com/scipy/scipy/pull/22743#pullrequestreview-3020065024 and a techical glitch, https://github.com/scipy/scipy/pull/22743#discussion_r2207857664 > If MKL works, I'd expect Accelerate to work as well pretty...

> This stood out more to me after the change to *_lp64_dep. A lot of these should already be ready for ILP64, e.g. special (gh-12085), lbfbgs (gh-12009), trlib (gh-12030). See...

> @ev-br the commit history is a bit messy, I'm considering squash-merging it - unless you have a different preference? +1 to squash merge, yes.

> Could you open a new tracking issue for per-module ILP64 support wherever that needs to be added back? https://github.com/scipy/scipy/issues/23351

This looks like a great start @ilayn! Apologies it took me a while to get to it. A couple of high-level comments first: - `surfit.f` is exposed to python in...

> But anyways, is it ok that I translate what I find in the dependency tree and then we salvage what needs to be salvaged? Absolutely, yes. We'll be able...

`_fitpack` / `dfitpack` duplication is historical [1], cf https://github.com/scipy/scipy/issues/16729. The goal is to have both of them gone together with the Fortran code they wrap. `_dierckxmodule` is new in scipy...

Thanks @ilayn. On that mermaid, we need the following top-level routines and their dependencies: - surfit - sphere - spgrid (unless @czgdp1807 has already found a way to add it...

> I don't quite see what is wrapped (or if anything used) directly from _dierckxmodule.c. But seems like nothing is used from F77 code. Not sure what you mean? These...