Evgeni Burovski

First and foremost, IMO this RFC is not only for C++ experts! It may impact us mere mortals as well. With the C++ momentum by @izaid and @steppi in scipy.special...

> compiler/ecosystem support remains a blocker for the foreseeable future? So if we drop the kokkos implementation somewhere into scipy, just as gh-20230 does, do we expect issues from the...

I'm going to close this issue since there was no follow-ups for several months and the general answer is https://github.com/scipy/scipy/issues/20152#issuecomment-1964871661. The difference between f32 and f64 hints at the delegation...

I wonder if this is related to https://github.com/scipy/scipy/issues/18533 Could you check if switching the lapack driver matters here?

- `propack` tests I think it's easiest to xfail / ignore: propack should be out of the door in not-so-distant future. Hopefully. - interpnd & qhull crashes might be related,...

There's also an einsum call in the qhull tests: https://github.com/scipy/scipy/blob/main/scipy/spatial/tests/test_qhull.py#L299 All in all a stack trace is needed to debug these further. Ideally it'd be helpful to see which python...

`lstsq` failures bring unpleasant memories :-). This was problematic back when windows XP was a thing, is skipped ATM on MUSL, and is problematic under conda-forge. So something is fishy,...

TestCloughTocher2D::test_dense and test_more_barycentric_transforms are very likely to have the same cause. There are too many moving parts though, to meaningfull troubleshoot ether of them without stepping through on a failing...

OK, meaning the lstsq crash is still there, just hiding better. Could you try this patch instead then: https://github.com/ev-br/scipy/commit/ed192aa0c7e54d642fe2e986b3bc5ca24bd1061e ?

Feel free to close if it's known