Henry Schreiner

Thanks, I can reproduce with that, I'll try to look at it soon. ~~Also reminds me that I need to make the coverage opt-in.~~ Ah, seems that's in https://github.com/scikit-build/scikit-build/pull/719. Maybe...

I'll try to work on this this week. I'll have more time for Scikit-build starting Sep 1 ([grant funded!](https://iscinumpy.dev/post/scikit-build-proposal/)), but part of that is on the rewrite (which has technically...

I believe the problem is you have setuptools_scm installed in the environment (seen via `pip list`). This adds everything in git to the SDist, which causes the failures seen here.

Yes, this is planned. Currently any option scikit-build needs has to be passed to setup, and can't be placed in setup.cfg. (Ones it does not need are fine).

This is https://github.com/pypa/setuptools/issues/2912. Since we use a setuptools backend (until/if https://iscinumpy.dev/post/scikit-build-proposal/), I don't think we should do better than setuptools (at least yet). This really should be fixed there.

Probably similar to https://github.com/pybind/cmake_example/blob/544111c7d6184559b54c5d8c6b1a613314904b0a/pyproject.toml#L25 (not scikit-build, but same issue from setuptools).

I'm pretty sure there's a bug lurking that https://github.com/scikit-build/scikit-build/pull/582 was trying to fix, but it's been hard to reproduce. Possibly related?

The fiddling with the package dir likely would affect more than just editable installs. We very much do not support PEP 621 configuration yet (see https://github.com/scikit-build/scikit-build/discussions/601 or our wiki for...

Not _right_ now - biggest issue is we have a wrapper around `setup` that's a pain to maintain and doesn't intact well with commands. We can't handle setup.cfg, much less...

I list some of them in https://iscinumpy.dev/post/scikit-build-proposal/. I'd probably do something like my page here: https://cibuildwheel.readthedocs.io/en/stable/working-examples/, but this would likely be done as part of that project if it gets...