SciPy does not use any of these but numpy.linalg does. We could replace it with a try/except and the with errstate(). That is slightly noticeable in speed (<2.5us of 10us for a single 3x3 inverse) due to the errstate overhead and slightly ugly.

Maybe starting with seterrobj and geterrobj will be easier. Although, I doubt the slowdown is bad enough to worry about (and can probably be slashed down in an eventual follow-up).

This also requires ripping out the C-API functions PyUFunc_GetPyValues and PyUFunc_handlefperr.

They are designed in an awkward way to allow use within a ufunc inner-loop. Something that we don't do (presumably anymore). I have not found any usage of them.

If people speak up with need, we (or rather they) could add the new PyUFunc_GiveFloatingpointErrors to the public API with backport. The function squashes both versions.

Basically, the proposal is now:

Remove the C-API functions and at least (set|get)errobj as well as the errobj= kwarg.

• np.errstate() supports everything linalg needs. The only caveat is overhead and I am sure that this can (and would presumably) be reduced by at least a factor of 2: it's fine.

Additionally, I would think it is a good idea to slash the API more, because few use it anyway:

• seterr, seterrcall, setbufsize could be folded into something like np.errstate(...).apply(). (Occasionally it is useful to set things globally; this would discourage that as well as reduce API.)
• The introspection geterr, geterrcall, and getbufsize could also be attributes on np.errstate(). I think that makes sense: they are probably practically never used, but no introspection capability would be weird.

Those changes are:

• Low impact: You may get an error and the replacement isn't super tricky and shouldn't be impossible to find.
• Few users affected: seterr will be used mainly, most of which should just switch to errstate anyway. A few people who set it globally may need a try/except hack. (If anyone worries about that, or people complain, we could keep seterr around or add it back.)

(The C-API change is in theory higher impact. But its hard to see why anyone needs to align with our machinery and we should be able to add work-arounds in the off-chance that someone complains. Plus, there are no known users, so it wouldn't matter even if it was higher impact.)

EDIT: I will add removing of log and print. For no particular reason beyond that call should be enough to log in practice and that print to stdout is weird and the only point for it is avoiding grabbing the GIL (which doesn't make sense in our use of this).

+1 for the cleanup. I would like to move to a contextvar as well