Hmm, ~1h 57m for an incomplete test... I doubt it's a good idea to run this on every pr. Perhaps, for tag push?

The sympy (why do we need sympy to test mpmath?!) failure seems to be related to the #704 (fixed by sympy/sympy#25290). It looks like Sage also uses mpmath.rational. Let me know if you think that we should do proper deprecation of this private module.

IIUIC, this integration will not run the mpmath tests suite (only own sage tests), so #705 should be kept open.

Hmm, ~1h 57m for an incomplete test... I doubt it's a good idea to run this on every pr. Perhaps, for tag push?

The SymPy CI runs this after a merge to master rather than on each push to a PR. It is a bit flakey and often fails for unrelated reasons.

Let me know if you think that we should do proper deprecation of this private module.

I would say that since the module is being used in some downstream codebases let's not remove it yet in the next mpmath release. I don't know what exactly was the history of all this or why it was being used but at least SymPy and Sage were using it so I would expect that something else is using it as well. It would be better to remove all known downstream usage first before removing or deprecating the module in mpmath.

IIUIC, this integration will not run the mpmath tests suite (only own sage tests),

This can be added from the Sage side by creating a test script build/pkgs/mpmath/spkg-check.in; PR welcome

The SymPy CI runs this after a merge to master rather than on each push to a PR.

Yeah I saw this. But I was hoping there is a faster way...

I would say that since the module is being used in some downstream codebases let's not remove it yet in the next mpmath release.

I think we can restore some minimal version of it (#728) and then - do proper deprecation.

This can be added from the Sage side by creating a test script build/pkgs/mpmath/spkg-check.in

Can this be patched right in the workflow, just as with update-pkgs.sh?

The SymPy CI runs this after a merge to master rather than on each push to a PR.

Yeah I saw this. But I was hoping there is a faster way...

Of course only a subset of the ~ 3000 modules of Sage may be enough to test. But we don't easily know which ones are relevant for proper coverage of mpmath.

This can be added from the Sage side by creating a test script build/pkgs/mpmath/spkg-check.in

Can this be patched right in the workflow, just as with update-pkgs.sh?

Sure.

But we don't easily know which ones are relevant for proper coverage of mpmath.

None. Mpmath has no non-optional dependencies.

see discussion in #705