sagemath: update to 10.2, flintlib: update to 3.0.1

[tornaria]

Testing the changes

• I tested the changes in this PR: YES

This is a major update for both sagemath and flintlib.

Sagemath moves to cython 3.

A couple of packages need rebuild or update:

• eclib: rbuild for flintlib-3.0.1
• singular: update to 4.3.2p10
• python3-pplpy: rebuild using cython 3
• python3-networkx: update to 3.2.1.

[dkwo]

Excellent. This builds and checks fine on aarch64, after rebasing.

[tornaria]

@dkwo sagemath is on its second RC so it will be released soon. This PR will be ready to merge once the release happens. It should ideally be tested together with #46785 and #46787.

[dkwo]

Can flint (and possibly sage) be built without ntl after this update?
i read somewhere that ntl might not be necessary anymore, and this would ease the burden of cross compiling.

[dkwo]

possibly eclib as well?

[tornaria]

Can flint (and possibly sage) be built without ntl after this update? i read somewhere that ntl might not be necessary anymore, and this would ease the burden of cross compiling.

I think flintlib is already cross-compiled.

As for eclib, it lists ntl-devel in makedepends. Have you tried building it cross without ntl-devel? If it works, we can just add a build option ntl that is disabled by default in cross builds. However, I think that eclib actually uses ntl for some things.

As for sagemath, I think it has a few more nocross makedepends. The comment in the template list some; including flintlib, but I'm not sure this is a mistake. It's possible that sagemath needs a flintlib that was built with ntl.

[dkwo]

The current iteration (rc4) has a couple of test failures on aarch64:

sage/libs/gap/element.pyx (killed due to segfault)
sage/matrix/matrix_integer_dense.pyx (killed due to segfault)
sage/graphs/generic_graph.py (timed out)

With regard to the crossbuild, I will investigate eclib and flint (in the past, our crossbuilt flintlib was giving a doctest error in sage).

[dkwo]

(previously, I was testing with XBPS_CHECK_PKGS=yes, yesterday I switched to full.)

[tornaria]

The current iteration (rc4) has a couple of test failures on aarch64:

sage/libs/gap/element.pyx (killed due to segfault)
sage/matrix/matrix_integer_dense.pyx (killed due to segfault)
sage/graphs/generic_graph.py (timed out)

With regard to the crossbuild, I will investigate eclib and flint (in the past, our crossbuilt flintlib was giving a doctest error in sage).

Can you try running those tests again? At least the matrix_integer_dense.pyx I think I've seen from time to time as a random segfault. The others I don't know, if you can reproduce consistently, then try to see what triggers it, get a minimal example that reproduces the issue, and submit a bug report upstream.

[tornaria]

@dkwo this is ready to merge, any comments?

[dkwo]

looks good to me.
i only see this failure on aarch64 with full check: sage/graphs/generic_graph.py (timed out).
@leahneukirchen can this be merged?

[dkwo]

that test passes when run from within installed sage.
all looks good :)