Hi Maintainer

The upload of python-sparse 0.14.0-1 is failing its own autopkgtest on
arm64 and s390x [1][2].  I've copied what I hope is the relevant part
of the log below.

Regards
Graham


[1] https://ci.debian.net/packages/p/python-sparse/testing/arm64/
[2] https://ci.debian.net/packages/p/python-sparse/testing/s390x/


60s autopkgtest [15:18:15]: test command1: set -e ; cp -r sparse/tests
"$AUTOPKGTEST_TMP" ; for py in $(py3versions -r 2>/dev/null) ; do cd
"$AUTOPKGTEST_TMP" ; echo "Testing with $py:" ; $py -m pytest -v ;
done
 60s autopkgtest [15:18:15]: test command1: [-----------------------
 60s Testing with python3.11:
 60s ============================= test session starts
==============================
 60s platform linux -- Python 3.11.6, pytest-7.4.3, pluggy-1.3.0 --
/usr/bin/python3.11
 60s cachedir: .pytest_cache
 60s rootdir: /tmp/autopkgtest-lxc.ia_llwxf/downtmp/autopkgtest_tmp
 63s collecting ... collected 5496 items
 63s
 67s tests/test_array_function.py::test_unary[mean] Fatal Python
error: Segmentation fault

At least on arm64, this appears to be a crash in a @numba.jit function.
Given numba's known issues on non-x86 (e.g. #1033907), this suggests
that actually fixing it _may_ not be reasonable.

In pandas I don't recommend/build/test-depend on numba on the problem
architectures, but python-sparse has a hard Depends on numba so that
doesn't appear to be an option here.

A potential option is to disable the tests on these architectures and
display a warning on import (for an example, see
https://salsa.debian.org/science-team/statsmodels/-/commit/a64430166b1e48135c58ec41330e3ce9b2206b46
).

Hi Nilesh,

Am Sat, Jan 13, 2024 at 04:15:07PM +0530 schrieb Nilesh Patra:

Thanks a lot.  I'll wait for debci whether bug #1055847 persists.

I wonder whether we could close simply #1058485 since the test is passing
with current numba.

Kind regards
     Andreas.

Hi Diane and Ghislain,

you are listed as Uploaders of python-sparse.  Since I now have other
tasks than maintaining team maintained packages I would be really happy
if you could subscribe upstream issue

https://github.com/pydata/sparse/issues/628

and continue what I have started.

Thanks a lot
   Andreas.

I subscribed, though I'm still fighting with updating numba.

Also I know there are many ways that numba can fail, so odds are
sparse's problems are caused by numba.

There are definitely problems with numba on arm64.
https://github.com/numba/numba/issues/9109

(Also congratulations on your election, I have respect your
organizational skills on the med team)

Diane

The autopkgtest of python-sparse 0.16.0a9-1 seems to be passing on
arm64 now, and numba is not yet built on s390x.