Final 9.6 upgrade is catastrophic on Debian testing

87 views
Skip to first unread message

Emmanuel Charpentier

unread,
May 16, 2022, 7:13:33 AM5/16/22
to sage-r...@googlegroups.com

Observed on two systems running Debian testing (updated daily).

On these two systems, 9.6.rc4 (resp 9.6.rc4) gave satisfying results (already reported). After daily system update, the compilation of 9.6 went well, bit ptestlong gave catastrophic results : 57 (fifty seven, count 'em !) segfaults...

Enclosed an archive containing :

  • config.log
  • config.status
  • logs\install.log
  • logs/ptestlong.log

I'm at loss

-- Emmanuel Charpentier

catastrophic.tar.gz

Dima Pasechnik

unread,
May 16, 2022, 7:17:53 AM5/16/22
to sage-release
Have you rebuilt from scratch after the system update? Any update of a dynamic library used by Sage can lead to lots of errors, unless the relevant components are rebuilt.


--
You received this message because you are subscribed to the Google Groups "sage-release" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-release...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/b6cb0005953d0196ed460cd7159e9165b0eb18b4.camel%40free.fr.

Emmanuel Charpentier

unread,
May 16, 2022, 7:24:04 AM5/16/22
to sage-release

Le lundi 16 mai 2022 à 13:17:53 UTC+2, Dima Pasechnik a écrit :

Have you rebuilt from scratch after the system update?

Nope. That’s usually unnecesary.

Any update of a dynamic library used by Sage can lead to lots of errors, unless the relevant components are rebuilt.

I’ll try that (i. e make distclean && ./configure --enable-download-from-upstream-url && make ptestlong) . Not especially fun…

Emmanuel Charpentier

unread,
May 16, 2022, 8:56:14 AM5/16/22
to sage-release

After make dist-upgrade, rebuilding Sage faiis opn the documentation :

[sagemath_doc_html-none] NameError: Singular library 'freegb.lib' not found
[sagemath_doc_html-none] 
[sagemath_doc_html-none]     Note: incremental documentation builds sometimes cause spurious
[sagemath_doc_html-none]     error messages. To be certain that these are real errors, run
[sagemath_doc_html-none]     "make doc-clean doc-uninstall" first and try again.

This happens even after make doc-clean doc-uninstall. Still stuck.

Do you need logs/install.log ?

Emmanuel Charpentier

unread,
May 16, 2022, 10:03:05 AM5/16/22
to sage-release
On another machine, Sage builds, but `ptestlong` fails with a slightly smaller list of failures :

```
----------------------------------------------------------------------
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/cyclic_covers/cycliccover_finite_field.py  # 1 doctest failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/dynamics/arithmetic_dynamics/projective_ds.py  # Killed due to abort
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/curves/projective_curve.py  # Killed due to abort
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/curves/affine_curve.py  # 48 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/rings/function_field/ideal.py  # 12 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/curves/closed_point.py  # 23 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/elliptic_curves/ell_field.py  # Killed due to abort
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/rings/function_field/differential.py  # 2 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/product_projective/rational_point.py  # 1 doctest failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/curves/curve.py  # 14 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/interfaces/singular.py  # 5 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/curves/constructor.py  # 4 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/schemes/curves/point.py  # 6 doctests failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/libs/singular/function.pyx  # 1 doctest failed
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/rings/polynomial/multi_polynomial_element.py  # Killed due to abort
sage -t --long --random-seed=292720829567961557971978070321731961282 src/doc/en/constructions/algebraic_geometry.rst  # Killed due to abort
sage -t --long --random-seed=292720829567961557971978070321731961282 src/sage/libs/singular/singular.pyx  # 1 doctest failed
----------------------------------------------------------------------
```  

Suggestions ?

Thierry Dumont

unread,
May 16, 2022, 11:48:22 AM5/16/22
to sage-r...@googlegroups.com

Starting from a fresh clone:

Compilation ok.

A warning when doing "sage -optional":

/usr/local/sages/sage/local/var/lib/sage/venv-python3.10.3/lib/python3.10/site-packages/sage/misc/package.py:119:
UserWarning: failed to fetch the version of pkg='ore_algebra' at
https://pypi.org/pypi/ore_algebra/json
warnings.warn("failed to fetch the version of pkg={!r} at
{}".format(pkg, url))


t.d.

Emmanuel Charpentier

unread,
May 16, 2022, 5:56:52 PM5/16/22
to sage-release

After perusing the failures log and some wild guesses, problem worked around by ./configure --with-system-python=no.

Does this problem deserve reporting to Debian (i. e. filing a ticket against Debianpython3 package) ?

Emmanuel Charpentier

unread,
May 17, 2022, 5:30:50 AM5/17/22
to sage-release

Le lundi 16 mai 2022 à 23:56:52 UTC+2, Emmanuel Charpentier a écrit :

[ Snip... ]

Does this problem deserve reporting to Debian (i. e. filing a ticket against Debianpython3 package) ?


I did it. But its precision may leave something to desire...

HTH,

Maxime Bombar

unread,
May 17, 2022, 1:56:12 PM5/17/22
to sage-r...@googlegroups.com, Emmanuel Charpentier
On 5/16/22 23:56, Emmanuel Charpentier wrote:

After perusing the failures log and some wild guesses, problem worked around by ./configure --with-system-python=no.

Does this problem deserve reporting to Debian (i. e. filing a ticket against Debianpython3 package) ?

I just tried on Debian unstable, with the same Python version as in Debian testing, and I didn't get any error at all (all tests passed) either with Sage's own Python (3.10.3), or with Debian' s version (3.10.4).

Maybe it is something fixed between testing and unstable, or there is another issue, but in any case I don't think this problem deserves reporting to Debian.

--
Maxime

OpenPGP_signature

Matthias Köppe

unread,
May 17, 2022, 6:54:38 PM5/17/22
to sage-release
We automatically test on debian-bookworm ("testing") on GH Actions. Here's the run for the 9.6 release tag: https://github.com/sagemath/sage/runs/6443847112
(all good).
Reply all
Reply to author
Forward
0 new messages