Sage 8.9.rc1 released

[Volker Braun]

As always, you can get the latest beta version from the "develop" git branch. Alternatively, the self-contained source tarball is at http://www.sagemath.org/download-latest.html 

2ab843c20b (trac/develop, tag: 8.9.rc1) Updated SageMath version to 8.9.rc1

26677b4e26 Trac #25026: Update jmol to version 14.29.52

582a9ef4fd Trac #28444: Fix backwards incompatibility of unpickling in Python 3

2eff516090 Trac #28059: Some citations are missing from pdf docs

be66263728 (tag: 8.9.rc0) Updated SageMath version to 8.9.rc0

[Eric Gourgoulhon]

On Ubuntu 18.04 running on bi-Xeon E5-2623 (8 cores) + 16 GB RAM:

- python2:

  - incremental build (from 8.9.rc0, with -j16) OK

  - all tests from ptestlong passed

- python3:

  - incremental build (from 8.9.rc0, with -j16) OK

  - tests from ptest-python3 passed, except for 1 permanent failure in

       src/sage/rings/polynomial/polynomial_rational_flint.pyx

       this is the error tracked at https://trac.sagemath.org/ticket/28334

  - tests from ptestlong passed, except for

----------------------------------------------------------------------
sage -t --long --warn-long 52.6 src/sage/libs/eclib/interface.py  # 1 doctest failed
sage -t --long --warn-long 52.6 src/sage/rings/polynomial/polynomial_rational_flint.pyx  # 1 doctest failed
sage -t --long --warn-long 52.6 src/sage/repl/attach.py  # 4 doctests failed
sage -t --long --warn-long 52.6 src/sage/libs/singular/function.pyx  # 1 doctest failed
sage -t --long --warn-long 52.6 src/sage/numerical/backends/generic_backend.pyx  # 2 doctests failed
----------------------------------------------------------------------

      The failed doctests are those tracked at https://trac.sagemath.org/ticket/28298

      The files

       - src/sage/libs/eclib/interface.py

       - src/sage/rings/polynomial/polynomial_rational_flint.pyx

      have permanent failures, while all doctests in

      - src/sage/repl/attach.py

      - src/sage/libs/singular/function.pyx

      - src/sage/numerical/backends/generic_backend.pyx

      pass when run standalone.

 

Eric.

[Emmanuel Charpentier]

Hmmm... This one is nasty...

On Debian testing running on core i5 + 8 GB RAM, upgrading from 8.9.rc0, I get:

Transient errors:

  src/sage/combinat/posets/posets.py               1 doctest failed

  src/sage/plot/animate.py                         7 doctests failed

  src/sage/combinat/matrices/hadamard~matrix~.py   2 doctests failed

  src/sage/misc/latex.py                           1 doctest failed

  src/sage/misc/persist.pyx                        2 doctests failed

  src/sage/coding/databases.py                     2 doctests failed

  src/sage/repl/load.py                            1 doctest failed

  src/sage/graphs/graph~latex~.py                  2 doctests failed

  src/sage/databases/findstat.py                   9 doctests failed

  src/sage/misc/remote~file~.py                    2 doctests failed

  src/sage/combinat/designs/ext~rep~.py            1 doctest failed

  src/sage/interfaces/magma~free~.py               2 doctests failed

  src/sage/interfaces/mathematica.py               Timed out

Permanent errors, already known:

  src/sage/libs/eclib/interface.py                          1 doctest failed

  src/sage/rings/polynomial/polynomial~rationalflint~.pyx   1 doctest failed

New permanent errors due to aborts:

  src/sage/crypto/mq/sr.py                                   Killed due to abort

  src/sage/rings/polynomial/multi~polynomialsequence~.py     Killed due to abort

  src/sage/rings/polynomial/pbori.pyx                        Killed due to abort

  src/sage/rings/polynomial/polynomial~ringconstructor~.py   Killed due to abort

  src/sage/sat/solvers/dimacs.py                             Killed due to abort

  src/doc/en/reference/sat/index.rst                         Killed due to abort

  src/sage/sat/boolean~polynomials~.py                       Killed due to abort

Details in SageTstAborts (attached).

Looking at those details, in all but one case, the abort is a "length error":

  src/sage/crypto/mq/sr.py                                   terminate called after throwing an instance of \'std::length~error~\'

  src/sage/rings/polynomial/multi~polynomialsequence~.py     terminate called after throwing an instance of \'std::length~error~\'

  src/sage/rings/polynomial/pbori.pyx                        terminate called after throwing an instance of \'std::bad~alloc~\'

                                                             terminate called recursively

  src/sage/rings/polynomial/polynomial~ringconstructor~.py   terminate called after throwing an instance of \'std::length~error~\'

  src/sage/sat/solvers/dimacs.py                             terminate called after throwing an instance of \'std::length~error~\'

  src/doc/en/reference/sat/index.rst                         terminate called after throwing an instance of \'std::length~error~\'

  src/sage/sat/boolean~polynomials~.py                       terminate called after throwing an instance of \'std::length~error~\'

The exception in pbori.pyx, seems to be multiple errors.

I suppose tat something may be broken in my setup. Hints about how to unhose my system are welcome...

HTH,

[Emmanuel Charpentier]

Forgot to specify that these results were for ptestalllong, which is now pertinent for Python 3-based Sage...

Incidentally, testing on another machine a patch for ticket Trac#28534 (which needed a large recompilation), ptest-python3 gives only the well-known failure related to FLINT. Currently running ptestalllong on this other machine.

[Emmanuel Charpentier]

On the second machine, upgrading a Python 3-based Sage to 8.9.rc1, adding Trac#28524 and ptestalllong'ing it gives the same failures as previously reported for 8.9.rc0, i. e. permanent failures for eclib/interface, Flint, and gap_packages. No killed tests.

My tentative conclusion is that I may have inadvertently hosed the first machine's Sage installation. Will rebuild it from scratch and report the results (but not for the next day...).

HTH,

[Markus Wageringel]

With Python 3 on CentOS, all tests from ptestlong passed.

With Python 2, I got a (transient) timeout in

    src/sage/parallel/map_reduce.py

There seems to be a race condition. It is difficult for me to reproduce. Has anyone seen this before? I am unsure whether this might have been caused by the recent changes to RecursivelyEnumeratedSet in #27967.

[Sébastien Labbé]

I am unsure whether this might have been caused by the recent changes to RecursivelyEnumeratedSet in #27967.

I don't think so as the changes in #27967 are not related to the search forest (and map_reduce.py) code.

 

[Samuel Lelièvre]

Thu 2019-09-26 06:21 UTC, Markus Wageringel:

>
> With Python 2, I got a (transient) timeout in
>
> src/sage/parallel/map_reduce.py
>
> There seems to be a race condition. It is difficult for me to reproduce.
> Has anyone seen this before?

Yes, it comes up once in a while. See previous mentions in sage-release:

https://groups.google.com/forum/#!searchin/sage-release/map_reduce%7Csort:date

and occasionally on sage-devel:

https://groups.google.com/forum/#!searchin/sage-devel/map_reduce%7Csort:date

[Emmanuel Charpentier]

After rebuilding after "make distclean" on the first machine (but without reinstallation of optional packages), "ptestalllong" gets me results close to those reported for 8.9.rc0, albeit with a slightly shorter list of transient failures:

src/sage/plot/animate.py                     

src/sage/combinat/matrices/hadamard_matrix.py

src/sage/misc/latex.py                       

src/sage/misc/persist.pyx                    

src/sage/coding/databases.py                 

src/sage/repl/load.py                        

src/sage/databases/findstat.py               

src/sage/misc/remote_file.py                 

src/sage/combinat/designs/ext_rep.py         

src/sage/interfaces/magma_free.py            

src/sage/interfaces/mathematica.py           

Sorry for the noise... HTH,

Le mercredi 25 septembre 2019 14:45:00 UTC+2, Emmanuel Charpentier a écrit :

[Samuel Lelièvre]

Thanks Volker for releasing Sage 8.9.rc1.

Any chance to include #28041 in Sage 8.9.rc2?
This would help with continuous integration.
https://trac.sagemath.org/ticket/28041

[Volker Braun]

I think rc1 is going to be equal to the release unless something happens by tomorrow or so... But I'll be happy to merge it in 9.0.beta0

[Samuel Lelièvre]

Le sam. 28 sept. 2019 à 00:03, Volker Braun:

>
> I think rc1 is going to be equal to the release unless something
> happens by tomorrow or so... But I'll be happy to merge it in 9.0.beta0

Alright. Merging a ticket that changes the Makefile
might fit better in a beta0 than in an rc2 anyway.
Good to know Sage 8.9 should be out soon,
I'm excited at all the progress that got into it.
Thanks for all your work preparing the releases.

[Justin C. Walker]

> On Sep 24, 2019, at 16:15 , Volker Braun <vbrau...@gmail.com> wrote:
>
> As always, you can get the latest beta version from the "develop" git branch. Alternatively, the self-contained source tarball is at http://www.sagemath.org/download-latest.html

Built from a fresh clone of the develop branch, on two macOS systems: 10.13.6 (2017 iMac Pro, 18-core Xeon W), and 10.14.6 (2017 MBP, Quad-core Core i7).

No problems with the build on either system.

Testing (‘ptestlong’) on 10.13.6 showed no problems. On 10.14.6, there was one failure:
sage -t --long --warn-long 72.6 src/sage/libs/singular/polynomial.pyx # 1 doctest failed

Justin

--
Justin C. Walker, Curmudgeon at Large
Institute for the Absorption of Federal Funds
-----------
I'm beginning to like the cut of his jibberish.
-----------