[Debian-science-sagemath] Sage 8.0 status (was Re: Sage 7.6 upload for RC bugfix and GAP 4.8.7 in unstable)

Fri Aug 4 13:16:56 UTC 2017

On 07/28/2017 03:06 PM, Jerome BENOIT wrote:
>
> On 28/07/17 17:53, Tobias Hansen wrote:
>> On 07/28/2017 02:38 PM, Jerome BENOIT wrote:
>>> On 28/07/17 17:28, Tobias Hansen wrote:
>>>> On 07/28/2017 02:24 PM, Jerome BENOIT wrote:
>>>>> On 24/07/17 19:28, Jeroen Demeyer wrote:
>>>>>> On 2017-07-24 17:17, Jerome BENOIT wrote:
>>>>>>> I think that the trouble comes from cysignals.
>>>>>> Do you have logs or anything that you want me to have a look at?
>>>>> At the time of writing, the cysignals package is in the tube (unstable):
>>>>> the doctest issue was fixed, it was merely a packaging mistake.
>>>>>
>>>>> Enjoy,
>>>>> Jerome
>>>> Hi,
>>>>
>>>> won't that break sagemath 7.6? At least it declares that it does. ;P
>>> Oops, I was no aware of that.
>>> Sorry for the mistake, Jerome
>> Nevermind. We should all get used to uploading stuff for the next Sage
>> version to experimental by default (unless we know unstable is ok).
>>
>> Now we can either live with the fact that sagemath will be uninstallable
>> in unstable until we upload 8.0 or we need to check if sagemath 7.6
>> works with the newer cysignals and then remove the "Breaks" from
>> cysignals. I'd say let's first wait and see how our sage 8.0 package
>> progresses. cypari2 is still in NEW.
>>
>> At least now we will see if the "Breaks" really prevents cysignals from
>> migrating to testing. (So please don't remove it unless we know that
>> sage 7.6 works with this cysignals version.)
> Ok, I guess you meant that we let it as it.
>
> Meanwhile, I tried to reproduce the fpylll issue on my box:
> 1] with cysignals 1.3.2+ds-1 (stable/testing), it could be built ;
> 2] with cysinagls 1.6.4+ds- and 1.6.5+ds-s, it cannot be built.
>
> In short, the issue remains but we know now it unlikely caused by cysignals.
>
> I may work on it later. Any hint is welcome, meanwhile.
>
> Cheers,
> Jerome

Hi,

good work on finding the fpylll issue! Now the main holdup for sage 8.0
is still cypari2 being stuck in NEW. I wrote on June 21 to
ftpmaster at debian.org and August 2 directly to the ftpmaster who helped
us get sagemath through NEW in time for stretch. Let's see if it helps.

I'm also not totally sure if the sagemath packages can be built at the
moment. When I build in a schroot with sage deps installed with
DEB_BUILD_OPTIONS=parallel=5 fakeroot debian/rules binary
after the build is almost completed, including running the tests, it
later tries again to run the test suite and fails (I attached the end of
the console output). I wanted to try if this also happens with sbuild
once the fpylll problem is fixed.

However I will be on vacation the next three weeks.

The accidental upload of cysignals 1.6.5 to unstable is now a RC bug
(#870688). Not sure if we should fix it by downgrading cysignals,
patching sage 7.6 or just let the bug sit until we upload sage 8.0. At
least the "Breaks: sagemath (<< 8.0~)" does indeed prevent cysignals
from migrating to testing, which is good.

Best,
Tobias
-------------- next part --------------
$DEB_BUILD_OPTIONS=parallel=5 fakeroot debian/rules binary
...
I: dh_python2 fs:322: renaming matroid.so to matroid.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming set_system.so to set_system.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming union_matroid.so to union_matroid.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming unpickling.so to unpickling.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming probability_distribution.so to probability_distribution.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming channels.so to channels.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming algebra_elements.so to algebra_elements.x86_64-linux-gnu.so
I: dh_python2 fs:322: renaming paths.so to paths.x86_64-linux-gnu.so
dh_numpy --package sagemath # stop lintian complaining at us
make[2]: Leaving directory '/home/thansen/src/sage/sagemath/sagemath'
   dh_python2 -Nsagemath
   dh_lintian
   dh_perl
   dh_link
   dh_strip_nondeterminism
   debian/rules override_dh_compress
make[2]: Entering directory '/home/thansen/src/sage/sagemath/sagemath'
dh_compress -X.pdf -X.pickle -X.doctree
make[2]: Leaving directory '/home/thansen/src/sage/sagemath/sagemath'
   dh_fixperms
   dh_missing
make[1]: Leaving directory '/home/thansen/src/sage/sagemath/sagemath'
   dh_testdir
   dh_update_autotools_config
   dh_autoreconf
   dh_auto_configure
   debian/rules override_dh_auto_build-arch
make[1]: Entering directory '/home/thansen/src/sage/sagemath/sagemath'
dh override_dh_auto_build-arch --with=python2,sphinxdoc
make[1]: Leaving directory '/home/thansen/src/sage/sagemath/sagemath'
   debian/rules override_dh_auto_build-indep
make[1]: Entering directory '/home/thansen/src/sage/sagemath/sagemath'
# Sometimes, parts of the docbuild fail with MemoryError or OSError but
# Sage doesn't detect this. Here we detect it and fail the build if so.
# Also filter out some false positives that are not hard errors.
! (grep Error sage/logs/dochtml.log | grep -v "citation not found: .*Error.*")
dh override_dh_auto_build-indep --with=python2,sphinxdoc
make[1]: Leaving directory '/home/thansen/src/sage/sagemath/sagemath'
   debian/rules override_dh_auto_test
make[1]: Entering directory '/home/thansen/src/sage/sagemath/sagemath'
cd sage && src/bin/sage-starts

Testing that Sage starts...
[2017-07-30 16:13:19] SageMath version 8.0, Release Date: 2017-07-21
Traceback (most recent call last):
  File "/home/thansen/src/sage/sagemath/sagemath/sage/src/bin/sage-eval", line 4, in <module>
    from sage.all import *
ImportError: No module named sage.all
Sage failed to start up.
Please email sage-devel (http://groups.google.com/group/sage-devel)
explaining the problem and send the log file
  /home/thansen/src/sage/sagemath/sagemath/sage/logs/start.log
Describe your computer, operating system, etc.
debian/rules:113: recipe for target 'override_dh_auto_test' failed
make[1]: *** [override_dh_auto_test] Error 1
make[1]: Leaving directory '/home/thansen/src/sage/sagemath/sagemath'
debian/rules:47: recipe for target 'binary' failed
make: *** [binary] Error 2