upgrade to pynac-0.3.5

[rwst]

We skip 0.3.3 (#18155) which couldn't be used because of a bug showing on OSX.

https://drive.google.com/folderview?id=0B4PmRyK1JXgHfmJmbWZ5NnVtMnNITTVHQ2hCci1WVnhLbC1PeC1YbzRQLUpidU9yY0x3Qjg&usp=sharing

* fixed the series expansion of zeta(s) around 1
* print_latex of wildcards
* clear the libtool versioning scheme
* sync with current GiNaC as much as possible
* revert GiNaC commit interfering with Pynac print order
* fix code that produced warnings

Upgrade will require a few doctests to adapt, see https://groups.google.com/d/topic/sage-devel/hwQXxFmutcU/discussion

Dependent tickets: #15047, #15846

Component: packages: standard

Author: Ralf Stephan

Branch/Commit: c5ac7dc

Reviewer: Benjamin Hackl

Issue created by migration from https://trac.sagemath.org/ticket/18237

[rwst]

Branch: u/rws/pynac035

[rwst]

New commits:

835d82b	pynac-0.3.5
2db2621	18155: upgrade to pynac-0.3.3
216e29c	18155: changed chksum for tarball made with make dist
839c49f	18155: file change
c5ac7dc	fix doctests

[rwst]

Author: Ralf Stephan

[rwst]

Commit: c5ac7dc

[behackl]

comment:3

On it, currently doing some more doctesting on Arch Linux (kernel 3.19.3-3, i7, 12GB RAM); will be finished soon.

Meanwhile: I can confirm that the bug with the unstable output of integrate(sec(x), (x, 0, pi/8)) on OSX (see pynac/pynac#18) is gone. Thanks to Volker for access to the OSX buildslave!

[behackl]

comment:4

MAKE="make -j4" make ptestlong gives

----------------------------------------------------------------------
sage -t --long --warn-long 81.8 src/sage/combinat/dyck_word.py  # 1 doctest failed
----------------------------------------------------------------------

The doctest fails with

sage -t --long --warn-long 81.8 src/sage/combinat/dyck_word.py
**********************************************************************
File "src/sage/combinat/dyck_word.py", line 3501, in sage.combinat.dyck_word.DyckWords_size.__init__
Failed example:
    TestSuite(DyckWords(4,2)).run()
Expected nothing
Got:
    Failure in _test_enumerated_set_iter_cardinality:
    Traceback (most recent call last):
      File "/home/behackl/Programming/sage-6.7.beta1/local/lib/python2.7/site-packages/sage/misc/sage_unittest.py", line 282, in run
        test_method(tester = tester)
      File "/home/behackl/Programming/sage-6.7.beta1/local/lib/python2.7/site-packages/sage/categories/finite_enumerated_sets.py", line 462, in _test_enumerated_set_iter_cardinality
        tester.assert_(isinstance(card, Integer))
      File "/home/behackl/Programming/sage-6.7.beta1/local/lib/python/unittest/case.py", line 422, in assertTrue
        raise self.failureException(msg)
    AssertionError: False is not true
    ------------------------------------------------------------
    The following tests failed: _test_enumerated_set_iter_cardinality
**********************************************************************
1 item had failures:
   1 of   2 in sage.combinat.dyck_word.DyckWords_size.__init__
    [574 tests, 1 failure, 2.98 s]

This happened two times. However, when testing the respective file manually:

sage -t --long --warn-long 81.8 src/sage/combinat/dyck_word.py
    [574 tests, 0.95 s]
----------------------------------------------------------------------
All tests passed!
----------------------------------------------------------------------

So this only seems to occur for make ptestlong.

Additionally, as I've not run make ptestlong on a clean install of 6.7.beta1 yet, I'm not even sure if this is related to Pynac at all. I'll do so and report back...

[behackl]

comment:5

See https://groups.google.com/forum/?fromgroups=#!topic/sage-devel/7BQPmEgfA8M -- until this is resolved, I can't review the new Pynac version reliably.

[sagetrac-jkeitel]

comment:6

I had the same doctest fail today on a clean 6.7beta1 install and, just like you, could not reproduce it when running this specific test again on the file, so I doubt that this is related to this ticket.

[behackl]

comment:7

I'm also quite sure that it has nothing to do with this ticket; especially after Vincent confirmed that this is happening for him too on both 6.7.beta0 and 6.7.beta1. This discussion, however, belongs to the sage-devel thread.

Still, I don't like the idea of giving this positive_review when I don't have a passing ptestlong, regardless of why the doctest is failing. Currently, I'm compiling sage on two more machines; one of those should enable me to review this.

[behackl]

Reviewer: Benjamin Hackl