doctest failures with Sage 4.3.4.alpha0

[mvngu]

From  sage-devel:

* The following tests failed on sage.math:

sage -t  -long local/lib/python2.6/site-packages/sagenb-0.7.5.1-py2.6.egg/sagenb/notebook/interact.py # 1 doctests failed
sage -t  -long local/lib/python2.6/site-packages/sagenb-0.7.5.1-py2.6.egg/sagenb/misc/sageinspect.py # 1 doctests failed
sage -t  -long devel/sage/sage/categories/finite_semigroups.py # 2 doctests failed
sage -t  -long devel/sage/sage/categories/examples/finite_semigroups.py # 1 doctests failed 

The doctest

sage -t  -long devel/sage/sage/categories/finite_semigroups.py # 2 doctests failed

also failed on SPARC Solaris 10 (the machine t2.math). The failure uncovered another bug relating to kpsewhich. See ticket #8445.

[jhpalmieri]

I've tried to address the problem with sageinspect.py in ticket #8324.

[jhpalmieri]

Note that the problem with sagenb/notebook/interact.py should be fixed by the sagenb patch at ticket #5601.

[hivert]

• The following tests failed on sage.math:

sage -t  -long devel/sage/sage/categories/finite_semigroups.py # 2 doctests failed
sage -t  -long devel/sage/sage/categories/examples/finite_semigroups.py # 1 doctests failed 

Those last two should be fixed by the submitted patch. A better fix together with improvements is being written on sage-combinat but I don't know how long finishing those improvements will take. So please review the patch.

By the way, should'nt I create a ticket for this patch ? It is certainly orthogonal to the two other one doctest failure.

Cheers,

Florent

[mhampton]

Since the other issues have there own tickets I don't think another one is necessary.

There is one doctest failure still:

File "/Volumes/E/sage-4.3.4.alpha0/devel/sage-t1/sage/categories/finite_semigroups.py", line 229:
    sage: sorted(S.j_transversal_of_idempotents())
Expected:
    ['a', 'ab', 'ac', 'acb', 'b', 'bc', 'c']
Got:
    ['a', 'ab', 'ac', 'acb', 'b', 'c', 'cb']
*********************************************

which is not covered by the patch. Seems like maybe the expected string is wrong, unless the sorted function has varying behavior.

[hivert]

Replying to mhampton:

There is one doctest failure still:

File "/Volumes/E/sage-4.3.4.alpha0/devel/sage-t1/sage/categories/finite_semigroups.py", line 229:
    sage: sorted(S.j_transversal_of_idempotents())
Expected:
    ['a', 'ab', 'ac', 'acb', 'b', 'bc', 'c']
Got:
    ['a', 'ab', 'ac', 'acb', 'b', 'c', 'cb']
*********************************************

which is not covered by the patch. Seems like maybe the expected string is wrong, unless the sorted function has varying behavior.

I'm not sure to understand:

• why I missed this one;
• why the result of sorted changed between 4.3.3 and 4.3.4;
• why despite of this change it seems to be stable on both version...

Anyway the newly uploaded patch passed the tests 10 times in a raw on the machine sage. Maybe it could be worth checking of it is also stable on other architectures.

Sorry for the mess and please review.

[mhampton]

All tests in categories passed on a 10.6 mac and on 64-bit Ubuntu 9.10, so I think I can give this a positive review.

[mpatel]

SageNB 0.7.5.2 includes fixes for the first two failures in the description (see #8435).

An important note: It's best to test SageNB files with -force_lib, e.g.,

sage -t -long -force_lib local/lib/python2.6/site-packages/sagenb-0.7.5.1-py2.6.egg/sagenb/notebook/interact.py

Running sage -t -sagenb or make test and its friends will set it implicitly. If the option is set, SAGE_LOCAL/bin/sage-doctest will treat the file(s) as Sage library code, even if it does not live under SAGE_ROOT/devel. In particular, sage-doctest will not attempt the equivalent of from filebase import *, which can raise false errors or cause segfaults.

I suppose I should have called it -library_code, instead.

Please remember to fillin

[drkirkby]

I'm not at home, and can't very easily start applying patches, but I can confirm I get the same results, based on Sage 4.3.4.alpha0 but with a few Solaris-specific patches on Solaris 10 (SPARC)

**********************************************************************
File "/export/home/drkirkby/32/sage-4.3.4.alpha0/devel/sage/sage/categories/fini
te_semigroups.py", line 232:
    sage: sorted(S.j_transversal_of_idempotents())
Expected:
    ['a', 'ab', 'ac', 'acb', 'b', 'bc', 'c']
Got:
    ['a', 'ab', 'ac', 'acb', 'b', 'c', 'cb']
**********************************************************************
File "/export/home/drkirkby/32/sage-4.3.4.alpha0/devel/sage/sage/categories/fini
te_semigroups.py", line 198:
    sage: S.j_classes()
Expected:
    [['acb', 'cab', 'bca', 'abc', 'bac', 'cba'], ['ac', 'ca'],
    ['ab', 'ba'], ['bc', 'cb'], ['a'], ['c'], ['b']]
Got:
    [['a'], ['acb', 'cba', 'bca', 'abc', 'bac', 'cab'], ['ac', 'ca'], ['ab', 'ba
'], ['cb', 'bc'], ['c'], ['b']]
**********************************************************************
2 items had failures:

So the problem is occuring on more that one architecuture. I hope to try the patch within the next 12 hours.