[Biopython-dev] Python 2.7 failure from test_Tutorial.py, was: Compiling Biopython on 64 bit Windows

[Peter Cock]

Hi Manlio,

Moving on to Python 2.7, apart from the TogoWS issue,
http://lists.open-bio.org/pipermail/biopython-dev/2014-April/011259.html
your system showed one other failure:

======================================================================
ERROR: test_doctests (test_Tutorial.TutorialTestCase)
Run tutorial doctests.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "test_Tutorial.py", line 167, in test_doctests
ValueError: 1 Tutorial doctests failed: test_from_line_04424

----------------------------------------------------------------------


You should be able to rerun the tutorial tests alone (from memory
only if Biopython is installed) with just:

c:\python27\python test_Tutorial.py

This is from the following doctest in the Tutorial.tex file,

%This example fails under PyPy 2.0, https://bugs.pypy.org/issue1546
%doctest examples lib:numpy
\begin{verbatim}
>>> import numpy as np
>>> from Bio import AlignIO
>>> alignment = AlignIO.read("PF05371_seed.sth", "stockholm")
>>> align_array = np.array([list(rec) for rec in alignment], np.character)
>>> align_array.shape
(7, 52)
\end{verbatim}

Could you try running that example "by hand" and seeing what
happens? Note the doctest framework is very very fragile and
the slightest output difference will break a test.

Thanks,

Peter

On Fri, Apr 18, 2014 at 7:40 PM, Manlio Calvi <manlio.calvi at gmail.com> wrote:
> And at last of this forward I'm posting the build and tests logs (more a
> dump than a log)  for those who want peek into it.
> As you'll see the only consistent error was the TogoWS' XHTML error.
>
> Cheers,
> Manlio
>
>
>
> On Fri, Apr 18, 2014 at 8:35 PM, Manlio Calvi <manlio.calvi at gmail.com>wrote:
>
>>
>>
>> ---------- Forwarded message ----------
>> From: Peter Cock <p.j.a.cock at googlemail.com>
>> Date: Fri, Apr 18, 2014 at 12:01 PM
>> Subject: Re: Compiling Biopython on 64 bit Windows, was: Volunteer
>> buildslave machines?
>> To: Manlio Calvi <manlio.calvi at gmail.com>
>> Cc: Tiago Antao <tra at popgen.net>
>>
>>  Manlio
>>
>> Excellent work - well done :)
>>
>> We should add some of this to the "Installing from source on Windows"
>> section of the installation notes (which we also need to update regarding
>> newer versions of Python and which compiler works):
>>
>>
>> https://github.com/biopython/biopython/blob/master/Doc/install/Installation.tex
>>
>> Step 6 sounds familiar, I think there is a similar batch file to set the
>> environment variables for the 43bit MSVC compiler.
>>
>> Regarding step 7, Python should use separate subdirectoroes under
>> build/ for each version of Python, so that shouldn't be a needed.
>> However, if you do use the same folder some of our tests use
>> hard coded temporary filenames so it is best to only test one version
>> of Python at a time.
>>
>> Regarding step 8, normally on Windows I don't bother with setting the
>> $PATH environment variable and simply call the explicit version of
>> Python I am testing, e.g.
>>
>> C:\Python27\python.exe setup.py build
>> C:\Python27\python.exe setup.py test
>> C:\Python27\python.exe setup.py install
>>
>> (This is what the current buildbot instructions do where we
>> have automated the tests).
>>
>> As to what's text - let's try to fix these test failures :)
>> Can you post the test output please?
>>
>> TogoWS is an online service, so from time to time it does have
>> transient failures (so does the NCBI Entrez API). The fact this
>> broke under both Python 2.7 and 3.4 suggests this is not a
>> Python 3 issue.
>>
>> I am a little surprised you didn't see this problem with Python 3.4,
>> but Windows does unicode a little differently so that may be why?
>> https://github.com/biopython/biopython/issues/305
>>
>> How did you install git & clone Biopython? Specifically, what
>> is it doing about text files (i.e. Unix versus DOS/Windows
>> new lines). This was something which has caused test
>> failures in the past - perhaps some new cases have arisen.
>>
>> Thanks!
>>
>> Peter
>>
>>
>
> _______________________________________________
> Biopython-dev mailing list
> Biopython-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/biopython-dev
>