BioPython: Different test suite results when debuild and pdebuild
Hi,
I have started anyother try to finalise BioPython which is prepared at
Vcs-Svn:
svn://anonscm.debian.org/debian-med/trunk/packages/python-biopython/trunk/
I try to run as much tests of the test suite as possible in the build
process and I also added autopkgtest based on upstream's unit tests.
Strangely enough I have some different results when building with
debuild on my local *testing* installation and when building the package
in pbuilder *unstable* chroot when it comes to Python 3 (3.3 as well as
3.4).
Usually this is due to some missing build depends but I have no idea
which one this might be since all these errors are about
UnicodeDecodeError. You might hopefully be able to reproduce this
by building the package using debuild which should work and compare
it by using pdebuild which shows some errors of type:
==
ERROR: test_ClustalOmega_tool
--
Traceback (most recent call last):
File run_tests.py, line 407, in runTest
suite = unittest.TestLoader().loadTestsFromName(name)
File /usr/lib/python3.3/unittest/loader.py, line 96, in loadTestsFromName
module = __import__('.'.join(parts_copy))
File
/tmp/buildd/python-biopython-1.63/.pybuild/pythonX.Y_3.3/build/Tests/test_ClustalOmega_tool.py,
line 29, in module
output = getoutput(clustalo --help)
File /usr/lib/python3.3/subprocess.py, line 703, in getoutput
return getstatusoutput(cmd)[1]
File /usr/lib/python3.3/subprocess.py, line 684, in getstatusoutput
data = check_output(cmd, shell=True, universal_newlines=True, stderr=STDOUT)
File /usr/lib/python3.3/subprocess.py, line 578, in check_output
output, unused_err = process.communicate(timeout=timeout)
File /usr/lib/python3.3/subprocess.py, line 907, in communicate
stdout = _eintr_retry_call(self.stdout.read)
File /usr/lib/python3.3/subprocess.py, line 479, in _eintr_retry_call
return func(*args)
File /usr/lib/python3.3/encodings/ascii.py, line 26, in decode
return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 170:
ordinal not in range(128)
==
I went via pbuilder hook into the schroot and tried to install
additional packages (for instance python3-unidecode which sounded
promising) but running the test manually via
cd /tmp/buildd/python-biopython-1.63/Tests
python3.3 run_tests.py -v test_ClustalOmega_tool
(while
python run_tests.py -v test_ClustalOmega_tool
is running nicely).
So I wonder whether there is some missing Build-Depends which is
usually the case if it runs in debuild but not pdebuild or if there
is a new problem with encodings in unstable which simply does not
occure in testing.
Any help is welcome
Andreas.
--
http://fam-tille.de
debuild_vs_run-unit-test.diff.gz
Description: Binary data
Re: BioPython: Different test suite results when debuild and pdebuild
Hi Andreas, File /usr/lib/python3.3/encodings/ascii.py, line 26, in decode return codecs.ascii_decode(input, self.errors)[0] That's may be because pdebuild sets LC_ALL=C (see #376404). Python3 uses it to determine how it needs to print unicode. Here it tries to decode it as ascii but this fails for characters outside of 0-127. It already bit me once as the results are different in pbuilder and sbuild. A solution is to explicitly set LC_ALL to tell dh_auto_tests that the terminal is able to interpret utf8: LC_ALL=C.UTF-8 dh_auto_tests -- ... HTH, -- Etienne Millon -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140425085733.GA6329@klow
Re: BioPython: Different test suite results when debuild and pdebuild
On Fri, Apr 25, 2014 at 10:57:33AM +0200, Etienne Millon wrote: LC_ALL=C.UTF-8 dh_auto_tests -- ... Works! Thanks a lot for this really helpful hint Andreas. -- http://fam-tille.de -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140425094523.gb13...@an3as.eu
BioPython: Some Python modules not found in tests at package build time
Hi, as I wrote in my last mail I have started anyother try to finalise BioPython which is prepared at Vcs-Svn: svn://anonscm.debian.org/debian-med/trunk/packages/python-biopython/trunk/ I hope the final problem with the tests is that I get output like test_GraphicsBitmaps ... skipping. Install PIL (Python Imaging Library) if you want to create bitmaps with Bio.Graphics. despite the required Python modules are installed. Interestingly enough if I manually run the test suite I get a different warning message for test_GraphicsBitmaps: test_GraphicsBitmaps ... Warn: Can't find .pfb for face 'Times-Roman' skipping. Check the fonts needed by ReportLab if you want bitmaps from Bio.Graphics Can't setFont(Times-Roman) missing the T1 files? Originally type 'exceptions.TypeError': makeT1Font() argument 2 must be string, not None My conclusion is that when not running the build (here it is not relevant whether I'm doing debuild or pdebuild!) PIL is not found but in case I run the test suite manually afterwards it is found an some other problem occures (hints how to get the .pfb files are for sure welcome as well). Any idea what the problem might be? Kind regards Andreas. -- http://fam-tille.de -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140425130542.ge13...@an3as.eu
Re: BioPython: Some Python modules not found in tests at package build time
Hi, On Fri, Apr 25, 2014 at 03:05:42PM +0200, Andreas Tille wrote: Hi, as I wrote in my last mail I have started anyother try to finalise BioPython which is prepared at Vcs-Svn: svn://anonscm.debian.org/debian-med/trunk/packages/python-biopython/trunk/ I hope the final problem with the tests is that I get output like test_GraphicsBitmaps ... skipping. Install PIL (Python Imaging Library) if you want to create bitmaps with Bio.Graphics. despite the required Python modules are installed. Interestingly enough if I manually run the test suite I get a different warning message for test_GraphicsBitmaps: test_GraphicsBitmaps ... Warn: Can't find .pfb for face 'Times-Roman' skipping. Check the fonts needed by ReportLab if you want bitmaps from Bio.Graphics Can't setFont(Times-Roman) missing the T1 files? Originally type 'exceptions.TypeError': makeT1Font() argument 2 must be string, not None My conclusion is that when not running the build (here it is not relevant whether I'm doing debuild or pdebuild!) PIL is not found but in case I run the test suite manually afterwards it is found an some other problem occures (hints how to get the .pfb files are for sure welcome as well). Well, this was rather me forgetting a python-imaging dependency - I (wrongly) assumed that python-pil would be sufficient. So any idea how to get .pfb for face 'Times-Roman' found by reportlab Kind regards Andreas. -- http://fam-tille.de -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140425133221.gg13...@an3as.eu
Re: BioPython: Some Python modules not found in tests at package build time
[Debian Python Modules Team python-modules-t...@lists.alioth.debian.org in CC since it maintains python-reportlab] Hi again, On Fri, Apr 25, 2014 at 03:32:21PM +0200, Andreas Tille wrote: test_GraphicsBitmaps ... Warn: Can't find .pfb for face 'Times-Roman' skipping. Check the fonts needed by ReportLab if you want bitmaps from Bio.Graphics Can't setFont(Times-Roman) missing the T1 files? Originally type 'exceptions.TypeError': makeT1Font() argument 2 must be string, not None Well, this was rather me forgetting a python-imaging dependency - I (wrongly) assumed that python-pil would be sufficient. So any idea how to get .pfb for face 'Times-Roman' found by reportlab Some websearch leaded me to #174639 and thus I applied the following patch to python-reportlab: Author: Andreas Tille ti...@debian.org Last-Update: Fri, 25 Apr 2014 16:02:07 +0200 Description: Fix TTFSearchpath (see #174639) --- a/src/reportlab/rl_settings.py +++ b/src/reportlab/rl_settings.py @@ -122,6 +122,7 @@ T1SearchPath = ( . # places to look for TT Font information TTFSearchPath = ( +'/usr/share/fonts/truetype/', # Here are Debian's TTF fonts kept 'c:/winnt/fonts', 'c:/windows/fonts', '/usr/lib/X11/fonts/TrueType/', Unfortunately the result remained the same. I somehow came to the conclusion that the problem is caused by a broken reportlab package but it seems my patch did not solved the issue. Any better idea? I do not really want to reopen #174639 / file a new bug before I'm really sure that this is the cause of the problem. Kind regards Andreas. -- http://fam-tille.de -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140425140841.gh13...@an3as.eu
Re: RFS: python-ebooklib
Hi Zygmunt, I think you are quite welcome to join DPMT https://wiki.debian.org/Teams/PythonModulesTeam/HowToJoin Thanks, I have sent a request via the web interface. I have also helped upstream create some Sphinx documentation and re-uploaded the package to include it. Cheers! Daniel On Thu, Apr 24, 2014 at 6:34 PM, Daniel James dan...@64studio.com mailto:dan...@64studio.com wrote: Hello, I am a non-DD looking for sponsorship of (or any helpful hints about) my first upload to Debian: http://mentors.debian.net/package/python-ebooklib EbookLib is a Python library for managing EPUB2/EPUB3 and Kindle files. It's capable of reading and writing EPUB files (Kindle support is under development). Upstream homepage is https://github.com/aerkalov/ebooklib I have had some general feedback from the Debian Mentors list and believe the package is now ready for review. I'd be happy to join a packaging team should that be possible. My username on Alioth is dhj-guest, if that helps :-) Thanks! Daniel -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/535a7e23.5010...@64studio.com
InVesalius - arch-dependent-file-in-usr-share
Hi all, I'm one of developer of InVesalius [1], an open source medical software. InVesalius is already packaged to Debian thanks to Debian med. Now I'm updating the package [2] to the last release of InVesalius. Until the penultimate version, InVesalius was developed exclusively in Python and all its source files in the Debian package were kept inside the /usr/share/invesalius folder. In the last version we started to use Cython in one of the modules we developed, so now we have a compiled file. In the part of the package I break the InVesalius in two parts: * invesalius - where are kept the python files (arch-independent) * invesaliu-bin - where this cython compiled file is kept All files from both packages are kept inside /usr/share/invesalius. But, as indicated by the lintian warning, /usr/share is reserved for architecture-independent files. So invesalius-bin isn't following the rule. I'm thinking to create a patch to update the setup.py to make it installs the compiled file inside the Python dist-packages folder and create other package to patch InVesalius to make it imports this module from the system dist-packages. But I don't know if it's the better strategy to fix this problem or if there are better ways of doing this. What do you suggest me? Many thanks! [1] - https://github.com/invesalius/invesalius3 [2] - http://anonscm.debian.org/viewvc/debian-med/trunk/packages/invesalius/trunk/debian/
