[Debichem-devel] updated python-ase / gpaw

Marcin Dulak marcin.dulak at gmail.com
Sat Jul 25 21:35:45 UTC 2015


Hi,

On Fri, Jul 24, 2015 at 6:14 PM, Andreas Tille <andreas at an3as.eu> wrote:

> Hi Graham,
>
> On Fri, Jul 24, 2015 at 05:53:20PM +0200, Graham Inggs wrote:
> >
> > On 24/07/2015 16:53, Andreas Tille wrote:
> > >I noticed that ase also should work with Python 3 and thus added a
> > >Python 3 package.
> >
> > OK.  I had read that Python3 wasn't supported yet [1]:
>
>
python3 should be fully supported by ASE.
The upstream documentation is out-of-date.


> I simply did:
>
> $ grep 3\.4 * 2>/dev/null
> PKG-INFO:Classifier: Programming Language :: Python :: 3.4
> setup.py:          'Programming Language :: Python :: 3.4',
>
> which seems to be a sign for 3.4 support.
>
> > The following packages are required for basic ASE functionality:
> >
> >     1. Python2 version 2.6 or newer. Python3 is not supported yet.
> >     2. NumPy.
> >
> > >   Unfortunately 1 of 142 fails.  Would you mind having a look into
> this?
> >
> > Weird, I tried building and it passed all the tests (although that was
> with
> > Python 3.4, not 3.5 as is in unstable).
>
> I was building in an unstable chroot which obviously used Python 3.4
> (not 3.5).  The log said:
>
> ...
> tasks/dcdft.py (ScriptTestCase) ... skipped ok
> tasks/gbrv.py (ScriptTestCase) ... ERROR
> tasks/htb.py (ScriptTestCase) ... ok
> tasks/mmbjmp.py (ScriptTestCase) ... ok
> turbomole/turbomole_H2.py (ScriptTestCase) ... skipped ok
> turbomole/turbomole_h3o2m.py (ScriptTestCase) ... skipped ok
> vasp/vasp_Al_volrelax.py (ScriptTestCase) ... skipped ok
> vasp/vasp_co.py (ScriptTestCase) ... skipped ok
> COCu111.py (ScriptTestCase) ... Test suite failed
> ok
>
> ======================================================================
> ERROR: tasks/gbrv.py (ScriptTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/tmp/buildd/python-ase-3.9.1.4567/ase/test/__init__.py", line 50,
> in testfile
>     exec(compile(fd.read(), self.filename, 'exec'),
>   File "/usr/lib/python3.4/encodings/ascii.py", line 26, in decode
>     return codecs.ascii_decode(input, self.errors)[0]
> UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 488:
> ordinal not in range(128)
>
> ----------------------------------------------------------------------
> Ran 142 tests in 46.282s
>
> FAILED (errors=1)
> E: pybuild pybuild:256: test: plugin custom failed with: exit code=1:
> python3.4 /tmp/buildd/python-ase-3.9.1.4567/setup.py test
> dh_auto_test: pybuild --test -i python{version} -p 3.4 --dir . returned
> exit code 13
>
> ...
>
> So if you build in an unstable chroot you do not get this error?
>

this is interesting. I have the same problem with python-ase and python3
on Fedora - can't reproduce this error on the command line in latest Fedora
Rawhide.
If anybody finds a way to reproduce the error (and possibly a fix) - please
report to upstream ASE.


>
> > I did get hundreds of Lintian warnings though, e.g.
> > W: python-ase source: binaries-have-file-conflict python-ase python3-ase
> > usr/bin/ase-build
>
> Well, the build at my side did not finalise so I can not look into the
> binary but it seems that the package does install not only binaries but
> also has some executables in /usr/bin.  Setup.py simply installs these
> for each run and so you get the files in /usr/bin in both packages which
> will lead to a conflict between both packages.
>

the scripts under tools are not python2/3 aware, they just start
with a "#!/usr/bin/env python" shebang.
I'm not sure how Debian handles switching between python 2 and 3.
The scripts shipping with python3 package can be just renamed to have 3
appended to the names, e.g. ase-build3. This is what I do on Fedora.

>
> I had a similar situation in DendroPy
>    git://anonscm.debian.org/debian-med/python-dendropy.git
> and solved this by a separate package delivering the tools in /usr/bin.
> I decided for the Python 3 version of the tool.
>
> > Is this normal for packages that produce Python2 and Python3 binaries?
>
> This is normal for all packages installing the same file name.
>
> > What about the recommended packages; python-gtk2, python-matplotlib and
> > python-scipy?
> > Should those become python3-matplotlib and python3-scipy?
>
> Sure.  I missed this piece.
>
> > It seems that
> > there is no python3-gtk2.
>
>
ASE does not support gui for python3 for that reason.
Therefore ase-gui should not be packaged for the python3 package.

Marcin


> I'm not sure.  May be you clarify this with
> debian-python at lists.debian.org.
>
> > I must admit I am not comfortable with this change you have made.
>
> If you want to push on a quick update I'm fine with reverting this
> change.  However, at some point in time you will need to do it anyway.
> Its youe choice when you want to do this.
>
> Kind regards
>
>         Andreas.
>
> --
> http://fam-tille.de
>
>
> --
> To UNSUBSCRIBE, email to debian-science-request at lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact
> listmaster at lists.debian.org
> Archive: https://lists.debian.org/20150724161403.GK26775@an3as.eu
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/debichem-devel/attachments/20150725/5f0b7004/attachment.html>


More information about the Debichem-devel mailing list