Menu
▾
▴
cclib-devel
|
From: Noel O'B. <bao...@gm...> - 2007-06-04 09:19:09
|
On 02/06/07, Karol Langner <kar...@kn...> wrote: > Hi, > > Since I haven't seen any debs of cclib on the web (I have seen queries about > it) and I have some experience in making them, I've decided to take a shot at > it. It would be nice to get cclib into the Debian packaging system, and from > what I gather it can be done pretty quickly. I need to find a bit of time to > tune things, but I made a quick build today - they install and seem to work > fine on my Debian etch installation. You can find all the related files here: > http://www.mml.ch.pwr.wroc.pl/langner/tmp/debs/ > > Notice that I also built a packages called 'cclib-data', which has the test > scripts and data files. At the moment, it installs into the same directory as > cclib, but cannot be accessed from Python. > > My question for you are: > > 1. Do you use cclib on Debian and would be willing to test these packages? No. > 2. Do you have colleagues that use cclib on Debian and would be willing to > test these packages? > > 3. Would you be interested in seeing debs of GaussSum and/or PyMOLyze? I'm not > aware if any exist already. GaussSum's already in there. > 4. A testing issue has occured to me while testing the packages. It should be > possible to run the test suite we have from the Python interpreter. I think > that adding this is a good idea for the next release, together with an option > in setup.py to install the test scripts and data files. What are you poinions > on this? Well, I think you should see how this is done for other packages first. The general recommendation is that "python setup.py test" should run the tests. But I think you are suggesting something beyond this. To be honest, I'm not very interested in getting cclib into any of the distributions. Of course, you are welcome to do it. However, since the API has not currently settled, you are going to have problems if some program, e.g. PyMOlyze/GaussSum, starts to depend on it. Basically, you will find it difficult ever to upgrade it if the API changes. Also, cclib has optional dependencies on openbabel, BioPytho, PyQuante (and pyvtk?), which will need to be sorted out. I don't mean to discourage you, but I don't know if it's a good thing to be stuck with a version of cclib in Debian that's 1.5 years out of date, especially when Python packages are pretty easy to install, and cclib has been changing a lot from one release to the next. I'd actually be more interested in getting a Python egg together for ez_setup.py/setuptools. I don't know what the current situation is with eggs and Debian though. > Cheers, > Karol > > -- > written by Karol Langner > Sat Jun 2 15:44:18 CEST 2007 > |
|
From: Karol L. <kar...@kn...> - 2007-06-04 11:03:23
Attachments:
cclib-0.7-py2.4.egg
|
On Monday 04 June 2007 11:19, Noel O'Boyle wrote:
> > 3. Would you be interested in seeing debs of GaussSum and/or PyMOLyze?
> > I'm not aware if any exist already.
>
> GaussSum's already in there.
Yes, now I see.
> > 4. A testing issue has occured to me while testing the packages. It
> > should be possible to run the test suite we have from the Python
> > interpreter. I think that adding this is a good idea for the next
> > release, together with an option in setup.py to install the test scripts
> > and data files. What are you poinions on this?
>
> Well, I think you should see how this is done for other packages
> first. The general recommendation is that "python setup.py test"
> should run the tests. But I think you are suggesting something beyond
> this.
Sure, "python setup.py test" is the standard. I was thinking of something like
numpy.test(), which runs the NumPy test suite for the installed package. What
are your feelings about something like this? For cclib, that means copying
the data files during installation
> To be honest, I'm not very interested in getting cclib into any of the
> distributions. Of course, you are welcome to do it. However, since the
> API has not currently settled, you are going to have problems if some
> program, e.g. PyMOlyze/GaussSum, starts to depend on it. Basically,
> you will find it difficult ever to upgrade it if the API changes.
> Also, cclib has optional dependencies on openbabel, BioPytho, PyQuante
> (and pyvtk?), which will need to be sorted out. I don't mean to
> discourage you, but I don't know if it's a good thing to be stuck with
> a version of cclib in Debian that's 1.5 years out of date, especially
> when Python packages are pretty easy to install, and cclib has been
> changing a lot from one release to the next.
I totally understand, this was more or less the answer I was expecting :), and
I made these debs mostly out of curiousity.
> I'd actually be more
> interested in getting a Python egg together for
> ez_setup.py/setuptools. I don't know what the current situation is
> with eggs and Debian though.
Well, if you have setuptools installed and put this in setup.py:
try:
from setuptools import setup
except ImportError:
from distutils.core import setup
and run "python setup.py bdist_egg", you get ann egg that should be ready to
go (I'm attaching it). Should I add this code along with other stuff
(dependency info, etc.) to the trunk? As far as Debian is concerned,
installing the package python-setuptools lets you use eggs.
Cheers,
Karol
--
written by Karol Langner
Mon Jun 4 11:42:16 CEST 2007
|
|
From: Noel O'B. <bao...@gm...> - 2007-06-04 11:17:42
|
On 04/06/07, Karol Langner <kar...@kn...> wrote: > On Monday 04 June 2007 11:19, Noel O'Boyle wrote: > > > 3. Would you be interested in seeing debs of GaussSum and/or PyMOLyze? > > > I'm not aware if any exist already. > > > > GaussSum's already in there. > Yes, now I see. > > > > 4. A testing issue has occured to me while testing the packages. It > > > should be possible to run the test suite we have from the Python > > > interpreter. I think that adding this is a good idea for the next > > > release, together with an option in setup.py to install the test scripts > > > and data files. What are you poinions on this? > > > > Well, I think you should see how this is done for other packages > > first. The general recommendation is that "python setup.py test" > > should run the tests. But I think you are suggesting something beyond > > this. > Sure, "python setup.py test" is the standard. I was thinking of something like > numpy.test(), which runs the NumPy test suite for the installed package. What > are your feelings about something like this? For cclib, that means copying > the data files during installation Sure, go for it - we can probably standardise the tests a bit more too. I'm not sure what you mean regarding the data files. They are currently installed by default, right? > > To be honest, I'm not very interested in getting cclib into any of the > > distributions. Of course, you are welcome to do it. However, since the > > API has not currently settled, you are going to have problems if some > > program, e.g. PyMOlyze/GaussSum, starts to depend on it. Basically, > > you will find it difficult ever to upgrade it if the API changes. > > Also, cclib has optional dependencies on openbabel, BioPytho, PyQuante > > (and pyvtk?), which will need to be sorted out. I don't mean to > > discourage you, but I don't know if it's a good thing to be stuck with > > a version of cclib in Debian that's 1.5 years out of date, especially > > when Python packages are pretty easy to install, and cclib has been > > changing a lot from one release to the next. > I totally understand, this was more or less the answer I was expecting :), and > I made these debs mostly out of curiousity. Sounds like I'm becoming predictable :-) > > I'd actually be more > > interested in getting a Python egg together for > > ez_setup.py/setuptools. I don't know what the current situation is > > with eggs and Debian though. > Well, if you have setuptools installed and put this in setup.py: > try: > from setuptools import setup > except ImportError: > from distutils.core import setup > and run "python setup.py bdist_egg", you get ann egg that should be ready to > go (I'm attaching it). OK, so that was easy! Go ahead and commit it, but please make sure that our regular distributions are unchanged by this. >Should I add this code along with other stuff > (dependency info, etc.) to the trunk? As far as Debian is concerned, > installing the package python-setuptools lets you use eggs. Sure, go ahead. > Cheers, > Karol > > -- > written by Karol Langner > Mon Jun 4 11:42:16 CEST 2007 > > |
|
From: Karol L. <kar...@kn...> - 2007-06-05 13:42:44
|
On Monday 04 June 2007 13:17, Noel O'Boyle wrote: > > > > 4. A testing issue has occured to me while testing the packages. It > > > > should be possible to run the test suite we have from the Python > > > > interpreter. I think that adding this is a good idea for the next > > > > release, together with an option in setup.py to install the test > > > > scripts and data files. What are you poinions on this? > > > > > > Well, I think you should see how this is done for other packages > > > first. The general recommendation is that "python setup.py test" > > > should run the tests. But I think you are suggesting something beyond > > > this. > > > > Sure, "python setup.py test" is the standard. I was thinking of something > > like numpy.test(), which runs the NumPy test suite for the installed > > package. What are your feelings about something like this? For cclib, > > that means copying the data files during installation > > Sure, go for it - we can probably standardise the tests a bit more > too. I'm not sure what you mean regarding the data files. They are > currently installed by default, right? I don't think so, at least not with my machine: langner@slim:~$ ls -l apps/python/lib/python2.5/site-packages/cclib total 24K drwxr-xr-x 2 langner langner 4.0K 2007-06-05 15:22 bridge -rw-r--r-- 1 langner langner 247 2007-05-20 01:03 __init__.py -rw-r--r-- 1 langner langner 479 2007-06-05 15:22 __init__.pyc drwxr-xr-x 2 langner langner 4.0K 2007-06-05 15:22 method drwxr-xr-x 2 langner langner 4.0K 2007-06-05 15:22 parser drwxr-xr-x 2 langner langner 4.0K 2007-06-05 15:22 progress Did you mean for them to be installed? > > > I'd actually be more > > > interested in getting a Python egg together for > > > ez_setup.py/setuptools. I don't know what the current situation is > > > with eggs and Debian though. > > > > Well, if you have setuptools installed and put this in setup.py: > > try: > > from setuptools import setup > > except ImportError: > > from distutils.core import setup > > and run "python setup.py bdist_egg", you get ann egg that should be ready > > to go (I'm attaching it). > > OK, so that was easy! Go ahead and commit it, but please make sure > that our regular distributions are unchanged by this. OK, I haven't changed any keywords to the call to setup() for now. You can build the egg now, though. -- written by Karol Langner Tue Jun 5 15:37:20 CEST 2007 |
|
From: Karol L. <kar...@kn...> - 2007-06-11 00:04:08
|
On Monday 04 June 2007 13:17, Noel O'Boyle wrote: > I'm not sure what you mean regarding the data files. They are > currently installed by default, right? They weren't, now they are when using 'python setup.py install'. Also, I added MANIFEST.in which is supposed to automatically take care of MANIFEST when building the source distirbution. I'm not removing manifest.py, though, because I don't know if this works also on Windows (someone try?). -- written by Karol Langner Mon Jun 11 02:00:57 CEST 2007 |
|
From: Noel O'B. <bao...@gm...> - 2007-06-12 16:34:41
|
On 11/06/07, Karol Langner <kar...@kn...> wrote: > On Monday 04 June 2007 13:17, Noel O'Boyle wrote: > > I'm not sure what you mean regarding the data files. They are > > currently installed by default, right? > They weren't, now they are when using 'python setup.py install'. Great, although I don't think we're are going to distribute the regression files as part of the normal distribution. When I create the source distribution, I get a file of size 68MB. We will probably require users to download separately the log file distribution if they want to run regression.py fully. > Also, I added MANIFEST.in which is supposed to automatically take care of > MANIFEST when building the source distirbution. I'm not removing manifest.py, > though, because I don't know if this works also on Windows (someone try?). Seems to work very well. I'm not sure why I didn't do this in the first place (perhaps not supported by Python 2.3??, or else I couldn't figure it out). > -- > written by Karol Langner > Mon Jun 11 02:00:57 CEST 2007 > |
|
From: Karol L. <kar...@kn...> - 2007-06-12 20:41:09
|
On Tuesday 12 June 2007 18:34, Noel O'Boyle wrote: > On 11/06/07, Karol Langner <kar...@kn...> wrote: > > On Monday 04 June 2007 13:17, Noel O'Boyle wrote: > > > I'm not sure what you mean regarding the data files. They are > > > currently installed by default, right? > > > > They weren't, now they are when using 'python setup.py install'. > > Great, although I don't think we're are going to distribute the > regression files as part of the normal distribution. When I create the > source distribution, I get a file of size 68MB. We will probably > require users to download separately the log file distribution if they > want to run regression.py fully. Hmm... when I use 'python setup.py sdist' I get a *.tar.gz file around 6MB, without the regression files. Are you doing this under Windows? I haven't tried it, maybe the MANIFEST syntax has a slightly different meaning. I will try it under Windows when I get the chance. -- written by Karol Langner Tue Jun 12 22:37:35 CEST 2007 |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X