Mailman 3 July 2008 - Distutils-SIG

Re: [Distutils] Setuptools documentation
by Phillip J. Eby 09 Jul '08

09 Jul '08

At 07:28 PM 7/9/2008 +0200, Florian Mayer wrote: >Phillip J. Eby wrote: > > > > The code is apparently correct here; the choice of example in the text > > is bad. Patches welcome. > > > >I do not seem to quite understand the answer. Which code is correct? parse_version() is correct: "-" should be dropped when it precedes a prerelease tag. Apparently I changed it as a bugfix, but didn't update the docs: http://svn.python.org/view?rev=41630&view=rev

1 0

Subversion 1.5 entries not parsed successfully
by Jason R. Coombs 06 Jul '08

06 Jul '08

Subversion 1.5 is released and official now. Any chance we can get the patch suggested at http://mail.python.org/pipermail/python-bugs-list/2008-May/051721.html or any other fix applied to the 0.6 branch of setuptools? I acknowledge this fix may not fit into the grand scheme of the development, but it seems to me to be a straightforward and logical fix to address what will be an increasingly prominent problem. Sincerely, Jason

1 0

Running 'ldconfig' after install command
by Milinda Pathirage 04 Jul '08

04 Jul '08

Hi all, I'am developing a extension for Python using C library. I have correctly follow the steps described in docs and now the setup is working fine. But the problem is we have some libraries deployed in a custom location other than /usr/lib or any other default location. So at the installation time I am copying a configuration file containing library location to /etc/ld.so.conf.d. But after copying this I have to run ldconfig to make system pickup the libraries. Can any one please suggest a method that I can run ldconfig when the instllation is finished (As the last step of installation run the ldconfig.). Thanks in advance Milinda -- http://mpathirage.com http://wso2.org "Oxygen for Web Service Developers" http://wsaxc.blogspot.com "Web Services With Axis2/C"

1 0

ctypes based modules
by Uwe Schmitt 04 Jul '08

04 Jul '08

Hi, how do you distribute ctypes based modules on a linux machine ? What does your setup.py look like ? I have to copy my shared library to /usr/lib or another common place, but I would prefer .../site-packages/... In the latter case I think I have to modify LD_LIBRARY_PATH oder some ld.so.conf configuration file. So, what is the prefered method for this task ? Greetings, Uwe -- Dr. rer. nat. Uwe Schmitt F&E Mathematik mineway GmbH Science Park 2 D-66123 Saarbrücken Telefon: +49 (0)681 8390 5334 Telefax: +49 (0)681 830 4376 uschmitt(a)mineway.de www.mineway.de Geschäftsführung: Dr.-Ing. Mathias Bauer Amtsgericht Saarbrücken HRB 12339

1 0

Optimization of C++ extensions to Python
by Michael Wieher 03 Jul '08

03 Jul '08

hi, I tried this question on the C++/Python SIG and was told to ask here. So. When using distutils to build an extension (shared library), a C++ library to be imported into python, I find the optimization level set for me, to -O2 This is great, I guess, but for certain debugging tools (valgrind) I need that to be -O0 I tried setting the extra_compile_args and get output like this gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fPIC -I/usr/include/python2.4 -c memrid.cpp -o build/temp.linux-i686-2.4/memrid.o -O0 Which as you can see, has -O2 (the default) following -DNDEBUG ... and my "extra" argument appended ... as -O0. I need to be sure that this overrides the -O2 setting (or know how to set it to -O0 correctly) ... I noted that the binaries produced are different and that the binary created when -O0 is in the gcc string is larger than the other... the GCC documentation I could find didn't address the issue of multiple flags on the same line and/or how the compiler would handle them, and I think there is/should be a "correct" way of doing this such that the confusion of "which optimization level is it using" doesn't exist. Can anyone a) tell me if my binary compiled w/extra_compile_args -O0 is, indeed, level zero b) tell me the correct way to adjust this flag? thank you -mike

2 1

Re: [Distutils] Correct way to allow C extension building to fail?
by Andreas Klöckner 03 Jul '08

03 Jul '08

On Mittwoch 02 Juli 2008, you wrote: > on failure have it raise any kind of exception (custom or otherwise) > and wrap the creation of the module in a try/except block. if it > fails, proceed on the alternate branch of execution This seems to do what I want. Thanks! Andreas

1 0

Correct way to allow C extension building to fail?
by Andreas Klöckner 03 Jul '08

03 Jul '08

Hi there, I'm working on a Python module (simplejson) that comes with a C extension that can optionally speed up some of its operations. It is however ok for this extension to fail to build, in which case the module will use Python fallbacks. How can I allow for this failure? Bob Ippolito, the original author, has this implemented as follows: 8< ------------------------------------------------------ class ve_build_ext(build_ext): # This class allows C extension building to fail. def run(self): try: build_ext.run(self) except DistutilsPlatformError, x: self._unavailable(x) def build_extension(self, ext): try: build_ext.build_extension(self, ext) except (CCompilerError, DistutilsExecError), x: self._unavailable(x) def _unavailable(self, exc): print '*'*70 print BUILD_EXT_WARNING print exc print '*'*70 ... setup(cmdclass={'build_ext': ve_build_ext},) 8< ------------------------------------------------------ On current setuptools, this fails with ** error: Setup script exited with error: can't copy 'simplejson.egg-info\native_libs.txt': doesn't exist or not a regular file How can I fix this? Andreas

2 3

setuptools "tests" sprint today
by Tarek Ziadé 02 Jul '08

02 Jul '08

Hi, Like discussed earlier, some people will work today on adding tests to setuptools. The idea is to add human-readable doctests to setuptools a bit like what zc.buildout has. I am planning to create one doctest per setuptools feature, and some general doctest that demonstrates the usage of the tool. (functional tests but preferrably with no os.popen/os.system magic) To synchronize all our work (I doubt we can all get commiter access to svn.python.org today :D ), I have created a launchpad project here: https://launchpad.net/setuptools-test It is a copy of the current setuptools trunk, you can get it with bzr: bzr branch lp:~tziade/setuptools-test/main I suggest we all push changes there then present that work for review to Phillip, Let's meet in #distutils today ! Regards Tarek -- Tarek Ziadé | Association AfPy | www.afpy.org Blog FR | http://programmation-python.org Blog EN | http://tarekziade.wordpress.com/

4 11

[issue26] Subversion 1.5 and tag_svn_revision do work together
by Jeroen Ruigrok van der Werven 02 Jul '08

02 Jul '08

New submission from Jeroen Ruigrok van der Werven <asmodai(a)in-nomine.org>: When using tag_svn_revision = true in setup.cfg with Subversion 1.5 the resulting revision number is 0. Example: creating /usr/local/lib/python2.5/site-packages/Trac-0.12dev_r0-py2.5.egg ---------- messages: 54 nosy: asmodai priority: bug status: unread title: Subversion 1.5 and tag_svn_revision do work together _______________________________________________ Setuptools tracker <setuptools(a)bugs.python.org> <http://bugs.python.org/setuptools/issue26> _______________________________________________

1 0