PMXBOT Log file Viewer

Help | Karma | Search:

#pypa logs for Thursday the 13th of August, 2015

(Back to #pypa overview) (Back to channel listing) (Animate logs)
[15:18:08] <chudler> "2015.4.28" is published and "2015.04.28" also, but the former seems to go nowhere. It killed my pip operations temporarily. https://pypi.python.org/pypi/certifi/2015.4.28
[15:18:57] <chudler> whoops sorry, I posted this by mistake. I meant to tell you about what I had wrong first. Can anything be done about that (broken?) module?
[15:21:09] <dstufft> 2015.04.28 and 2015.4.28 are equivilant
[15:22:12] <chudler> Gotcha, I got "No requirements found" using pip, but its probably misled me. I understand now that it treated them as equivalent
[15:22:35] <chudler> I am continuing troubleshooting that without including this poor innocent module
[17:20:14] <egc> i'm trying to override a value from a setup.cfg section on a python setup.py commandline... is this possible, in general?
[17:20:56] <egc> for example, my setup.cfg has "[_ldap]\nlibs = bar"
[17:21:22] <egc> can i say something like, "python setup.py build _ldap --libs=foo" ?
[17:22:59] <egc> (i'd like to avoid editing the cfg file)
[17:29:29] <egc> i suppose this case is tripping me up because the [_ldap] syntax disallows those options for the build command...
[19:08:33] <dstufft> agronholm: did you forget to write a message in your email to python-ideas?
[19:14:18] <agronholm> dstufft: no
[19:14:24] <agronholm> dstufft: my reply is further down
[19:14:39] <agronholm> "Amen to this. This is EXACTLY where I've hoped Python would go :)"
[19:14:51] <dstufft> ahhh
[19:14:52] <dstufft> oops
[19:14:58] <dstufft> it blended in with my lines
[19:15:04] <dstufft> missed it
[19:15:06] <dstufft> my bad :D
[19:15:25] <agronholm> btw, now that I have your attention -- when do you plan to make a new pip release?
[19:15:39] <agronholm> I've been anxiously waiting for one since the distlib upgrade
[19:15:43] <dstufft> um
[19:16:19] <agronholm> that upgrade fixes a problem that essentially makes tox pointless on jython
[19:17:27] <agronholm> *useless, not pointless
[19:19:37] <dstufft> agronholm: I can cut one later today or tomorrow
[19:19:42] <agronholm> dstufft: great!
[19:20:08] <agronholm> and to fix tox, virtualenv needs a new release too obviously
[19:21:46] <dstufft> yea, we always cut a virtualenv release with a new pip release
[20:09:45] <jchen> hi, i'm using `setup.py develop` for deving on my flask app, which uses node and has the local `node_modules` directory with thousands of files. setup.py gets hung up checking each of those files, none of which are python files we care about. how do I exclude them?
[20:10:06] <jchen> i added recursive-exclude node_modules to MANIFEST.in, but seems that setup.py develop doesnt respect that
[20:12:22] <dstufft> jchen: you might try ``prune node_modules`` instead
[20:13:23] <jchen> k, i'll give that a shot
[20:15:40] <jchen> looks like it's still looking into those directories... https://gist.github.com/fly/19c32f7f31b62d178557
[20:17:20] <jchen> maybe it's because i'm using `packages=find_packages(exclude=['ez_setup'])` in setup.py?
[20:18:32] <xafer> just in case, prune should be used as `prune path/to/node_modules`
[20:18:36] <dstufft> oh right
[20:18:45] <dstufft> jchen: you might want to exclude node_modules too
[20:18:57] <jchen> yep, tryinng that now
[20:19:47] <jchen> nop, still looking into that dir with addition of 'node_modules' to find_packages exclude list
[20:23:51] <xafer> what does your MANIFEST.in look like ?
[20:27:41] <jchen> https://gist.github.com/fly/f6b35e5a5334ae95917a
[21:04:31] <xafer> jchen, cf https://docs.python.org/2/distutils/sourcedist.html#commands
[21:05:09] <xafer> you should have 3 prunes: prune node_modules, then prune dockerfiles and prune logs
[21:06:21] <xafer> but they are not needed since they are included anyhow
[21:06:29] <xafer> *are not included anyhow
[21:07:18] <xafer> and recursive-exclude expect a dir followed by a list of patterns
[21:08:21] <xafer> but I've no idea for your slow setup.py :-/
[21:09:53] <pjdelport> Have you considered one of the setuptools extensions that produces your source manifest from version control instead?
[21:10:19] <pjdelport> It's much better than trying to manually maintain a manifest template, in my experience.
[21:28:26] <ronny> hi
[21:42:10] <ark3> How does `python setup.py bdist_wheel` decide the stuff after package_name-version- in the name (and presumably metadata) of the .whl file? In particular, I have two pure-Python2 packages that I'm building; one shows up with -py2-none-any while the other has -cp27-none-macosx_10_10_intel.
[21:42:24] <ark3> I think I'd like the latter to be -py2-none-any as well.
[22:13:50] <xafer> ark3, for some reason bdist_wheel thinks the second one is not a pure python package
[22:14:34] <ark3> xafer: Okay. How might I figure out why that is? How does bdist_wheel decide?
[22:15:11] <ark3> FWIW, the second wheel has no module or package, just scripts.
[22:16:08] <xafer> well, thats the reason :)
[22:17:45] <xafer> cf https://hg.python.org/cpython/file/3a1db0d2747e/Lib/distutils/dist.py#l999
[22:18:45] <xafer> if it has no pure python module, is_pure returns False
[22:20:13] <xafer> you can use an hack in your setup.py to tell distutils it is pure
[22:21:39] <xafer> cf http://lucumr.pocoo.org/2014/1/27/python-on-wheels/#building-wheels for an example of such hack (but in your case, you want is_pure to return True)
[22:34:28] <ark3> xafer: Perfect. Thanks.
[22:36:05] <ark3> xafer: Out of curiosity, is this considered a bug?
[22:36:39] <ark3> Oops, gotta run. Again, thanks for your help.
[23:09:15] <aleatoire> Hi everyone
[23:10:31] <aleatoire> I have a bunch of questions, anyone around feel like answering?
[23:15:49] <nibalizer> aleatoire: just ask your questions
[23:15:54] <nibalizer> if people know the answer they will speak up
[23:16:19] <aleatoire> Ok. Is there a way to upgrade setuptools without sudo?