pmxbot IRC Log Viewer

[17:49:54] <Wooble> Is it wrong for me to completely judge a module by the fact its docs suggest using sudo pip to install it? :)

[18:32:26] <ionelmc> Wooble: does ability to produce quality code correlate with ability to produce quality package, or meet specific packaging practices for that matter?

[18:38:54] <Wooble> ionelmc: hard to say.

[19:32:35] <ronny> ionelmc: in my experience the idea that sudo pip install is fine correlates with more strange practices to be found in the codebase

[19:57:38] <dstufft> I sudo pip

[19:57:51] <dstufft> it's pretty OK in most situations

[20:06:11] <ronny> dstufft: using it is very different from advocating it as default install method for a project

[20:21:58] <preyalone> help, i get a 500 error when i try to publish my first python package: https://github.com/mcandre/frozen/issues/1

[20:22:40] <Wooble> maybe it doesn't like version 0.0? :/

[20:23:47] <Wooble> preyalone: although you also have a previous 403 there; you're not allowed to register "frozen" apparently?

[20:24:40] <Wooble> preyalone: the name appears to already be taken ( https://warehouse.python.org/project/frozen/0.1.0/ )

[20:26:34] <preyalone> but there is no https://pypi.python.org/pypi/frozen

[20:27:38] <Wooble> yes, the default pypi doesn't show unreleased by registered things.

[20:27:51] <Wooble> s/by/but/

[20:28:15] <preyalone> this ghostly frozen hasn't been touched since january... would it be okay if i used it to publish something?

[20:28:36] <preyalone> 0 downloads

[20:28:50] <Wooble> You can't, unless the owner makes you the maintainer, as far as I know.

[20:28:57] <preyalone> no description

[20:29:22] <Wooble> (although I'm not a pypi admin; maybe they'd be willing to nuke it if it's never used; I have no idea what the policy is)

[20:30:06] <preyalone> i'd ask the maintainer nikola, but no contact details are available: https://warehouse.python.org/user/nikola/

[20:30:10] <Wooble> maybe that package is being held up in a long legal dispute with Disney :)

[20:30:16] <preyalone> haha

[20:30:53] <preyalone> "nikola@generic.company" https://github.com/nikola

[20:33:53] <preyalone> i've asked him via a GitHub issue: https://github.com/nikola/htpc-updater/issues/2

[20:34:20] <preyalone> i'll wait, then ask an admin what he/she thinks, then rename my project if necessary

[20:35:12] <preyalone> pypi does seem to allow 0.0, or at least version 0.0.0: https://encrypted.google.com/#safe=off&q=pypi+version+0.0

[20:39:53] <dstufft> the 500 was my fault

[20:39:58] <dstufft> I did a bad deploy

[20:44:43] <preyalone> 500 fix confirmed. it's just a name conflict in my case.

[22:09:48] <cmyers> greetings all - because reasons, I need to build a 32-bit python in a 64-bit docker container. python 2.6.1 specifically. I've done this, and after much gnashing of teeth, I have a python install with easy_install and pip. but it seems broken.

[22:10:05] <cmyers> When I try to pip install things, I get a crazy error. full stack trace here: http://paste.pound-python.org/show/vXhofD0xpChuTxy83LB4/

[22:10:37] <cmyers> short version is warnings about InsecurePlatformWarning: followed by stacktrace ending in "TypeError: unsupported operand type(s) for &: 'NoneType' and 'int'

[22:10:40] <cmyers> "

[22:10:50] <cmyers> all of it comes from pip-7.1.0-py2.6.egg

[22:24:18] <ronny> cmyers: i suspect your python is built without threading support?

[22:24:37] <ronny> cmyers: also why did you install pip as a egg?

[22:29:20] <cmyers> ronny: I have no idea why I installed pip as an egg. I just did /build/.../path/to/python ./ez_setup.py && /path/to/easy_install pip

[22:29:30] <cmyers> if there is a better / more correct way to have pip installed, I can try that

[22:30:12] <ronny> cmyers: usually the most sane way is get_pip.py on older python

[22:30:19] <ronny> (it will install pip as wheel then)

[22:30:37] <cmyers> cool, does get_pip.py come with python then? *goes to look*

[22:30:49] <ronny> nope

[22:31:02] <ronny> cmyers: python 2.7 grew python -m ensurepip

[22:31:05] <cmyers> is there a set of directions somewhere I should look at?

[22:31:17] <ronny> get_pip.py must be taken from the pip repo on github if i recall correct

[22:31:34] <tdsmith> or https://bootstrap.pypa.io/get-pip.py

[22:31:45] <cmyers> that's sorta where ez_setup.py came from, so that makes sense

[22:31:55] <ronny> oh, that one is probably better

[22:32:23] <cmyers> yeah, just found that, thx

[22:32:33] <ronny> cmyers: ez_setup is for ensuring a project can use setuptools in any case, it should be avoided to use easy_install

[22:32:57] <tdsmith> generic.company is a registered domain fwiw

[22:32:58] <cmyers> so I should still run ez_setup but just not use setuptools to install pip?

[22:33:29] <ronny> cmyers: why would you run ez_setup at all?

[22:33:35] <cmyers> I have no idea

[22:33:38] <cmyers> some things need setuptools

[22:33:46] <cmyers> should I use pip to install that?

[22:33:50] <ronny> if you just make your complete python, just install pip and setuptools in the latest versions directly

[22:34:03] <ronny> (pip uses setuptools, get_pip should take both as far as i remember)

[22:34:10] <cmyers> in case it isn't obvious, you should just pretend I know a lot about computers and linux and very littel about python =/

[22:34:14] <cmyers> oh, excellent

[22:34:44] <cmyers> re-running my docker build now. ah, the joys of timetravel.

[22:35:11] <cmyers> ok, so now invokeing /path/to/python get-pip.py causes that error I pastebin'd

[22:35:26] <cmyers> so further evidence, perhaps, that my python is build wrongly?

[22:35:36] <ronny> cmyers: do you have threading enabled?

[22:35:55] <cmyers> does that not happen by default? if not, tell me what arg to add to the ./configure script and I'll shove it in

[22:36:15] <cmyers> my current configure line is: CFLAGS="-m32" LDFLAGS="-m32" EXTRA_CFLAGS="-m32" OPT="-m32" ./configure --prefix=/build/toolchain/lin32/python-2.6.1 --build=i686-pc-linux-gnu --host=i686-pc-linux-gnu --enable-unicode=ucs4 --with-ssl

[22:36:37] <cmyers> don't even get me started about the -m32's all over the place

[22:40:33] <cmyers> /build/toolchain/lin32/python-2.6.1/bin/python -m threading => works

[22:40:45] <cmyers> so I think my python is built with threading support

[22:50:38] <cmyers> ok, having read the comments in get-pip.py, I can now confidently say the issue is pip is exploding when run under the python I built, and the nearby warning implies it is SSL related

[22:50:59] <cmyers> I am passing the configure script --with-ssl and the system ssl is openssl-1.0.1e

[23:07:33] <cmyers> ugh, this is so frustrating, nothing works

[23:07:44] <cmyers> flashback to the perl 5.6.0 days

[23:09:29] <cmyers> pip 7.1.0 is the newest version...is it really compatible with both new python and old python?

[23:09:33] <cmyers> current situation implies no?

[23:10:28] <cmyers> python2.6 -m virtualenv .ve && source .ve/bin/activate && pip install -r dev-requirements.txt && python ./tasks/generate.py fails on the current checkout of pip

[23:10:43] <cmyers> with ImportError: No module named request

[23:10:59] <cmyers> module "request" doesn't exist, but "requests" does and is installed

[23:14:00] <tdsmith> cmyers: it looks like the error is in the lockfile module?

[23:14:31] <cmyers> well, it's importing import urllib.request

[23:14:46] <cmyers> but that's probably a big nasty rabbit hole of don't care anyways

[23:14:57] <cmyers> I shouldn't have to build a customer get-pip.py here

[23:15:04] <cmyers> it was just a crazy idea

[23:15:26] <cmyers> the real problem is, I build python, I run python get-pip.py, and it errors out

[23:15:52] <tdsmith> do you have a link to that traceback?

[23:16:02] <cmyers> let me get the newer one

[23:17:12] <cmyers> http://paste.pound-python.org/show/3AYIuud2RfNKRIb9UEfW/

[23:18:18] <cmyers> the entire stacktrace occurs in pip.zip, the embedded copy of pip in get-pip.py

[23:18:53] <cmyers> so this version of pip seems incompatible with this version of python (which I compiled from python 2.6.1 src with some crazy configure flags but otherwise unaltered)

[23:19:19] <tdsmith> this looks like a known issue in lockfile; this bug report https://code.google.com/p/pylockfile/issues/detail?id=8 blames this bug in python 2.6 https://bugs.python.org/issue5632

[23:19:43] <cmyers> ooh, that sounds useful *reads*

[23:20:34] <tdsmith> hmhmmhmm and lockfile's been assimilated by openstack

[23:20:39] <cmyers> says downgrading fixed it... but we would need to downgrade the version of lockfile bundled in the zip?

[23:21:13] <cmyers> for incredibly stupid reasons, I need actual python 2.6.1 not 2.6.X whereX>1

[23:21:32] <cmyers> although /me is about to go murder someone over that...

[23:22:40] <tdsmith> i think ideally you convince the openstack team to accept a patch and release it, and then pip vendors the new version, but i don't actually work on pip

[23:23:26] <tdsmith> some relevant information at https://github.com/smontanaro/pylockfile

[23:23:43] <cmyers> ah, cool, thx

[23:25:00] <cmyers> so the "right thing" to do here might be to add special case code to pylockfile so in the case where python version is old enough to have this bug, it still handles it correctly?

[23:25:10] <cmyers> and hope that openstack folks will accept that

[23:25:26] <cmyers> and get the pip guys then to roll a new get-pip.py with that new dep in there.

Log file Viewer

Help | Karma | Search:

#pypa logs for Wednesday the 5th of August, 2015