[00:28:59] <gp> I guess this doesn't seem very straight forward. Is it not common to build python projects on one machine and deploy on another? I can't really find anything on google but maybe I am not searching correctly
[00:30:05] <tomprince> gp: The tools for doing that heterogenously on linux aren't there.
[00:30:17] <tomprince> (which is why pipy doesn't support linux wheels)
[00:30:59] <tomprince> If all the machines run the same os, it shouldn't be a problem.
[00:31:03] <gp> Right... I read all the discussions about that decision on mailing lists. But I thought that was for general consumption of packages
[00:31:55] <gp> They are the same minus the build tools and libraries
[00:32:30] <tomprince> Well, you need the libraries (but not the build tools) on both.
[00:32:42] <dstufft> gp: no it's common, but generally people do dynamic linking
[04:22:43] <collinanderson> it would be nice if there were any easy way to know what development headers you need when you "pip install pillow", or any package in general
[04:23:31] <dstufft> collinanderson: the problem is, how do you define that?
[04:26:19] <collinanderson> dstufft: not sure, but now that system_site_packages=False by default, it makes it a pain to install pillow
[04:27:21] <collinanderson> dstufft: even if package maintainers had to list off different dependencies for deb, rpm, etc, that would still be better than the current situation
[04:33:42] <collinanderson> dstufft: maybe you could say, something like: requires python.h
[04:48:12] <collinanderson> yeah, so you list out the required header files, like Python.h, once it's there, someone else can figure out a way to translate that into something like "yum whatprovides '/usr/include/*/Python.h'"
[04:59:05] <dstufft> collinanderson: sorry went to go get something to eat
[04:59:40] <dstufft> collinanderson: the files thing is interesting, I suggest posting it to distutils-sig
[09:49:33] <mgedmin> " Could not fetch URL http://pypi.python.org/simple/PyYAML/: <urlopen error [Errno -2] Name or service not known>" -- your DNS is broken
[09:49:47] <mgedmin> or you got a temporary network error; have you tried again?
[13:14:04] <jax> does pip use wget or curl to fetch the index?
[13:26:57] <[Tritium]> pip uses python's networking to make connections
[13:34:21] <[Tritium]> pip ships with the requests library (they vendor a copy), which uses either urllib or the python socket primatives to make the http requests. Ultimatly it does not matter what the requests library uses, its what pip uses, and IMHO what you should be using to do HTTP in python
[13:41:48] <jax> ok i'm missing root certificates i think
[13:41:56] <jax> because pip fails to connect to https://pypi.python.org/simple/
[13:42:14] <jax> this is some funky NAS, i need to figure out how to tell python to use other ca-certs
[13:48:18] <[Tritium]> jax: what version of pip are you on?
[15:47:57] <gp> Will the client cert option in pip 1.6 https://github.com/pypa/pip/issues/1424 work with find-links urls and download links found from the find-links?
[15:56:09] <gp> Cool. Is there an approximate ballpark on when 1.6 might be released?
[15:58:27] <dstufft> gp: sometime in the near future, I want to get it out but there's a few things I want to get in before we do, and there might be things other people want to get into 1.6 before it's released too