Log file Viewer

[00:36:11] <dstufft> xafer: "sort of"

[00:36:35] <dstufft> xafer: virtualenv only ever uses it with a copy of pip bundled with virtualenv, same with ensurepip and get-pip.py

[00:42:08] <StevenK> dstufft: Hai! Can I bug you for packaging 16.1 again?

[00:42:28] <dstufft> ah right

[00:42:30] <dstufft> moment

[00:47:33] <dstufft> StevenK: https://github.com/pypa/packaging/releases/tag/16.1

[00:47:57] <StevenK> Yay

[00:48:55] <StevenK> dstufft: Do you have time to update pip

[00:48:58] <StevenK> Sigh

[00:49:04] <StevenK> dstufft: Do you have time to update pip's copy, or shall I?

[00:49:17] <dstufft> if you can that'd be great :)

[00:49:25] <dstufft> I'm gonna go watch some more football

[14:58:30] <quinoa> @dstufft what is the proper way to upgrade the vendored libs in pip ?

[14:58:44] <quinoa> The Makefile removes more than it adds

[16:57:13] <quinoa> what is canonical way to install pip from git repo via a local checkout? It really wants to either use the network or local cache

[17:04:54] <odyssey4me> howdy everyone - when compiling wheels, does pip check its index and extra-index with equal weight? or does it take from the index if it finds something, and only look at the extra-index if it doesn't find it in the index?

[17:06:35] <odyssey4me> I'm suspecting that first-found wins, and therefore it does not give the index and extra-index equal weight. I'd just like to understand if this is working as designed?

[17:09:23] <dstufft> quinoa: I tend to just rm -rf whatever && pip install -t pip/_vendor whatever==version

[17:10:05] <dstufft> odyssey4me: I forget the exact order, but highest version wins, if extra index and index both have all the same versions, then I think extra index might win (but I could be wrong)

[17:10:58] <odyssey4me> dstufft thanks, I'm in the process of testing that theory - will know in a bit

[17:11:26] <odyssey4me> interestingly though, in my index I have a zip file but in my extra-index I have a wheel... and the zip won

[17:11:44] <odyssey4me> busy verifying though, so don't quote me on it just yet :)

[17:11:47] <dstufft> is the zip file higher version?

[17:12:16] <dstufft> The thing is basically Highest Version -> Prefer Wheel over Sdist -> Index Order

[17:13:50] <odyssey4me> dstufft well, MySQL-python-1.2.5.zip vs MySQL_python-1.2.5-cp27-none-linux_x86_64.whl

[17:14:37] <dstufft> gotcha

[17:15:05] <dstufft> assuming you're running on CPython 2.7 and Linux the wheel should be preferred

[17:15:48] <odyssey4me> CPython 2.7 relates to the cp27 part of the file name, right?

[17:16:05] <dstufft> odyssey4me: If this is for Openstack and the Openstack Wheel Mirror though, I think the problem is that the wheel mirror needs adjusted to normalize the names, at least krotscheck was investigating this earlier and that seemed to be the conclusion

[17:16:06] <dstufft> yea

[17:17:04] <odyssey4me> ah, and yes this does relate to the OpenStack wheel mirror - I'm trying to use it for gating as a preference to using RPC's wheel repo and pypi

[17:19:02] <odyssey4me> dstufft what do you mean by normalise the names - my apologies for what may be rookie questions

[17:19:46] <dstufft> Yea, pip normalizes the name using the function in PEP 503 when it generates the URLs it fetches, essentially it lower cases and turns . _ and - into -, so "MySQL-Python" becomes "mysql-python" and it does GET $ROOT/mysql-python/

[17:19:58] <dstufft> the Openstack Wheel mirror was using the "real", unnormalized name

[17:20:17] <dstufft> so it was expecting $ROOT/MySQL-Python/

[17:21:30] <odyssey4me> ah, and it all burns in a fire

[17:22:06] <dstufft> yea, pip does the GET $ROOT/mysql-python/, get's a 404 and skips it

[17:22:30] <dstufft> Once the openstack wheel mirrors normalize the names correctly, that shoudl work for you

[17:25:28] <quinoa> @dstufft can I use the pip in my local checkout to install itself?

[17:25:57] <quinoa> it wants to hit either the cache or the network

[17:26:16] <quinoa> if I use setup.py then it uses easy_install

[17:26:23] <quinoa> a tots diff code path

[17:26:24] <dstufft> quinoa: I'm not sure what you mean, you're trying to install pip itself? I'd just do ``pip install /path/to/pip``

[17:26:40] <quinoa> I have no pip in that virtualenv

[17:26:53] <quinoa> and the pip I require is the one in tree I just modified

[17:28:03] <dstufft> ah

[17:28:25] <quinoa> $ jython -c 'import sys, pip; sys.exit(pip.main(["install", "--ignore-installed"] + sys.argv[1:]))' pip

[17:28:31] <quinoa> put it pulls my cache

[17:28:36] <quinoa> Using cached pip-8.0.2-py2.py3-none-any.whl

[17:29:05] <dstufft> quinoa: remove "pip" at the end of that with /path/to/pip/

[17:30:48] <quinoa> nice, progress. !

[17:50:05] <quinoa> @dstufft when pip is installed local path, it short circuits through setuptools

[17:50:20] <quinoa> totally different codepath and different bugs

[21:19:23] <xafer> dstufft: what will be the next pip version ? 8.1 ? Or do you plan a 8.0.3 with small fixes ?

[21:27:44] <dstufft> xafer: *shrug*

[21:27:54] <dstufft> if we need a 8.0.3 for some reason I'll cut one

[21:27:57] <dstufft> otherwise, 8.1

[21:41:13] <xafer> Well https://github.com/docker-library/python/pull/83 might appreciate a 8.0.3

[21:42:58] <xafer> https://github.com/pypa/pip/pull/3434 might also be included

[21:45:27] <xafer> and maybe even https://github.com/pypa/pip/pull/3468

[21:51:01] <xafer> I guess this also depends on the ETA of 8.1

[21:57:11] <dstufft> xafer: seems fine, 8.1 probably isn't for some months

[22:02:08] <xafer> btw now that unit tests are passing on windows on 2.7/3.3/3.4, I'd like to have an official appveyor link for pypa/pip

[22:15:54] <xafer> dstufft: are you planning to release today ? Or should I create a 8.0.3 milestones with these 3 issues ?

[22:26:56] <dstufft> xafer: milestone please

[22:40:02] <xafer> https://github.com/pypa/pip/issues?utf8=%E2%9C%93&q=milestone%3A8.0.3+