pmxbot IRC Log Viewer

[01:01:02] <lifeless> WAT

[01:01:16] <lifeless> tox raising DistributionNotFound errors on things installed as wheels on 2.6

[01:01:31] <lifeless> (globally, outside of any venv)

[01:01:39] <lifeless> e.g. usr-local

[01:18:13] <dstufft> lifeless: old version of setuptools?

[01:18:20] <dstufft> lifeless: like, < 0.6.28

[01:18:34] <lifeless> dstufft: likely

[01:18:53] <lifeless> yes

[01:18:54] <lifeless> 0.6.10

[01:19:08] <lifeless> if thats known broken, shouldn't pip automatically disable wheels?

[01:20:26] <dstufft> lifeless: possibly yea

[01:20:34] <dstufft> older setuptools didn't support the dist-info directory

[01:20:34] <lifeless> should I file a tissue?

[01:20:37] <dstufft> sure

[01:22:10] <lifeless> any idea why e.g. TestUpgradeSetuptools.test_from_setuptools_7_to_setuptools_7 passes locally, as does the install_builds_wheels test

[01:22:14] <lifeless> but fails in travis?

[01:27:09] <dstufft> lifeless: looking

[01:27:49] <lifeless> the uninstall_wheel test fails because we're deleting the delete marker file from site-packages

[01:29:22] <dstufft> lifeless: I didn't look at the PR yet, do you unamigiously add $CACHEDIR/wheels to --find-links?

[01:29:33] <lifeless> yes

[01:29:40] <lifeless> except when its disabled

[01:29:45] <dstufft> assert 'setuptools (0.9.8)' in 'pip (6.1.0.dev0)\nsetuptools (14.3)\n'

[01:30:16] <dstufft> it looks like we used a --find-links in combination with --no-index to explicitly control what version of setuptools was available

[01:30:29] <dstufft> but for some reason it's getting 14.3

[01:30:32] <lifeless> shouldn't test isolation prevent that though ?

[01:30:42] <lifeless> I checked and there's a unique homedir set for each test

[01:30:48] <lifeless> or meant to be

[01:32:11] <dstufft> lifeless: I bet the virtualenv creation is doing it

[01:32:23] <dstufft> lifeless: guess what version of 14.3 is bundled inside of virtualenv

[01:32:24] <dstufft> er

[01:32:29] <dstufft> what version of setuptools

[01:32:31] <dstufft> (too bad I ruined the surprise)

[01:32:35] <lifeless> what is setuptools

[01:32:48] <lifeless> ok, so isolation - thanks

[01:33:05] <dstufft> and virtualenv uses pip to install setuptools

[01:33:13] <lifeless> yeah

[01:33:23] <lifeless> I'll nuke the cache dir after the venv is setup

[01:33:23] <dstufft> so probably you just need to delete the the cache dir at the start of the test

[01:33:29] <dstufft> yea

[01:33:30] <dstufft> or that

[01:37:27] <lifeless> pushed up, hmm, probably should be an isolated commit. Meh.

[01:37:42] <lifeless> now looking at the uninstall one

[01:59:19] <lifeless> dstufft: ahahaha

[01:59:24] <lifeless> dstufft: tracked down the uninstall one

[01:59:31] <lifeless> I think

[01:59:42] <lifeless> the virtualenv install code ends up with a build-dir of site-packages

[01:59:48] <lifeless> so it gets marked with the delete marker

[02:22:13] <lifeless> wow this is freaking bizarre

[02:43:19] <lifeless> ahhha

[02:43:47] <lifeless> we cp -a /tmp/pip-build-kkw90P/setuptools

[02:43:49] <lifeless> into

[02:43:55] <lifeless> /tmp/pytest-20/test_uninstall_wheel0/workspace/venv/lib/python2.6/site-packages

[03:27:40] <lifeless> ah, enlightenment

[03:45:46] <lifeless> bombs away, lets see if thats better

[07:37:36] <lifeless> dstufft: seems to be somewhat sporadic failures now. Not sure whats up. I had a clean run other than flake8, fixed that, now dirty...

[17:16:20] <DanielHolth> OK if I go ahead and merge https://github.com/pypa/pip/pull/2561 ?

[19:38:22] <lifeless> morning folk

[19:38:41] <lifeless> dstufft: its now easter public holidays for me

[19:39:21] <lifeless> dstufft: it seems there's something flaky about the new tests/code- its passing for me every time,but travis has failed a couple of jobs

[19:39:36] <lifeless> dstufft: I rebased and it then passed those and failed different ones

[19:39:44] <lifeless> dstufft: but on the same test - my test functional test

[20:59:28] <qwcode> lifeless, any concerns https://github.com/pypa/pip/pull/2625

[21:02:44] <dstufft> qwcode: I'm tihnking I might cut a 6.1 before the implicit wheel building or setup-requires stuff lands, and then put all of that inside of a 7.0

[21:02:49] <dstufft> qwcode: what do you think of that

[21:03:08] <qwcode> dstufft, sure

[21:03:51] <qwcode> dstufft, I guess you saw the guy wanting 6.1 for the broken setuptools issue on ubuntu 12.04

[21:03:57] <dstufft> yea

[21:07:11] <lifeless> qwcode: looking

[21:10:39] <lifeless> qwcode: commented. I think its ok as-is but I've put a couple of notes of potential improvements

[21:11:40] <qwcode> lifeless, noted, I'll make those changes.

[22:47:30] <lifeless> dstufft: dunno if you saw but https://travis-ci.org/pypa/pip/builds/56846005 is at fine tuning - just need to figure out this sporadic failure thing

[22:49:09] <dstufft> lifeless: interesting

[22:49:22] <dstufft> lifeless: fwiw you can normally get a VM image from Travis CI to debug stuff on

[22:49:34] <dstufft> s/image//

[22:49:44] <dstufft> it looks like it's failing to install the upper wheel

[22:55:16] <lifeless> dstufft: yes, and it looks to me like the wheel unpacked dir will be bogus

[22:55:20] <lifeless> dstufft: but its odd - it passes here

[22:55:26] <lifeless> dstufft: and it passes on travis on some jobs

[22:55:42] <lifeless> dstufft: and rebase + push leads to which ones fail shifting around

[22:56:10] <dstufft> lifeless: hmm

[22:56:14] <dstufft> that's very odd

[22:56:38] <lifeless> yes

[22:56:50] <lifeless> I'd almost blame dictionary order

[22:57:13] <lifeless> if we didn't pin that

[22:57:15] <dstufft> tox should publish what hashseed it's using

[22:57:36] <dstufft> I don't think we do pin it do we ?

[22:57:46] <dstufft> I mean tox does at the begining of a run

[22:57:49] <dstufft> but it's different each time

[22:57:56] <lifeless> oh ok

[22:58:03] <lifeless> so hats worth checking then

[22:58:20] <dstufft> yea

[22:58:46] <lifeless> I

[22:59:04] <lifeless> I'll do some more trading - nearly @ 20M saved in e:d, then have a poke

[22:59:45] <dstufft> e:d?

[22:59:45] <lifeless> plus just put battlestar galactica S1 on

[22:59:52] <lifeless> elite dangerous

[23:00:26] <dstufft> ahh

[23:00:55] <lifeless> public holiday here today and monday

[23:01:39] <dstufft> I dunno if we get easter off or not

[23:02:25] <dstufft> probably not

[23:02:28] <dstufft> since it's on a weekend

[23:04:30] <lifeless> I have saved up 16M, I need 25M to get the next class of ship. I want those pixels.

[23:05:52] <lifeless> so I'm watching BSG S1 on one monitor, and trading back and forth at 400K profit per round trip on the other

[23:12:23] <dstufft> I've been playing bloodborne the last few days

[23:12:25] <dstufft> it's pretty good

[23:41:48] <lifeless> just read up on it, sounds fun

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Thursday the 2nd of April, 2015