pmxbot IRC Log Viewer

[00:00:15] <dstufft> tchaypo: lol

[00:16:50] <tchaypo> how did you end up working on pypa stuff? Did it start as as just another yak?

[00:22:22] <pmxbot> jaraco pushed 1 commit to setuptools (https://bitbucket.org/pypa/setuptools/) :

[00:22:22] <pmxbot> Conditionally remove __PYVENV_LAUNCHER__

[01:45:45] <dstufft> lifeless: is https://github.com/pypa/pip/issues/2888 fixed now?

[01:46:20] <lifeless> let me see

[01:46:30] <lifeless> yes

[01:46:38] <lifeless> that was tchaypo's patch

[01:52:38] <dstufft> lifeless: if you're looking for something to do, https://github.com/pypa/pip/issues/2683 would be great. the next version of OSX has "rootless" mode (basically like SELinux afaik) and --user is gonna be important there soon I think since people won't be able to upgrade things soon without it. If not I'll poke at it tomorrow

[01:54:23] <lifeless> I can put my hands on a mac

[01:54:28] <lifeless> totally out of practice though

[01:55:28] <lifeless> dstufft: or maybe, I think I know someone that has a mac and uses it even.... tchaypo

[01:55:48] <tchaypo> yo

[01:56:04] <dstufft> it's not exactly related to a mac, you just need a ~/.pydistutils.cfg that specifies a prefix option

[01:56:05] <tchaypo> next version meaning el capitan?

[01:56:32] <dstufft> the error just happens alot on a mac because Homebrew is popular and Homebrew python ships a distutils config that has a prefix option

[01:56:33] <dstufft> tchaypo: yea

[01:56:55] <lifeless> tchaypo: there's a smallish patch dstufft would like in 7.1 that hasn't been written yet

[01:57:14] <lifeless> tchaypo: since 7.1 is on our critical path

[01:57:23] <lifeless> tchaypo: I'm looking to help pip get there

[01:58:41] <tchaypo> so to check my reading of the bug

[01:59:08] <tchaypo> the way to reproduce is to have ~/.pydistutils.cfg that specifies a prefix, and then to install with —user

[01:59:37] <tchaypo> and the suggested fix is to have pip always add “—install-option=—prefix=“ to something when it’s given —user?

[01:59:51] <dstufft> tchaypo: yea

[02:00:09] <tchaypo> ah - have pip always act as though “—install-option=—prefix” was on the commandline

[02:00:45] <tchaypo> that seems simple enough that I can probably get it done today, and getting 7.1 out the door is important enough that I can pretend I’m doing something useful instead of just distracting myself from harder work

[02:02:26] <lifeless> anything pip is hard

[02:02:31] <tchaypo> I’ve never done any pip dev on my mac though, because too many of the tests fail because I’m missing things like svnadmin and don’t want to install them

[02:02:39] <lifeless> from the sounds of it you don't need to

[02:02:49] <tchaypo> in my head mac is a “it’s nice if it works there but we don’t really support it” platform

[02:03:05] <lifeless> add a phunctional test doing install with a .pydistutils.cfg in the temp homedir that the fixtures create

[02:05:46] <tchaypo> "ERROR: unknown environment ‘py35'"

[02:05:48] <tchaypo> thanks tox

[02:06:10] <tchaypo> hrm. I was hoping that upgrading tox would fix that

[02:06:32] <tchaypo> oh well. ‘tox -epy27’ it is

[05:28:07] <tchaypo> https://www.irccloud.com/pastebin/BOX9KFIP/

[05:28:15] <tchaypo> I’m hoping that I’m just seeing mirror skew

[06:05:41] <tchaypo> I’m not seeing it any more, although I’m not sure if that’s due to mirror updates or due to getting a new vagrant box.

[06:30:15] <pjdelport> tchaypo: Have you tried to apt-get update before running that again?

[06:30:30] <pjdelport> Sometimes Vagrant images seem to need that after a fresh start.

[06:31:21] <tchaypo> Yes, I had -but this wasn’t a fresh vagrant image

[06:34:30] <pjdelport> What does aptitude say?

[06:34:40] <pjdelport> Maybe the current package state is broken?

[06:35:12] <tchaypo> sorry, I can’t check - it went away after I did a vagrant box update

[06:35:28] <tchaypo> I’m not sure if it was the box update or the mirror getting up to date

[07:52:43] <lifeless> tchaypo: ubuntu seems to be borked

[07:52:52] <lifeless> tchaypo: trove are seeing that in tests too

[07:53:01] <tchaypo> awesome

[07:53:14] <tchaypo> well, in my case, a fresh install helped

[07:53:21] <tchaypo> perhaps a dist-upgrade would have helped me too

[08:03:41] <tchaypo> Nakato: do you still have an environment where all pip tests pass?

[08:05:01] <Nakato> tchaypo: Any repo in-particular?

[08:05:44] <tchaypo> https://github.com/pypa/pip

[08:06:36] <Nakato> Oh derp, checking

[08:06:56] <tchaypo> I think you mentioned getting one running the tests cleanly last week, is all

[08:07:32] <tchaypo> I’ve currently got

[08:07:32] <tchaypo> tests/functional/test_install_user.py ....FFFF.

[08:07:40] <tchaypo> and another F on test_uninstall_user

[08:08:25] <Nakato> I never bothered fixing anything that failed because HG was missing. But I've got that installed now. So lets see.

[08:28:09] <Nakato> tchaypo: tests/functional/test_install_user.py ......... Py34

[08:28:21] <Nakato> One failure on test_freeze though, not sure what yet

[08:29:23] <tchaypo> doing better than me. You’d be running on arch I assume?

[08:29:47] <Nakato> tchaypo: Yea, but it's under a virtualenv with tox as usual

[08:31:18] <Nakato> uninstall_user is fine too

[08:33:25] <tchaypo> Nakato: are your 27 failures in test_install_user?

[08:33:51] <Nakato> tests/functional/test_freeze.py and tests/unit/test_finder.py

[08:34:01] <tchaypo> heh. they pass for me :)

[08:34:17] <Nakato> IOError: [Errno 13] Permission denied: '/usr/lib/python2.7/site-packages/voluptuous-0.8.7-py2.7.egg-info/dependency_links.txt'

[08:34:18] <Nakato> E IOError: [Errno 2] No such file or directory: '/var/adm/inst-log/info'

[08:37:04] <Nakato> tchaypo: What are the errors out of yours?

[08:54:16] <Nakato> tchaypo: Py34 succeeds on my system

[09:38:36] <tchaypo> running again, will pastebin output when it’s done

[11:46:00] <tchaypo> Nakato: https://dpaste.de/igFC

[16:40:23] <dstufft> tchaypo: Nakato pip's tests are so bad

[16:40:40] <dstufft> I almost never bother to run them locally

[20:46:40] <tchaypo> source .tox/py27/bin/activate; py.test tests/functional/test_install_reqs.py

[20:46:56] <tchaypo> at least it means I don’t waste my time running tests that I know are going to fail

[20:49:40] <tchaypo> oh goody

[20:50:49] <tchaypo> oh, and that’s what .pth files do

[20:53:01] <tchaypo> Can not perform a '--user' install. User site-packages are not visible in this virtualenv.

[20:54:09] <tchaypo> So in order to test this, I need to make pip believe it’s not running in a virtualenv, and give it a user site-packages folder it can play with

[20:54:12] <tchaypo> wheeeeeeeee

[21:01:33] <pmxbot> jaraco pushed 1 commit to setuptools (https://bitbucket.org/pypa/setuptools/) :

[21:01:33] <pmxbot> Update changelog. Along with f44e81f2f62f, fixes #401.

[21:03:10] <pmxbot> jaraco pushed 3 commits to setuptools (https://bitbucket.org/pypa/setuptools/) :

[21:03:10] <pmxbot> Bumped to 18.0.1 in preparation for next release.

[21:03:10] <pmxbot> Added tag 18.0.1 for changeset e364795c1b09

[21:03:10] <pmxbot> Bumped to 18.0.2 in preparation for next release.

[21:19:51] <lifeless> dstufft: tchaypo: Nakato: found a new quirk in constraints

[21:20:15] <lifeless> with a constraints file that includes THING===VERSION

[21:20:34] <lifeless> pip install -c constraints.txt -e /path/to/thing

[21:20:42] <lifeless> installed everything but THING itself

[21:20:50] <lifeless> I think this is a bug

[21:20:50] <tchaypo> Hah

[21:20:53] <tchaypo> Useful

[21:21:06] <lifeless> I'm not sure how to generalise the desired override behaviour here though

[21:21:16] <lifeless> e.g. should 'cli arguments' win

[21:21:35] <lifeless> or should 'editable specifiers override version specifiers'

[21:21:59] <tchaypo> I think coo arguments

[21:22:01] <lifeless> we could say 'don't have THING in your constraints file like that'

[21:22:06] <lifeless> and in devstack just edit out THING

[21:22:39] <tchaypo> I think CLI arguments

[21:22:43] <lifeless> I'm not sure :)

[21:22:49] <lifeless> seeking thoughts

[21:23:01] <tchaypo> As a user, I told you to install thing

[21:23:27] <tchaypo> The other arguments are there to help you figure out how to handle stuff around thing, but thing is the thing that I want

[22:02:16] <lifeless> tchaypo: thats one argument

[22:02:20] <lifeless> tchaypo: on the flip side

[22:02:24] <lifeless> tchaypo: its much harder to reason about

[22:02:42] <lifeless> tchaypo: I think I'd be more comfortable with this erroring

[22:02:51] <lifeless> silently doing something odd is hard on users

[22:02:58] <tchaypo> True

[22:03:37] <tchaypo> I'd be much happier with "your instructions aren't clear, try again" then "your instructions aren't clear so I made a guess, good luck!"

[22:04:58] <dstufft> in the face of ambigiuity, refuse the temptation to guess

[22:05:09] <dstufft> setuptools forgot that particular rule, and look where that left us :)

[22:15:56] <lifeless> I think it would be ok to add 'it errors' as a 7.1.1

[22:16:13] <lifeless> but if tchaypo or Nakato could pick up making this error as a thing that would be super helpful

[22:16:22] <lifeless> I'm filing an issue now

[22:17:38] <Nakato> lifeless: Making this error as a thing?

[22:18:16] <lifeless> Nakato: https://github.com/pypa/pip/issues/2928

[22:18:23] <lifeless> Nakato: closing that ticket

[22:18:31] <Nakato> lifeless: Roger

[22:18:50] <lifeless> right now I need to unwedge the reverse checks within infra

[22:18:57] <lifeless> then I can come and help here again

[22:39:03] <tchaypo> I'm not likely to be able to pick it up today or tomorrow

[22:39:19] <tchaypo> But it sounds like Nakato can?

[22:39:35] <Nakato> tchaypo: lifeless: yep

[23:18:58] <lifeless> dstufft: hey

[23:19:08] <lifeless> dstufft: why would freeze have -f in it ?

[23:19:18] <lifeless> http://logs.openstack.org/31/194931/4/check/check-requirements-integration-dsvm/909ed3b/console.html

[23:19:21] <dstufft> dunno

[23:19:25] <dstufft> I don't know the freeze code well

[23:19:47] <lifeless> k

[23:19:58] <lifeless> I think its because we have a PIP_FIND_LINKS env variable

[23:20:04] <lifeless> I'm going to try trashing that stuff ;)

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Wednesday the 24th of June, 2015