pmxbot IRC Log Viewer

[01:51:39] <Arfrever> dstufft: Could you upgrade pkg_resources bundled in pip?

[01:54:16] <Arfrever> dstufft: E.g. to allow to fix https://bugs.python.org/issue24320

[02:04:54] <lifeless> Arfrever: open a ticket in github/pypa/pip please; it should be straight forward

[02:11:50] <Arfrever> lifeless: dstufft should have already received notificational e-mail from above bug. If he is not reading e-mails, then opening of another bug will not help :( .

[02:13:06] <lifeless> Arfrever: certainly it will

[02:13:28] <lifeless> Arfrever: because you're asking for something to be done in the pip project, and thats a localised database which gets queried routinely

[02:13:39] <lifeless> Arfrever: and multiple people do patches based on it

[07:52:22] <lifeless> dstufft: oh hi :)

[17:14:48] <Arfrever> jaraco: ping

[17:14:59] <jaraco> Hi Arfrever

[17:15:13] <Arfrever> jaraco: Please see my comment in https://bitbucket.org/pypa/setuptools/commits/f572ec9563d647fa8d4ffc534f2af8070ea07a8b

[17:20:29] <jaraco> Arfrever: Thanks for raising that to my attention. I see you raised that to my attention promptly, but for whatever reason, I missed the notice.

[17:31:17] <pmxbot> jaraco pushed 3 commits to setuptools (https://bitbucket.org/pypa/setuptools/) :

[17:31:17] <pmxbot> Backed out unintended changes made in f572ec9, restoring use of 'imp' module for dynamic module creation originally committed in 06ac3674 and 4c121bd24f.

[17:31:17] <pmxbot> Merge fix

[17:31:17] <pmxbot> Update changelog

[17:39:21] <pmxbot> jaraco pushed 3 commits to setuptools (https://bitbucket.org/pypa/setuptools/) :

[17:39:22] <pmxbot> Bumped to 17.1.1 in preparation for next release.

[17:39:22] <pmxbot> Added tag 17.1.1 for changeset 4a0d01d690ff

[17:39:22] <pmxbot> Bumped to 17.1.2 in preparation for next release.

[17:56:15] <lifeless> dstufft: morning

[17:56:26] <dstufft> lifeless: heya

[17:56:29] <lifeless> dstufft: I'd like a ruling please on -c or not -c + the code merged :)

[17:56:53] <lifeless> dstufft: my final argument for -c is YAGNI: there's no other planned use for -c atm

[17:57:54] <dstufft> lifeless: it looks like marcus feels similarly to me, wary but not a strong opinion, so- first come first served i suppose :)

[18:08:45] <lifeless> dstufft: wicked, thanks

[18:55:51] <lifeless> dstufft: do you want me to rebase the patch or anything ?

[22:17:55] <lifeless> dstufft: hi; ping

[22:18:02] <dstufft> lifeless: heya

[22:18:27] <lifeless> dstufft: two things; a) can has merge, so I can forget about the -c patch [since you seemed to hve ruled now :)]

[22:18:34] <lifeless> dstufft: or do you need a rebase etc done ?

[22:18:50] <lifeless> secondly, the bug with no-binary in the config file - I'll take that on now unless someone else has

[22:19:40] <dstufft> yea I can merge that first one in a few, let me just read over it again

[22:22:54] <lifeless> thanks!

[22:23:56] <dstufft> lifeless: it occurs to me that constraints files could be used as an override mechanism rather than relying on the "if you specify X at a higher level, then that takes precidence" crap

[22:24:02] <dstufft> maybe

[22:24:03] <dstufft> not sure

[22:24:24] <lifeless> possibly

[22:24:42] <lifeless> I intend for them to do that in the resolver branch :)

[22:24:56] <dstufft> (maybe not, I gave it like 5s worth of thought, so maybe a crazy idea)

[22:25:11] <lifeless> so my plan all along is that these will become global conditions in the resolver

[22:25:26] <lifeless> which will short-circuit anything thats lookedup, and constraint the search space

[22:25:31] <lifeless> of course, requirements do that too

[22:26:23] <dstufft> so an override would be something like -> "foo depends on bar <1.0, but I want to install bar>4 anyways even though that's not techinally allowed"

[22:26:33] <dstufft> not sure if cosntraints are additive or instead of

[22:26:43] <lifeless> ah

[22:26:45] <lifeless> so

[22:26:57] <lifeless> I think we need to design a UI for that

[22:26:59] <lifeless> for the resolver

[22:27:11] <lifeless> mechanism will be that we need a lookaside of things that are overridden

[22:27:29] <lifeless> it could be syntax in the requirements/constraints files

[22:27:44] <lifeless> or it could be a new file that has the same basic syntax we have today

[22:28:02] <dstufft> yea

[22:28:21] <lifeless> like, imagine a new == operator

[22:28:31] <lifeless> ==== <- 'STOMP ON YOUR REQUIREMENTS'

[22:29:22] <dstufft> lol

[22:29:30] <dstufft> if all else fails, add more =

[22:29:42] <lifeless> hey, JS has four doesn't it ?

[22:29:49] <dstufft> we're obviously slacking

[22:31:01] <dstufft> lifeless: ^

[22:31:51] <lifeless> woo

[22:32:10] <lifeless> release release release. kidding. Mostly.

[22:34:28] <dstufft> I need to get around to getting all the pieces we need to automate releases

[22:35:17] <lifeless> dstufft: are they written up somewhere? you might get patches...\

[22:35:34] <lifeless> wow #2874 went south fast

[22:35:51] <dstufft> https://github.com/pypa/pip/milestones/Automated%20Release%20Process is the things I thought of when I was thinking about it

[22:36:27] <dstufft> lifeless: heh, yea. I'm surprised it took this long TBH

[22:36:29] <lifeless> opened for later preusing

[22:36:34] <lifeless> perusing

[22:37:07] <dstufft> lifeless: I have half a crazy idea to steal some bits from pbr, but I'd need to write patches to pbr because I don't want something that needs a setup_requires at install-from-sdist time

[22:37:14] <dstufft> install-from-dev time is ok

[22:37:29] <lifeless> dstufft: oh look, you @mentioned me :)

[22:37:36] <lifeless> so

[22:37:55] <lifeless> I'm strongly against mangling setup.py such that its contents are different in sdist vs vcs

[22:37:58] <lifeless> but

[22:38:11] <lifeless> I'd be entirely happy if we can come up with a way that avoids that specific misfeature

[22:38:25] <lifeless> e.g. a lookaside file and a setup.py template that knows how to grab it

[22:38:41] <lifeless> I'd be very happy to figure out a way to make you happy such that you can just use pbr

[22:39:25] <ionelmc> lifeless: is there a good overview of all the pbr features?

[22:39:36] <ionelmc> like comparatively, what you gain vs plain setup.py

[22:39:47] <lifeless> http://docs.openstack.org/developer/pbr/#

[22:40:00] <lifeless> ionelmc: http://docs.openstack.org/developer/pbr/#what-it-does

[22:40:17] <lifeless> ionelmc: its nearly all opt-in too, since users.

[22:40:50] <lifeless> ionelmc: the stuff around requirements.txt will squick you - but I'm rapidly converging that on setup.cfg anyhow; and since pip has no dependencies, its not an issue for pip

[22:41:25] <dstufft> I have complaints about pbr, the requirements.txt being one of them, but I think lifeless has heard that form me enough and there's plans to move that to setup.cfg so :D

[22:41:47] <lifeless> dstufft: preaching to the choir

[22:42:04] <ionelmc> sphing stub generator?

[22:42:05] <lifeless> takes a little bit of time to move a 1000 dev ship is all

[22:42:13] <dstufft> complaining about pbr was the reason I joined #openstack-infra

[22:42:13] <ionelmc> what a odd feature for pbr ....

[22:42:17] <lifeless> ionelmc: yeah, predates sphinx getting its own

[22:42:17] <ionelmc> sphinx*

[22:42:59] <lifeless> ionelmc: pbr started as 'we have lots of copy-pasted setup.py code lets consolidate'

[22:43:11] <lifeless> ionelmc: so when 2+ projects in openstack needed something, it lands in pbr

[22:43:44] <ionelmc> lifeless: does it still not support distributing modules?!

[22:43:49] <lifeless> ionelmc: and then, setup.py got made a fixed value in openstack, so anything projects need is done by teaching pbr about the concept, and setup.cfg to control.

[22:43:56] <dstufft> woo history

[22:43:57] <lifeless> ionelmc: I don't know what you mean

[22:44:13] <dstufft> lifeless: foo.py instead of foo/__init__.py I think

[22:44:27] <ionelmc> lifeless: like six.py

[22:44:37] <lifeless> ionelmc: oh, no idea. is there a bug open? I think it would be reasonable to do that.

[22:44:37] <ionelmc> it's a single file (a module, not a package)

[22:44:52] <ionelmc> dunno, i'm asking :)

[22:44:52] <lifeless> I don't know why it wouldn't work

[22:45:20] <ionelmc> lifeless: did pbr gain automatic package collection? ala setuptools.find_packages

[22:46:08] <lifeless> dstufft: I think 2871 can be closed

[22:46:14] <lifeless> ionelmc: years and years ago

[22:46:30] <lifeless> ionelmc: it sucks everything from git in

[22:46:32] <lifeless> by default

[22:47:04] <ionelmc> lifeless: so you only need to specify the top-level package?

[22:47:09] <lifeless> ionelmc: mmm, possibly not exactly what you mean. But happy for it to get it if it doesn't have it. Certainly I never need to change setup.cfg when subpackages are added etc

[22:47:43] <lifeless> ionelmc: yah

[22:48:40] <dstufft> I wish github let us open up ticket control to anyone

[22:48:44] <dstufft> not just contributors

[22:50:28] <lifeless> I'm a contributor

[22:50:30] <dstufft> (There's this weird thing where I really don't want to run our own infrastructure if I can help it, and github's UX is pretty awesome, but some of their constraitns makes mea bit sad and I wish I could have other options)

[22:50:32] <lifeless> I think you mean committer :)

[22:50:44] <lifeless> erm

[22:50:44] <lifeless> pusher

[22:50:46] <lifeless> whatever

[22:51:18] <dstufft> yea that thing

[22:51:30] <dstufft> Collaborators Github calls them I guess

[22:51:36] <dstufft> which is a silly name

[23:00:41] <lifeless> 2869 is fun. Its ez-install/etc stuff, so the subject could be tweaked

[23:01:00] <lifeless> I don't know if we hold our hands in the air like we don't care - its a 2008 distribution thats glitching

[23:03:05] <lifeless> 2868 is distribute workaround biting again

[23:11:24] <lifeless> dstufft: is there a helper for config file setting in tests?

[23:13:19] <dstufft> nope!

[23:13:50] <lifeless> ah there is

[23:14:01] <lifeless> found test_install_config

[23:14:39] <lifeless> its just horrid

[23:20:23] <lifeless> oh man

[23:20:26] <lifeless> update_defaults is evil

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Monday the 8th of June, 2015