pmxbot IRC Log Viewer

[00:03:00] <lifeless> dstufft: where is the markerlib bugtracker?

[00:03:19] <dstufft> lifeless: um, not sure? it might just be part of setuptools now

[00:04:22] <lifeless> do you happen to know if there is a bug open about the lack of < and <= in marker support today?

[00:04:32] <lifeless> 426 specifies it, but its not implemented :)

[00:05:15] <lifeless> I can't see one, so I'll create it

[01:03:56] <dstufft> lifeless: I don't think so, though I also think that we should make comparison operators function sanely for markers

[01:04:02] <dstufft> lifeless: right now, they rely on string comparison

[01:07:09] <lifeless> dstufft: it compiles to python in pkg_resources

[01:07:20] <lifeless> dstufft: so it should be just a matter of accepting the operators AFAICT

[08:06:35] <lifeless> welp figured out my test suite performance

[08:06:40] <lifeless> Downloading http://pypi.region-b.geo-1.openstack.org/packages/3.4/s/setuptools/setuptools-15.2-py2.py3-none-any.whl (501kB)\n'

[08:06:52] <lifeless> during the make virtualenv

[08:07:20] <lifeless> I'm guessing the index ignoring is not quite right in my branch :)

[08:09:19] <lifeless> (and I have no idea why its taking 83 seconds to download that file but there you go

[08:19:43] <lifeless> dstufft: would you be ok with a replacement venv for testing that makes one venv then stashes it and cp's, rather than doing the whole actual venv logic each time ?

[08:20:01] <lifeless> [not to fix this issue, just as a general, tune-this-thing approach]

[08:28:17] <lifeless> ahahahahahahaha

[08:28:20] <lifeless> found it

[08:36:26] <lifeless> dstufft: you'll laugh. You'll cry. You'll issue a CVE.

[08:37:31] <lifeless> yeah 6 seconds, thats better

[09:49:17] <lifeless> dstufft: fspkg 0.1.dev0 incompatible with ===0.1.dev0

[09:49:26] <lifeless> dstufft: that surprises me, in install_as_egg

[09:50:48] <lifeless> thats from if not specifier.contains(candidate) where specifier is a SpecifierSet, and candidate is a Version

[09:55:30] <lifeless> welp, thing for tomorrow, but I bet this is the source of many if not most of the failures

[19:47:08] <lifeless> dstufft: morning

[19:48:07] <dstufft> lifeless: yea

[19:48:08] <dstufft> er

[19:48:15] <dstufft> lifeless: heya*

[19:48:22] <dstufft> lifeless: sorry, doing a deploy heh

[19:49:29] <lifeless> dstufft: kk; so I need some input on the odd x.dev===x.dev -> False

[19:49:46] <lifeless> dstufft: also the virtualenv bug I listed may be worth a cve

[19:49:57] <lifeless> which is of course what you need just now

[19:50:42] <dstufft> lifeless: gimme a moment to put away pypi a sec, deployed the start of migrating pypi-prod to S3 and it shook loose a bug or two

[19:50:56] <dstufft> should have them fixed now though

[19:51:06] <lifeless> totally

[19:51:08] <lifeless> no rush

[19:51:17] <lifeless> just putting up a queue for when you get ycles

[19:53:42] <lifeless> right now I'm cooking brekkie for family etc ;)

[19:54:29] <lifeless> dstufft: oh and I should plug https://rbtcollins.wordpress.com/2015/05/13/improving-dependency-handling-upstream-for-openstack/ <- my TODO lits

[19:54:32] <lifeless> list

[19:56:58] <dstufft> lifeless: in 5 minutes or so I have to go pick up Alyssa from the bus stop, I'll be back in 20-30 minutes after that and then, unless another fire pops up, I'm free to do anything

[20:01:22] <lifeless> dstufft: cool

[20:01:35] <lifeless> dstufft: I'll be fully on deck in ~80m

[20:36:57] <dstufft> lifeless: ok, I'm back now

[20:41:52] <lifeless> cool

[20:41:57] <lifeless> I'm just about to run C down to kindy

[20:42:06] <lifeless> but the two current q's for you are in backscroll

[20:42:15] <lifeless> (=== behaviour and venv . vulnerability)

[20:58:52] <dstufft> lifeless: I'm not sure I grok the == behavior, in the packaging library 0.1.dev0 == 0.1.dev0 is True - unless you mean pip is overriding that somewhere

[21:03:20] <dstufft> lifeless: the virtualenv thing, sigh yea

[21:03:31] <dstufft> that goes back to 2009 when ianb first added pip installation

[21:12:45] <lifeless> dstufft: so I don't know why, but it looks like === is returning False for parse_requirements('foo===0.1.dev0')[0].specifier.contains(parse_version('0.1.dev0'))

[21:13:12] <dstufft> lemme try that exact code

[21:13:27] <lifeless> its not quite exact, making a copy-pastable one now

[21:13:47] <lifeless> list(parse_requirements('foo===0.1.dev0'))[0].specifier.contains(parse_version('0.1.dev0'))

[21:13:51] <lifeless> False

[21:13:57] <lifeless> list(parse_requirements('foo==0.1.dev0'))[0].specifier.contains(parse_version('0.1.dev0'))

[21:14:00] <lifeless> True

[21:14:06] <lifeless> since === is meant to be the escape hatch

[21:14:36] <lifeless> I use it when things are pinned down - e.g. by file path requirements, or when reconstructing InstallRequirements after resolution

[21:14:37] <dstufft> oh

[21:14:38] <dstufft> wait

[21:14:41] <dstufft> === not ==

[21:14:51] <lifeless> yes

[21:14:52] <dstufft> let me try that

[21:14:56] <lifeless> list(parse_requirements('foo==0.1.dev0'))[0].specifier.contains(parse_version('0.1.dev0'))

[21:14:59] <lifeless> True

[21:15:00] <lifeless> bah, sorry

[21:17:06] <lifeless> the virtualenv thing; is that entrenched now? Seems like its just plain broke

[21:18:10] <dstufft> no I agree

[21:18:19] <dstufft> it shouldn't be grabbing things from random directories you happen to have on pwd

[21:19:11] <lifeless> k

[21:19:17] <lifeless> should I put up a patch

[21:21:00] <dstufft> sure

[21:21:04] <dstufft> I see why the === doesn't work

[21:21:34] <dstufft> I forgot to include === in the lists of specifiers which are inclusive

[21:21:42] <lifeless> should it? Like, is this a bug in the plumbing ?

[21:21:45] <dstufft> so it doesn't detect that ===0.1.dev0 allows pre-releases

[21:22:01] <lifeless> do I fix it in this branch, or do you land something in develop that I rebase on ?

[21:22:03] <dstufft> so when you pass in 0.1.dev0, it short circuits because it's a pre-release

[21:22:13] <dstufft> it's a bug in packaging, so I'll fix it there real quick

[21:31:19] <lifeless> cool

[21:32:21] <dstufft> lifeless: https://github.com/pypa/packaging/pull/32

[21:35:47] <dstufft> lifeless: I'll cut a release shortly and then pull the new version into develop

[21:38:34] <lifeless> kk

[21:49:40] <dstufft> lifeless:

[21:49:41] <dstufft> ^

[21:54:33] <lifeless> dstufft: cool

[21:54:43] <lifeless> I just edited the file locally for now

[23:07:05] <lifeless> dstufft: wow @ the fails on that upgrade

[23:20:18] <lifeless> oh I hope you don't mind, but I've proposed a talk about pip's internals and this stuff to pyconau

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Wednesday the 13th of May, 2015