Log file Viewer

[07:33:30] <[Tritium]> I need to install a version of pip which supports 3.2... I dont seam to be able to do that

[07:41:28] <[Tritium]> I figured it out, ignore that

[10:35:03] <zmo> hi!

[10:35:21] <zmo> is there a way to prevent people from installing a package with python2?

[10:35:41] <zmo> currently I do a test within my setup.py to halt if sys.version_info is wrong

[10:35:56] <zmo> but it would be nice if pip would tell, nope dude, just don't

[12:15:17] <KeyWeeUsr> hello! pip ignores the first requirement while building docs on RTD. any ideas why? https://readthedocs.org/projects/krysa/builds/4378797/

[12:17:06] <[Tritium]> KeyWeeUsr: this is something you have to ask the RTD people. They run pip with --no-download. And I am POSITIVE they are unhappy you want to run cython on their servers

[12:19:31] <KeyWeeUsr> Tritium: their irc channel is dead, no answer on github issue too, so I came to ask here (because of pip). also, I run cython, because package that's required for my docs needs cython even in mocked-up version

[12:19:49] <KeyWeeUsr> is there some kind of priority sorting of requirements from .txt file

[12:20:36] <KeyWeeUsr> it _worked_ before, but since ~9days ago it just throws an error and nothing else.

[12:21:06] <KeyWeeUsr> and cython _is_ installed on other projects, which is weird... otherwise I'd just think they forbid cython entirely

[12:21:14] <[Tritium]> Did you change requirements file or setup.py?

[12:21:33] <KeyWeeUsr> requirements no, wait a sec I'll check setup.py

[12:21:52] <KeyWeeUsr> no, 14days ago, so that's probably not the reason

[12:21:57] <KeyWeeUsr> https://github.com/KeyWeeUsr/KrySA/blob/master/setup.py

[12:22:26] <KeyWeeUsr> last successful build https://readthedocs.org/projects/krysa/builds/4358435/

[12:27:40] <[Tritium]> honestly, there is nothing there that can be debugged from this end. you kinda need to wait for someone at RTD to wake up

[12:28:09] <[Tritium]> (its 4-8 am on a weekend for many of them)

[12:28:49] <[Tritium]> The support team is on US pacific time, and its 4:18 AM there

[12:29:02] <[Tritium]> er... 5:18

[12:30:17] <KeyWeeUsr> oh, nice to know, anyway I can see my old posts in irc logs, so not much activity there and the issue is 3 days old :/

[12:46:13] <zmo> hi again… I'm getting bitten by an issue with uritemplate.py's rename to uritemplate

[12:46:40] <zmo> basically @sigmavirus24 renamed the project, and made uritemplate.py have a requirement to uritemplate

[12:46:49] <zmo> it works nice when installing with pip

[12:47:24] <zmo> but when I'm installing this directly with setuptools (either through python setup.py or zc.buildout) then it's totally breaking

[12:47:31] <zmo> https://travis-ci.org/guyzmo/git-repo/jobs/158945200#L354

[12:48:31] <zmo> I told sigmavirus24 about this, and asked him to actually change the dependency on his project (github3.py) that has a requirement to uritemplate.py (so it's now uritemplate), so the issue gets away

[12:48:57] <pombreda> KeyWeeUsr: "Statistical analysis for rats." :D ... I am not sure I would want get that :P

[12:48:58] <zmo> but I find it quite inconsistent that dependencies are understood differently from installing with pip or installing with setuptools

[12:49:33] <KeyWeeUsr> pombreda: well, it's not _with_ rats :P (though some are included as images) :D

[12:50:18] <pombreda> KeyWeeUsr: I mean, hum, as a possible user, I would not want to be compared to a rat. No offense meant to rodents of course

[12:50:46] <KeyWeeUsr> hm, good note

[12:50:53] <zmo> pombreda - in Douglas Adams books, the rats is said to be the most intelligent species on earth

[12:50:59] <zmo> dolphin being second, and humans only third

[12:51:02] <KeyWeeUsr> and they are! ^^

[12:51:13] <pombreda> zmo: I think these were mices, not rats?

[12:51:14] <KeyWeeUsr> but nowadays is almost everyone butthurt

[12:51:35] <pombreda> zmo: sry, I menat KeyWeeUsr

[12:51:55] <zmo> pombreda - hm… that was a while back, I remembered rats, but maybe you're right

[12:52:13] <zmo> that == when I read the books

[12:52:34] <KeyWeeUsr> pombreda: no, I'm sure those were rats. my friend tought them some fancy stuff

[12:53:14] <KeyWeeUsr> https://www.goodreads.com/quotes/475365-these-creatures-you-call-mice-you-see-they-are-not what....

[12:53:34] <pombreda> KeyWeeUsr: http://hitchhikers.wikia.com/wiki/Rat

[12:53:44] <pombreda> http://hitchhikers.wikia.com/wiki/Mice

[12:54:09] <pombreda> I guess wee are sloping away from pypa quickly :D

[12:54:14] <pombreda> *we

[12:54:19] <zmo> 🙇

[12:54:21] <zmo> ok

[12:55:06] <KeyWeeUsr> I think I'll reword the description a little bit. Still rats are awesome creatures :P

[12:56:40] <KeyWeeUsr> pombreda: anyway, I've already mentioned inteligence there https://github.com/KeyWeeUsr/KrySA/blob/master/README.md ^^

[13:02:08] <pombreda> KeyWeeUsr: you sure can handle your project prop anyway you want :P

[13:41:19] <ninjaaron> Is there a place where all possible arguments to setuptools.setu() are listed? I frequently use the py_modules attribute, for example, and I just discovered it isn't documented anywhere in setuptools docs. I had to go digging through distutils to find any additional information. Makes me wonder what else I'm missing out on.

[14:28:06] <ronny> gah

[14:28:33] <ronny> dstufft: manylinux has a massive flaw :( its opt out instead of opt-in

[15:29:19] <pombreda> binary compat is a difficult matter ;)

[15:29:39] <pombreda> ronny: I can understand your point there

[15:37:38] <pombreda> ronny: too recent or too old?

[15:38:50] <pombreda> imho the problem is not really with manylinux, but with the docker image which is like a really ancient centos 5, so old that there many things that do not even build on it by default like lxml for instance

[15:39:15] <pombreda> but the oldness is part of the compat deal I guess

[15:55:19] <ronny> pombreda: neither, incompatible with FHS

[15:57:02] <pombreda> FHS ?

[15:57:09] <ronny> filesystem hierachy standard

[15:57:15] <ronny> nixos is a differentb east there

[17:09:06] <pombreda> ronny: :) you are on nixos... neat

[17:12:29] <dstufft> ronny: it's a bit hard right, of we make manylinux1 opt-in then very few people will use it even though most of them would be perfectly fine with it, if we amke it opt-out then most people will use it but a few people might get weird situations where it doesn't work

[17:12:51] <dstufft> ronny: my suggestions are to opt out yourself, and if this is something detectable, improve pip's detection code to auto opt-out

[17:17:02] <pombreda> that sounds a sensible approach

[17:17:32] <pombreda> I can see more issues with opt-in than opt-out indeed for the larger user base

[17:19:12] <pombreda> dstufft: frankly except for edge, off broadway use cases I am quite happy with manylinux1. And handling these few edge cases my self is fine by me

[17:21:55] <ronny> dstufft: i think its a good idea to need following the FHS

[17:23:34] <ronny> dstufft: also it should be possible to set it in pip.conf?