Log file Viewer

[00:22:04] <_habnabit> is there a pypi server that can automatically build wheels?

[00:22:29] <_habnabit> or something i could integrate with pypiserver/devpi to automatically build wheels

[00:54:06] <dstufft> _habnabit: there's a thing that will build wheels and upload it to devpi

[00:54:22] <_habnabit> dstufft, is it https://github.com/blue-yonder/devpi-builder or something else?

[00:54:31] <dstufft> yea I think that's it

[00:54:36] <_habnabit> ah okay

[00:54:40] <_habnabit> i'll have to look into it

[01:38:07] <Yasumoto> whoa, that looks awesome

[01:38:21] <Yasumoto> we're trying to think through how to build something like that

[01:38:25] <Yasumoto> that's huge :)

[02:07:28] <buck1> pypi seems sad/slow

[02:09:05] <dstufft> buck1: are you in APAC

[02:09:16] <buck1> don't believe so

[02:09:26] <dstufft> Oh

[02:09:28] <dstufft> Hmm

[02:09:28] <dstufft> http://status.fastly.com/incidents/2hy5x4nkqjf2

[02:09:38] <dstufft> that's only in APAC it appears though

[02:09:42] <buck1> since i don't know what that is

[02:09:44] <dstufft> unless people started attacking other DCs

[02:10:13] <dstufft> Asia-Pacific

[02:10:30] <dstufft> e.g. do you live in asia/australia-ish area

[02:11:37] <tdsmith> hey dstufft: there's a change in master on distlib I'd like to see bundled into pip; when would Vinay have to make a distlib release for it to have a high probability of landing in the next pip release?

[02:11:45] <buck1> dstufft: i'm in SF

[02:16:37] <dstufft> tdsmith: anytime before the next pip release really, there isn't a defined date for that though :/

[02:16:37] <dstufft> buck1: ok , probably hitting SJC DC then, which isn't marked as effected

[02:17:22] <buck1> i got a "real name of requirement coverage is coverage"

[02:17:31] <buck1> which i think is the case when it can't list simple/coverage

[02:19:07] <tdsmith> okay, thanks!

[02:27:51] <dstufft> buck1: yea

[02:28:10] <buck1> i tried to factor that one out before =/

[02:28:25] <buck1> dstufft: seems like your proposal to deprecate it was successful

[02:28:41] <buck1> the requirement-casing double check

[02:29:03] <dstufft> looks like the fastly status got upgraded to other DCs and transit across all DCs

[09:43:30] <doismellburning> I cannot express sufficiently just how much easier pex looks like it will make my life

[09:44:20] <doismellburning> I have one client building chroots with mock to virtualenv inside; another spinning up Docker instances, another that does python-package-fetch-and-install at _deploy time_

[11:45:50] <dAnjou> hi, i'm getting log output http://paste.ubuntu.com/8985494/

[11:46:16] <dAnjou> what's happening? the internet says i should use pip 1.2.1 but i'm already using that

[11:49:28] <dAnjou> this is what jenkins tells me http://paste.ubuntu.com/8985529/

[11:58:19] <dAnjou> i must sound like an idiot ...

[12:51:37] <dAnjou> nvm, it mysteriously works again

[13:01:42] <mgedmin> dAnjou, you may want to pip install -U pip; this seems like a very old version if it defaults to HTTP and not HTTPS

[13:01:52] <mgedmin> unless you've got a config file somewhere that overrides the default index URL?

[13:04:18] <dAnjou> well, never change a running system

[13:04:20] <dAnjou> ;)

[13:04:56] <mgedmin> heh

[13:05:29] <dAnjou> but yeah, this thing definitely needs some work

[21:14:13] <gchristensen> Hi, when I do: pip install git+ssh://git@github.com:account/repo.git it runs the git command: /usr/bin/git clone -q ssh://git@github.com:account/repo.git /tmp/pip-Uq2SiI-build which fails to authenticate using my SSH key. oddly enough, when I run that same git command just without the ssh:// at the start, it connects fine and clones ok. is there a way to make pip not prefix the clone url with ssh://, or has

[21:14:16] <gchristensen> anyone seen this / know a way to resolve it? FWIW I'm using a read-only deploy key for this repository. (cc EWDurbin, warmwaffles told me to ping you.)

[21:14:43] <EWDurbin> oh man, git deps

[21:14:46] <EWDurbin> pip version?

[21:15:30] <gchristensen> 1.54

[21:15:32] <EWDurbin> gchristensen: an example of a working git dep i have on hand is:

[21:15:33] <gchristensen> 1.5.4*

[21:15:47] <EWDurbin> -e git+ssh://git@github.com/kissmetrics/data-thing.git@0.9.0#egg=data-thing

[21:15:53] <EWDurbin> in a requirements.txt

[21:16:03] <EWDurbin> git deps are hard :-x

[21:17:17] <gchristensen> it is worth noting that I'm not attached to git deps, I'm just not sure of a good way to let my devs have private deps which can be resolved in our production packaging environment and locally

[21:18:39] <gchristensen> hrmm -e looks interesting

[21:19:02] <gchristensen> I guess the core of my issue, is prefixing the git repo with ssh:// is somehow preventing my ssh keys from being used for authentication :|

[21:31:37] <gchristensen> EWDurbin: turns out git+ssh://git@github.com:account/repo.git should have been git+ssh://git@github.com/account/repo.git

[21:31:53] <EWDurbin> ahhhhhhh

[21:32:21] <gchristensen> what was like, a good 4 hours on that colon.

[21:34:44] <nicksloan> haha. yikes

[21:43:16] <dstufft> it wouldn't be python packaging if we didn't punish you for minor mistakes with little ot know useful information

[21:43:25] <dstufft> gchristensen: open a ticket about how crappy this failure condition is?

[21:44:43] <gchristensen> can-do

[21:45:32] <gchristensen> yeah ... while we're at it, looking for a "this is the 1-2-3 for making a pip-compatible package" guide was pretty disappointing as well ...

[21:46:04] <dstufft> packaging.python.org has some stuff

[21:46:08] <dstufft> I think

[21:46:32] <dstufft> if there's things missing from there, there's a github link and that repo would love issues (or even better, PRs) too

[21:55:40] <gchristensen> dstufft: how about this? https://github.com/pypa/pip/issues/2124

[21:56:09] <dstufft> lgtm

[21:56:30] <gchristensen> great

[22:05:59] <tomprince> gchristensen: Note that the ssh:// syntax (without the leading 'git+') is also supported by git.

[22:07:58] <gchristensen> tomprince: that is something I came across

[22:08:01] <gchristensen> yes

[22:08:29] <gchristensen> before I found the : vs / error, I had asked questions in #git and ##linux going deeper and deeper into the stack (none of them were able to help much :P)