Log file Viewer

[08:06:09] <ronny2> glyph: say, is there a channel for automate

[08:06:16] <ronny2> eh automat

[13:23:25] <daniel90> I have an issue installing a package with pip

[13:23:29] <daniel90> See this snippet: https://dpaste.de/qP1d#L

[13:24:51] <daniel90> It tells me "disk quota exceeded" but I'm almost certain that is not the case. As you can see in the snippet the destination folder lies on the filesystem "/scratch" which has 1.4P space available and I don't have a quota on that drive

[13:26:16] <dstufft> daniel90: might be a quota on the temporary directory?

[13:27:30] <daniel90> dstufft: At least `quota` doesn't tell me that there is one on /tmp. And it is the source of the copy operation

[13:27:43] <dstufft> mmm yea that's true

[13:28:37] <dstufft> I'm not sure I'd ever seen that error, maybe try dropping a pdb in there and see what files it's failing on? Make sure the underlying assumptions are correct wrt what files are being touched

[13:28:59] <apollo13> also try a manual cp

[13:30:29] <daniel90> dstufft: I put a print output in python3.4/shutil.py directly before the failing line, that's where the 'src: <filename>, dst: <filename>' output in my snippet comes from

[13:30:47] <daniel90> apollo13: I wanted to try a manual copy but pip seems to clean up the temporary directory

[13:31:01] <gchristensen> a 1.6P scratch partition, incredible!

[13:31:01] <dstufft> ah, missed that line

[13:31:16] <daniel90> gchristensen: yeah, it's not your everyday machine :D

[13:31:32] <dstufft> daniel90: I *think* --no-clean will stop it from cleaning up

[13:31:45] <gchristensen> by the looks of it, it isn't an every day cluster of machines either.

[13:31:46] <daniel90> dstufft: let's try that...

[13:32:16] <dstufft> (I say I think, because our temp file handling is kind of janky and not all of the tempf iles we make respect that flag)

[13:39:51] <daniel90> ok, it doesn't seem to be a pip issue, a manual copy also fails

[13:40:20] <daniel90> Or trying to create a bigger file with `dd`

[13:40:24] <daniel90> I have no idea why htough

[13:43:40] <mgedmin> too much free space overflows into negative numbers? :P

[13:43:50] <mgedmin> something something inodes rather than bytes?

[13:45:18] <apollo13> yeah, df with inodes should help too

[13:45:49] <apollo13> maybe you reached a limit there, but then it should not complain about quotas but rather disk full

[13:47:31] <daniel90> I guess I found the issue. The creation of new files works on /scratch but not in the specific subfolder of the virtual environment. That subfolder has me as the owner but a wrong group. That group has a quota of 10M whereas my user has a quota of 0.

[13:47:46] <daniel90> One would think that user quotas overwrite group quotas, but who knows...

[13:47:48] <apollo13> that would do it

[13:47:48] <mgedmin> haha

[13:49:25] <daniel90> I'll try to recreate that virtual environment

[13:54:11] <daniel90> It worked :) Thanks for your help

[18:01:32] <flupke> Hi, I'm looking for the function that normalizes distribution keys (e.g. 'foo.bar' -> 'foo-bar'), where is it defined in pip?

[18:07:31] <dstufft> flupke: https://packaging.pypa.io/en/latest/utils/#packaging.utils.canonicalize_name

[18:08:10] <flupke> dstufft: thank you very much

[18:08:49] <dstufft> flupke: no problem!

[18:15:53] <ngoldbaum> dstufft: ugh, just yelled at sean about being a jerk for no reason on the pypi issue tracker. Sorry you have to deal with this crap...

[18:30:39] <dstufft> ngoldbaum: :) It's OK. I'm used to some level of push back on things (and if Bitbucket had the capability I would have just blocked it)

[18:31:36] <ngoldbaum> dstufft: might be difficult with him since he works there....

[18:32:18] <mgedmin> ask him to implement user blocking on bbucket!

[18:32:44] <Wooble> bitbucket never implements new features for the issue tracker. How would they sell any Jira that way?

[18:33:02] <ngoldbaum> ^ this is the #1 reason why the bitbucket issue tracker is terrible and will never imporve

[18:33:08] <ngoldbaum> for reals

[18:33:13] <ngoldbaum> atlassian doesn't let them

[18:36:22] <mgedmin> butbutbut Jira is also terrible!

[18:37:41] <dstufft> ngoldbaum: I don't genreally use Mercurial much, mostly because I only have enough room in my head to be really proficient at only one VCS at a time, and Git has the mindshare so I go there :)

[18:38:07] <ngoldbaum> dstufft: sure, i can use git and hg just fine, i just don't like to use git :)

[18:38:47] <ngoldbaum> it cost a huge amount of social capital for me to propose moving projects i work on from hg to git, even if that might be a net benefit in the end since new contributors are more likely to know git at this point

[18:39:01] <ngoldbaum> so i just grit my teeth and deal with bitbucket's crappyness

[18:40:48] <dstufft> ngoldbaum: yea, I was involved in the discussion and such for moving CPython to Git (and Github), the pain is something I'm well aware of :)

[18:51:16] <Wooble> mgedmin: yeah, we've had it here for a few weeks and I'm hating it. (also, self-hosted bitbucket, which oddly doesn't support hg)

[18:51:52] <mgedmin> (self-hosted bitbucket doesn't support README.rst files, which is mind-boggling to me)

[18:52:21] <ngoldbaum> self-hosted bitbucket and cloud bitbucket share zero code

[18:52:33] <ngoldbaum> it was purely a marketing decision to give them the same name

[18:52:38] <ngoldbaum> which is really confusing to people

[18:52:46] <ngoldbaum> since you'd think that they have similar featuresets

[18:52:47] <ngoldbaum> nope

[18:53:30] <Wooble> oh good, I look forward to being confused. Fortunately I was the only one here using a VCS at all before so I don't also have to deal with other people's confusion.