pmxbot IRC Log Viewer

[00:17:40] <msabramo> Why can I not get the installed_version property of some InstallRequirement objects?

[00:17:45] <msabramo> (Pdb) req

[00:17:45] <msabramo> <pip.req.req_install.InstallRequirement object at 0x103c6ff90>

[00:17:46] <msabramo> (Pdb) req.name

[00:17:46] <msabramo> 'translationstring'

[00:17:46] <msabramo> (Pdb) req.installed_version

[00:17:46] <msabramo> *** OSError: [Errno 2] No such file or directory: '/var/folders/gw/w0clrs515zx9x_55zgtpv4mm0000gp/T/pip-build-M0Abpt/translationstring/pip-egg-info'

[00:18:31] <msabramo> in fact, calling the pkg_info method is what causes this error

[00:18:33] <msabramo> (Pdb) req.pkg_info()

[00:18:33] <msabramo> *** OSError: [Errno 2] No such file or directory: '/var/folders/gw/w0clrs515zx9x_55zgtpv4mm0000gp/T/pip-build-M0Abpt/translationstring/pip-egg-info'

[00:18:54] <dstufft> uh

[00:18:58] <dstufft> that's strange

[00:19:00] <dstufft> I don't know

[00:21:46] <msabramo> it was installed from a wheel if that matters

[00:22:15] <msabramo> (Pdb) req.url

[00:22:16] <msabramo> u'https://pypi.python.org/packages/2.7/t/translationstring/translationstring-1.3-py2.py3-none-any.whl#md5=81bd24a147483dffc8faf0dc91c5673d'

[00:23:40] <msabramo> ok this is probably a clue

[00:23:48] <msabramo> (Pdb) req.pkg_info()

[00:23:48] <msabramo> *** OSError: [Errno 2] No such file or directory: '/var/folders/gw/w0clrs515zx9x_55zgtpv4mm0000gp/T/pip-build-cHKTz5/translationstring/pip-egg-info'

[00:23:58] <msabramo> ❯ ls -l /var/folders/gw/w0clrs515zx9x_55zgtpv4mm0000gp/T/pip-build-cHKTz5/translationstring

[00:23:58] <msabramo> total 0

[00:23:58] <msabramo> drwxr-xr-x 7 marca staff 238 Dec 16 16:20 translationstring/

[00:23:58] <msabramo> drwxr-xr-x 8 marca staff 272 Dec 16 16:20 translationstring-1.3.dist-info/

[00:24:14] <msabramo> there is no pip-egg-info but there is a .dist-info

[00:24:34] <dstufft> msabramo: possibly that method just doesn't work with wheels

[00:24:59] <msabramo> how do you get the version of a wheel then?

[00:25:17] <dstufft> dunno!

[00:25:22] <dstufft> I normally use pkg_resources

[00:29:08] <msabramo> ok I'll dig around

[00:35:56] <msabramo> https://github.com/pypa/pip/blob/develop/pip/req/req_install.py#L415

[00:36:03] <msabramo> that's where it fails

[12:03:11] <xafer> any chance https://setuptools.pypa.io could be updated or redirected to https://pythonhosted.org/setuptools/ ?

[16:50:42] <DaZ> hai

[16:52:27] <DaZ> in my humble opinion this is kind of borked https://github.com/pypa/pip/blob/master/pip/util.py#L53, mostly by the fact that there are only three values in that tuple :u

[16:53:00] <DaZ> is there a chance for a quick fix, or nobody cares about windows users and we have to wait until devel branch gets less developy

[16:53:52] <DaZ> i'm new to this.

[16:55:01] <sigmavirus24> DaZ: there's an inherent strawman in that argument - the idea that pip developers don't care about windows users. There is a release of pip planned for this week or next.

[16:55:40] <sigmavirus24> That release includes the fix you're looking for and the quick fix is entirely unnecessary at this point given the imminent release of the next version of pip (which will fix that)

[16:56:18] <DaZ> i'm joking obviously, i know you love all your subjects equally ;3

[16:56:33] <sigmavirus24> Trying to claim to be shunned as windows user is kind of a terrible thing to do too because there are several people who specifically work on making sure windows users have the best possible pip experience

[16:56:38] <DaZ> yes, i've seen it fixed in devel branch, good to hear

[16:56:45] <sigmavirus24> You're insulting their effort (that is mostly done in their free time) and it's not cool

[16:58:33] <DaZ> this was never my intenion

[17:08:05] <msabramo> anyone see this problem with using virtualenv to create py3 envs on OS X? https://gist.github.com/msabramo/b3a12a570e49fd1a1858

[17:08:35] <msabramo> This used to work fine and I'm pretty sure I !@#$'d something up with my config but can't figure out what

[17:11:18] <msabramo> possibly relevant: https://bugs.launchpad.net/ubuntu/+source/python-virtualenv/+bug/1156704

[17:14:58] <msabramo> and https://bitbucket.org/ambv/configparser/issue/5/installing-configparser-is-dangerous

[17:21:26] <msabramo> @dstufft: While you're doing Travis stuff, want to set up Travis for pypa/readme?

[17:22:19] <msabramo> https://travis-ci.org/pypa/readme

[17:22:44] <msabramo> I have a few PRs for you as well - https://github.com/pypa/readme/pulls

[17:23:24] <dstufft> msabramo: done

[17:23:48] <dstufft> msabramo: I know, I'm mostly ignoring warehouse / readme until after pip 6

[17:23:51] <dstufft> only so many hours :/

[17:23:57] <msabramo> \o/

[17:23:59] <msabramo> no problem

[17:24:24] <msabramo> so pip 6 is shorthand for pip 1.6?

[17:24:38] <msabramo> not jumping from 1.5.6 to 6.0 I assume? :)

[17:28:17] <msabramo> ah, configparser that was it!

[17:28:22] <dstufft> we are jumping from 1.5.6 to 6

[17:28:27] <dstufft> dropping the leading 1.x

[17:29:07] <msabramo> oh interesting

[17:29:09] <msabramo> ❯ pip freeze | grep -i configparser

[17:29:09] <msabramo> configparser==3.3.0.post2

[17:29:31] <msabramo> shame on me for installing in global site-packages

[17:29:44] <sigmavirus24> tsk tsk tsk

[17:29:46] <sigmavirus24> =P

[17:30:03] <msabramo> <- administering self-flagellation

[17:30:20] <msabramo> holy crap, I've had that problem for weeks

[17:30:54] <msabramo> toying with the idea of sending a PR to virtualenv to check for configparser - that's probably silly though

[17:30:55] <sigmavirus24> msabramo: just don't defenestrate yourself. that would be a bad idea.

[17:31:09] <sigmavirus24> msabramo: yeah, snowflakes like that aren't a good idea to try to maintain

[17:31:35] <msabramo> not always good to create a PR out of anger

[17:32:15] <msabramo> wish there was a way to defend against hanging myself like that though

[17:32:33] <msabramo> me doing stupid things is pretty much going to be a constant

[17:33:51] <dstufft> msabramo: the best is when you get to do stupid things with critical pieces of infrastructure

[17:33:52] <dstufft> like pypi

[17:34:07] <dstufft> then everyone notices when you do something stupid and accidently take it down :V

[17:34:47] <msabramo> doh

[17:35:45] <msabramo> been there done that

[17:35:52] <msabramo> not pypi but other things

[17:38:11] <msabramo> ah the configparser/virtualenv bug was fixed in configparser 3.5.0 — https://pypi.python.org/pypi/configparser/

[17:40:17] <msabramo> unfortunately I had 3.3.0r2 in my local devpi

[17:42:19] <msabramo> oh no that's not it

[17:42:48] <msabramo> it wasn't because I had the old one in my devpi, it's because configparser 3.5.0 doesn't exist yet; there is only configparser 3.5.0b2 and that's a pre-release so pip wasn't picking that up

[17:43:43] <carreau> Hi PyPa-dev. I am trying to hack on warehouse, is there any dumy (or real) data I can use to populate postgres warehouse table ?

[17:44:46] <dstufft> er

[17:44:47] <msabramo> carreau: I wondered same thing in https://github.com/pypa/warehouse/issues/343

[17:44:52] <dstufft> I have some laying around here somewhere

[17:45:07] <dstufft> a db dump from PyPI with the PII removed

[17:45:22] <msabramo> best would be some abstraction that lets you avoid postgres and elasticsearch

[17:45:23] <carreau> msabramo: yes that me the other on the issue :-)

[17:45:26] <msabramo> but that's a lot of code

[17:45:38] <msabramo> carreau: ha ha, oops :-)

[17:45:56] <msabramo> didn't look at the name

[17:46:01] <dstufft> msabramo: -1 on those abstractions, because they ultimately mean that you're limited to whatever features the abstraction has not waht features postgresql has

[17:46:07] <carreau> Indeed, I tried, and also found an issue speeking on mock for test. Was judged too complicated

[17:46:49] <msabramo> dstufft: well yeah if you're ok with being tied to postgresql

[17:46:59] <dstufft> Warehouse has one deployment target

[17:47:01] <carreau> dstufft: the goal of the mock is for non-backend people to be able to work on HTML, so just a static dataset wihtout correct querry woudl be fine

[17:47:42] <dstufft> It doesn't support MySQL or SQlite or anything else :D

[17:47:44] <msabramo> so it is not a goal for warehouse to be something that folks could deploy their own?

[17:48:26] <carreau> I'm not a postgres person either, but if dstufft could upload this somewhere that woudl be awesome. Maybe a bit of explanation on the ocumentation on how to use it bu we can take care of that

[17:48:28] <msabramo> not that I know if I would even want to do that; devpi seems like a good solution here

[17:48:47] <dstufft> msabramo: currently no. In the future, maybe but even then I'd probably want to mandate the tech stack

[17:49:09] <dstufft> I don't see much/any benefit to letting someone run on MySQL or sqlite when it means I can't take advantage of postgresql's features

[17:49:10] <msabramo> did you consider using devpi for PyPI? I'm guessing someone thought of it and it wasn't a good fit

[17:49:51] <msabramo> curious of what the reasons would be

[17:50:16] <dstufft> msabramo: mostly that devpi and pypi are two different targets

[17:50:19] <dstufft> like

[17:50:23] <dstufft> target audience

[17:50:48] <msabramo> in what way?

[17:50:57] <msabramo> audience for both is python devs

[17:51:23] <msabramo> I guess Warehouse is for external/public; what does it need to do differently to fit that?

[17:51:50] <msabramo> if it's just that Warehouse is better for a wider audience, then maybe devpi could get better at that

[17:52:24] <sigmavirus24> msabramo: devpi doesn't support HTTPS

[17:52:31] <msabramo> again I'm just curious; not seriously advocating a switch

[17:52:32] <dstufft> msabramo: well self registration is one, devpi also does not have per package ACLs

[17:53:09] <dstufft> WArehouse is designed up front with the assumption that Fastly (or varnish at least) is going to be in front of it

[17:53:13] <msabramo> sigmavirus24: but can't you frontend devpi with nginx or something to do HTTPS?

[17:53:42] <sigmavirus24> msabramo: I've been shouted down about suggesting people use devpi with https so I assume not

[17:53:53] <sigmavirus24> msabramo: I haven't tried it personally

[17:53:55] <dstufft> you can use https with devpi

[17:54:01] <dstufft> if you front it with something

[17:54:04] <msabramo> dstufft: ah per package ACLs

[17:54:04] <carreau> I also think that warehouse is nice. It should be more advertised and more developped. It just too hard to participate in dev for now I find.

[17:54:17] <sigmavirus24> warehouse is better written too I've found

[17:54:31] <dstufft> devpi's data model doesn't scale out to multiple machines, since it uses the local FS as a database

[17:54:38] <carreau> Trying to help though.

[17:54:38] <sigmavirus24> I tried to help with the md5 => sha256 migration on devpi as my first contribution and the code is neither well-factored nor easy to work with

[17:54:44] <msabramo> I guess conversely devpi has a bunch of stuff that Warehouse doesn't need like multiple indices

[17:55:02] <msabramo> sigmavirus24: yeah, I also found the code a bit unapproachable

[17:55:16] <dstufft> carreau: yea, sorry about that :/ I'm the primary dev on warehouse and I've been focusing on the next release of pip before I switch back to warehouse

[17:55:50] <msabramo> dstufft: need some people to help out with warehouse and relieve some of your burden?

[17:55:59] <carreau> No problem i know it's hard to be only dev on a project.

[17:57:12] <carreau> dstufft: if/when you have time if you can put some link/instructions on https://github.com/pypa/warehouse/issues/343 for postgres db, I can take care of writing some docs on how to do it.

[17:57:58] <dstufft> msabramo: well contributions are always welcome! I think my first thing back will probably be re-evalating more popular frameworks and deciding if the on-boarding ease makes up for the fact I hate their global state

[17:58:04] <carreau> msabramo: we all need nore dev on our projects :-)

[17:58:37] <tomprince> dstufft: klein doesn't have global state! :) Or doesn't need to have it, anyway.

[17:58:37] <dstufft> I should probably jsut commit a db dump into the repo with the top 100 packages or something

[17:59:42] <carreau> that would work also I guess, but would this expose some private data ?

[18:00:10] <carreau> maybe we can only export a few project/user that would accept...

[18:00:26] <msabramo> dstufft: https://github.com/pypa/warehouse/pulls :)

[18:00:45] <msabramo> yeah, I don't know if I love the thread-local stuff in Flask; I am a Pyramid fan myself

[18:00:57] <msabramo> if you use Pyramid, I will totally help hack on Warehouse

[18:01:23] <msabramo> +1 on db dump with top 100 packages

[18:01:30] <dstufft> tomprince: ha, that comes with it's own problems ;) mostly that async hurts people's brains and we probably won't take much advantage of that since we attempt to offload as much of the actual traffic to Fastly

[18:01:33] <msabramo> maybe even 1000

[18:01:59] <dstufft> carreau: it won't be a raw dump, I'd take it and remove any non-public data and replace it with place holder data

[18:02:23] <dstufft> msabramo: pyramid is on my list of things to evaluate, I've never used it. I went from Django -> Flask -> Werkzeug

[18:03:11] <msabramo> dstufft: we are very happy with Pyramid

[18:03:41] <dstufft> msabramo: you underestimate my ability to find things I hate

[18:03:42] <dstufft> :D

[18:03:55] <msabramo> it feels like it has a good combination of easy to get started and create a simple single file app, but it can also scale well to larger apps

[18:04:47] <tomprince> Although, for warehouse, being good for single-file doesn't matter.

[18:04:49] <msabramo> dstufft: :) well, no thread locals in Pyramid, so you can cross that one off your list

[18:05:08] <msabramo> Pyramid is very flexible and pluggable

[18:05:10] <carreau> :-P. Whatever is used is fine with me.

[18:05:21] <dstufft> msabramo: it totally has them, but it's an optional escape and not a core design so that doesn't make me very upset

[18:05:23] <dstufft> :D

[18:06:01] <msabramo> dstufft: Ah right, it does have some things that _can_ do thread locals but you don't have to use them and it's discouraged

[18:06:11] <dstufft> I'm gonna rewrite it in node.js, that's what all the cool kids are doing anyways

[18:06:32] <msabramo> bah

[18:09:06] <sigmavirus24> dstufft: A+

[18:09:10] <sigmavirus24> would use io.js again

[18:09:48] <sigmavirus24> dstufft: node is so webscale too

[18:09:51] <sigmavirus24> can we use mongo with it?

[18:10:11] <dstufft> only mangodb

[18:10:24] <dstufft> https://github.com/dcramer/mangodb

[18:10:35] <msabramo> CSV FTW

[18:10:46] <dstufft> does your database have an auto sharting algorithm? I don't think so

[18:11:14] <msabramo> dstufft: Hmmm not seeing anything still at https://travis-ci.org/pypa/readme/requests … ?

[18:11:25] <carreau> nah, cool kids move to go and rust now.

[18:11:26] <sigmavirus24> dstufft: https://github.com/mathias/yolodb

[18:11:45] <dstufft> msabramo: http://d.stufft.io/image/0Q2m0H3i3Y1X

[18:11:47] <sigmavirus24> carreau: I'm rewriting IronPython in rust this weekend

[18:12:04] <dstufft> I'm rewriting pip in rust

[18:12:12] <dstufft> pypi in node.js

[18:12:15] <dstufft> virtualenv in go

[18:12:28] <carreau> https://github.com/mame/quine-relay ?

[18:12:33] <msabramo> dstufft: hmmmm, maybe it takes time

[18:12:40] <msabramo> ugh, go

[18:12:41] <sigmavirus24> doismellburning: would like buildout rewritten in java so he can maven while he builds things

[18:12:51] <sigmavirus24> msabramo: travis has been backed up lots lately

[18:12:59] <sigmavirus24> (lately = ~last 2 weeks)

[18:14:06] <doismellburning> sigmavirus24: :P

[18:14:25] <msabramo> I wonder if pushing something to pypa/readme would wake travis up? because normally it displays something in yellow as soon as you push, even if the job doesn't run for hours

[18:14:32] <sigmavirus24> doismellburning: don't think I've forgotten your appreciation for how maven handles things!

[18:14:47] <sigmavirus24> msabramo: their hook processors could be backed up

[18:15:05] <msabramo> sigmavirus24: yeah that's possible too

[18:15:46] <msabramo> cool, https://github.com/pypa/pip/pull/2212 passed

[18:16:26] <msabramo> it's weird that "sudo: false" gives you a whole different infrastructure; not intuitive

[18:16:48] <msabramo> weird they didn't make it "docker: true" or something

[18:16:52] <sigmavirus24> looks like it sped up 32

[18:17:19] <msabramo> yay, I hate those long 32 jobs

[18:17:43] <msabramo> still long but I think twice as fast as it was

[18:19:59] <sigmavirus24> so better than nothing

[18:20:26] <dstufft> everything else is twice as slow

[18:20:36] <sigmavirus24> hah

[18:20:52] <sigmavirus24> new fastererer environments for your rubby

[18:20:54] <msabramo> dstufft: want to see if https://github.com/pypa/readme/pull/11 gets things moving?

[18:20:59] <sigmavirus24> not for your scaley python

[18:21:25] <msabramo> it's the GIL

[18:21:29] <msabramo> it's always the GIL

[18:21:40] <dstufft> I suspect the reason it's slower is because pip's tests are IO bound

[18:21:47] <msabramo> even when it can't be the GIL, it's the GIL

[18:22:57] <sigmavirus24> msabramo: even when the GIL has had its heart broken many times over, it's teh GIL

[18:23:23] <sigmavirus24> I'm really surprised the GIL hasn't just up and left

[18:24:42] <msabramo> dstufft: Thanks for merging the PR. Note that the tests might fail because of https://github.com/pypa/readme/pull/10 :)

[18:24:55] <dstufft> YOLO

[18:25:09] <sigmavirus24> dstufft: did you check out my friend's db that I linked you?

[18:25:16] <sigmavirus24> specifically you should look at the commit history

[18:25:19] <sigmavirus24> https://github.com/mathias/yolodb

[18:54:45] <xafer> still no feedback on https://github.com/pypa/pip/pull/2153 ? first commit should not be too controversial

[18:55:32] <sigmavirus24> xafer: every other commit should be though? ;)

[18:56:09] <xafer> the second one could :p

[18:56:22] <dstufft> xafer: I haven't had a chance to poke at it yet

[18:56:45] <dstufft> I should probably add it to the 6.0 milestone though

[18:57:20] <xafer> would be nice :)

[20:06:49] <msabramo1> dstufft: https://github.com/pypa/readme/pull/12

[20:16:25] <msabramo1> dstufft: Weird. asarih said in #travis: "I don't see any commit coming through for pypa/readme in the last 3 hours"

[20:17:15] <dstufft> msabramo1: heh

[20:17:17] <dstufft> I see why

[20:17:27] <dstufft> toggling it on in travis didn't turn on the push hook

[20:17:27] <msabramo1> oh?

[20:17:28] <dstufft> for some reason

[20:17:30] <sigmavirus24> dstufft: not turned on in github settings?

[20:17:31] <sigmavirus24> heh

[20:17:35] <msabramo1> ah

[20:17:42] <dstufft> maybe they changed it so it doesn't do that anymore

[20:17:52] <msabramo1> maybe you hit their limit; they used to have some limit for using their UI

[20:21:07] <dstufft> maybe fix now

[20:25:03] <msabramo1> dstufft: OK, try merging https://github.com/pypa/readme/pull/12

[20:25:18] <dstufft> yolo

[20:25:35] <msabramo1> yup

[20:25:37] <msabramo1> working

[20:27:35] <msabramo1> green - love it

[20:44:46] <carreau> dstufft: form earlier discussion on database dump, maybe https://github.com/emirozer/fake2db would be an alternative ... I was not able to make it work though

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Wednesday the 17th of December, 2014