Log file Viewer

[02:21:12] <dstufft> zackc: you shouldn't be getting 403's, if you try it again it works?

[04:08:47] <jeromelanteri> how to have the sources of packages from pip ?

[09:52:56] <chris2> hi. i'm experiencing an issue with pypi mirrors/s3

[09:53:16] <chris2> i can download http://pypi.python.org/packages/source/A/Attic/Attic-0.15.tar.gz using curl, but -0.16 doesnt work

[09:53:30] <chris2> somehow the http -> https rewriting is broken i think

[11:27:13] <dstufft> chris2: you'e right

[11:27:17] <dstufft> chris2: I'll get it fixed

[11:27:33] <chris2> thanks :)

[11:27:53] <chris2> kinda had me stumped because the old links still worked :)

[14:21:08] <ztane> HTTP Error 400: Invalid classifier "Programming Language :: Python :: 3.5"

[14:21:10] <ztane> :D

[14:22:12] <ztane> dstufft: can you fix? :D

[14:25:43] <ztane> agronholm: added "Programming Language :: Python :: 3.5" in trove classifiers for Tonnikala, now I guess I need to revert that edit since PyPI is not accepting :P

[14:26:17] <agronholm> ztane: what, pypi rejects your package because of that??

[14:26:35] <ztane> yes,

[14:26:42] <ztane> invalid classifier, you know :D

[14:26:51] <agronholm> damnit

[14:27:15] <ztane> just tested with a4 and fixed 1 outstanding bug, I thought "nice to show we're on the edge" :D

[14:27:47] <ztane> and pypi answers with HTTP Error 400: Invalid classifier "Programming Language :: Python :: 3.5"

[14:27:51] <agronholm> I'm already using PEP 484 type hints in my framework :)

[14:29:12] <ztane> I can't for the templating language, still supporting python 2.6 or 2.7, though I am contemplating on dropping the support, there is very minimal risk that I'd be using that on Python 2.x and I'd not give a **** if anyone else wanted to use it on 2 :D

[14:29:54] <agronholm> new projects won't use python 2.6

[14:30:01] <agronholm> so you could drop that at least

[14:30:18] <agronholm> if you care about a maximal audience, you will want to support 2.7 though

[14:31:17] <ztane> I don't

[14:31:26] <ztane> they'd complain about bugs

[14:31:42] <agronholm> unlike 3.x users?

[14:31:59] <ztane> well 3.x users would be complaining about bugs that were worthwhile to fix

[14:32:12] <ztane> 2.x users would be complaining about bugs in the 2.x support :D

[14:32:26] <agronholm> well...naturally

[14:32:54] <ztane> in this case 2.6 requires 1 extra fix at least that I remember...

[14:33:06] <ztane> 1 test uses ordereddict

[14:33:18] <ztane> and 1 module

[14:33:24] <ztane> not a big deal

[16:27:35] <sontek> If I'm using parse_requirements from pip, whats the best way to get the actual line that pip will be using? I've found if the requirements file has 'foo>2.0' then str(r.req) will give me that

[16:27:59] <sontek> but if it has '-e git+git@foo.com#egg=foo' then it just gives me foo instead of the editable line

[16:28:49] <sontek> I'm working with pip.req.req_install.InstallRequirement objects

[16:57:02] <sontek> I see I can check `is_editable` and then use URL if it is set and name if it isn't but I feel like I'm just missing something in the API

[18:18:22] <tos9> sontek: (this isn't an answer to your question but just in case you didn't know, AIUI that is not a public AIP)

[18:18:25] <tos9> *API

[18:22:06] <sontek> tos9: Yeah, unfortunately the options are using a non-public API or write my own =)

[18:22:25] <sontek> tos9: but I traced back the history to 2008 and the API hasn't changed outside of adding a `PipSession` so it is a very stable API

[18:24:59] <dstufft> sontek: look at what pip freeze does, it gets it back out somehow

[18:28:08] <tos9> sontek: well, I mention that because I also make use of it anyhow, and it's broken me before :)

[18:28:12] <tos9> (within the past 2 months)

[18:28:48] <sontek> tos9: Yeah, I'm kind of worried about it but I think its the best solution. This is for tox to do dependency resolution

[18:29:18] <tos9> sontek: You're modifying tox itself? I still don't understand why tox doesn't use pip personally but probably I am underinformed

[18:30:31] <sontek> tos9: It will start using pip after this, it did use pip in 1.9.1 but I broke other features, so trying to make it not break things so that it can continue using pip

[18:30:38] <sontek> tos9: I have it all working except for editables

[18:31:29] <sontek> But once I get an `InstallRequirement` it doesn't seem like I can easily get the `-e git://...` line back, I just have to manually construct it

[18:57:24] <sontek> dstufft: as a side note, I saw a tweet recently about search changing and now I'm seeing weird results: https://pypi.python.org/pypi?%3Aaction=search&term=SQLAlchemy&submit=search

[18:57:39] <sontek> dstufft: I would expect SQLAlchemy to be the highest weighted project in that result

[19:28:49] <sontek> tos9: dstufft: https://bitbucket.org/hpk42/tox/pull-request/160/use-pip-to-parse-requirements-so-we-can/diff if you want to see what I'm attempting to use pip's parse_requirements for

[19:29:31] <tos9> sontek: cool I'll have a look (although what I meant before is I don't understand why tox doesn't use pip to install deps)

[19:29:59] <tos9> tox's dep resolution never works right, I have to use -r fairly frequently to tell it "no, you're wrong."

[19:30:20] <dstufft> tox does use pip

[19:30:47] <tos9> dstufft: to do the installs, or for everything?

[19:30:56] <dstufft> tos9: to install everything

[19:31:05] <tos9> the last time I looked (and from experimental current experience), it doesn't use pip to check "are all the deps satisfied"

[19:31:08] <sontek> tos9: We pretty much exclusively use -r for tox deps=.. line

[19:31:09] <tos9> it implements its own thing.

[19:31:16] <dstufft> the problem with tox is that it does virtualenv caching which is useful for fast tests, but when it fails to figure out something changed it will skip trying to install

[19:31:21] <tos9> right

[19:33:10] <sontek> That PR I just sent should help with the resolution. The problem is that tox did have its own dep parsing but if you had `-r` it would fall back to pip. Now those theoretically will match up