PMXBOT Log file Viewer

Help | Karma | Search:

#pypa logs for Wednesday the 17th of September, 2014

(Back to #pypa overview) (Back to channel listing) (Animate logs)
[00:00:02] <buck1> ammended with a comment
[00:00:35] <buck1> dstufft: elarson: https://github.com/ionrock/cachecontrol/pull/41/files
[00:02:15] <buck1> test a bit cleaner, now, too
[00:06:08] <dstufft> this bug has been plaguing the pip test suite for awhile heh
[00:06:25] <dstufft> we kept "fixing" but it kept poping back up
[00:06:30] <buck1> dstufft: do i have your +1 now?
[00:06:55] <dstufft> buck1: does self.__getattribute__("_CallbackFileWrapper__fp") work? or does it need the object__getattribute__?
[00:07:08] <buck1> uhm
[00:07:21] <buck1> i dont believe objects have a getattribute
[00:07:23] <buck1> checking
[00:08:20] <buck1> ah seems to work. i'll update.
[00:08:39] <elarson> buck1: I agree with dstufft that doing self.getattribute feels cleaner. is that the change that works?
[00:09:15] <dstufft> elarson: sorry my code has been so edge casey :((
[00:09:40] <buck1> elarson: it works, updating now
[00:10:00] <buck1> ammended https://github.com/ionrock/cachecontrol/pull/41
[00:10:05] <elarson> dstufft: no worries. I think it is a good use case to avoid caching when the response was never consumed.
[00:11:06] <buck1> elarson: final answer ^
[00:11:13] <dstufft> buck1: oh, do the pip test suites work if you delete the self.__calback = None in cachecontrol?
[00:11:18] <dstufft> I added that trying to fix this problem before
[00:11:24] <dstufft> I wonder if it's no longer needed with your patch
[00:11:34] <buck1> the real question is do they fail without the patch
[00:16:04] <buck1> trying it now..
[00:16:15] <buck1> without both patches
[00:16:41] <buck1> if nothing fails, i'd rather leave it be
[00:20:22] <buck1> dstufft: yea there's no test that fails by removing the callback=None
[00:20:33] <buck1> i'd like to leave it alone, for my branch
[00:28:21] <Yasumoto> buck1: so you guys figured it out? O_o
[00:28:32] <buck1> Yasumoto: close enough
[00:28:52] <Yasumoto> heh
[00:28:54] <buck1> i still dont understand gc, but there was an infinite __getattr__ recursion in cachecontrol
[00:29:01] <Yasumoto> dang
[00:29:09] <dstufft> basically
[00:29:13] <buck1> Yasumoto: https://github.com/ionrock/cachecontrol/pull/41
[00:29:16] <dstufft> the GC would delete self.__fp
[00:29:18] <Yasumoto> is there an issue/ML post/comment to track?
[00:29:20] <Yasumoto> ah
[00:29:26] <dstufft> then something else would try to access something on the object
[00:29:30] <buck1> what is ML
[00:29:34] <dstufft> and the object's __getattr__ referenced __fp
[00:29:47] <Yasumoto> *mailing list
[00:29:49] <Yasumoto> (sorry)
[00:29:51] <dstufft> so it would recurse forever into various __getattr__ calls trying to get the __fp instance
[00:30:05] <buck1> only took me six hours to figure out
[00:30:12] <Yasumoto> dstufft: roger
[00:30:14] <buck1> =(
[00:30:24] <Yasumoto> how to give virtual beer/coffee/tea/water ?
[00:30:41] <buck1> anyone bay area?
[00:31:05] <Yasumoto> indeed, actually- I'm in SF
[00:31:19] <Yasumoto> (but won't make it to this week's SF Python Meetup, sadly)
[00:31:54] <buck1> Yasumoto: the meetup is at my office :)
[00:32:32] <buck1> erk except not this week
[00:33:15] <Yasumoto> ah, you're at yelp, right? I went when Guido gave his asyncio talk- super awesome
[00:33:26] <dstufft> Philadelphia ftl :(((
[00:33:55] <Yasumoto> dstufft: we just need to figure out how to fly you out to get some #pypa-related talks
[00:34:15] <Yasumoto> (or you host some pypa mini-conf over there.. :)
[01:55:31] <dm7freek> I think I've found a bug on PyPI, am I in the right place?
[02:01:16] <Alex_Gaynor> dm7freek: yup
[02:01:47] <dm7freek> PyPI doesn't seem to allow login if your password contains a colon.
[02:02:36] <Alex_Gaynor> dstufft: ^
[02:02:53] <Alex_Gaynor> What kind of error do you get?
[02:05:53] <dm7freek> It keeps asking for credentials so I'm guessing Incorrect Password? Perhaps the sanitizer is overzealous?
[02:06:26] <dm7freek> If I reset my password to contain only upper and lowercase letters and a number, i can log in
[02:07:09] <Alex_Gaynor> dm7freek: pff, I think I see the issue
[02:08:00] <elarson> buck1, dstufft: re cachecontrol PR #41, just to be sure I understand the commentary correctly, the name munging we are talking about is the use of _CallbackFileWrapper__fp right?
[02:08:17] <Alex_Gaynor> dstufft: https://bitbucket.org/pypa/pypi/pull-request/50/dont-explode-on-passwords-with-colons/diff
[02:08:23] <Alex_Gaynor> dm7freek: ^ I suspect that fixes the issue
[02:08:41] <Alex_Gaynor> I just randomly grepped some code and sent a PR without ever testing it though lol, so someone should seriously review it
[02:09:34] <dm7freek> I could see how that would do it haha right on!
[02:09:44] <Alex_Gaynor> elarson: I mean, I loaded up webui.py in my browser and used cmd-f :-)
[02:27:28] <dm7freek> Just out of curiosity (I know I could figure this out on my own, but I don't feel like resetting my password), can you log in to PyPI using the email address on file instead of username?
[02:32:14] <elarson> dstufft, buck1: did you guys need another cachecontrol release?
[02:32:54] <elarson> if so, just let me know and I'll push one out
[02:57:00] <buck1> elarson: a release would let us sqash this bug in the pip test suite, yes
[03:02:44] <buck1> elarson: you seem to be wrapping your comments to 70 columns?
[03:02:59] <buck1> erm, period. is that true?
[03:11:09] <buck1> Alex_Gaynor: do you have access to the apache errors? You should see that line in stack traces ifyour hypothesis is correct.
[03:11:46] <Alex_Gaynor> buck1: I don't, but if you look, it catches ValueError, so there's nothing in the logs
[03:11:52] <buck1> oh.
[03:13:19] <buck1> Alex_Gaynor: the comment seems to indicate a 1:1 between ValueError and not-exactly-one-colon, i wonder if it should go away
[03:13:35] <buck1> elarson: is asleep i suppose
[03:13:43] <Alex_Gaynor> good catch
[03:14:52] <buck1> hah i meant the ValueError, but you did the right thing
[03:14:58] <buck1> a bad hash could have no colon.
[05:47:14] <buck1> dstufft: i looked at this a bit more, and IOBase is real weird: http://paste.pound-python.org/show/mBCJDztX1aZbshpWEfuh/
[05:47:33] <buck1> how can i stuff attrs onto it if there's no __dict__
[05:48:06] <buck1> it defines setattr, but it's just object.__setattr__
[05:49:45] <Alex_Gaynor> it has a dict, but it's in C
[05:49:59] <Alex_Gaynor> I didn't realize vars couldn't acess it, you might want to file a bug on bugs.python.org
[05:50:26] <Alex_Gaynor> actually, looks like this is fixed in python3.4
[05:50:28] <Alex_Gaynor> so don't bother
[05:56:18] <buck1> but i'm using 2.7 :(
[05:56:45] <buck1> Alex_Gaynor: where did you get the 2.8 tshirt
[11:24:15] <FirePowi> Hello, is there a problem with irclib ? https://pypi.python.org/simple/irclib/
[11:25:10] <FirePowi> !logs
[11:25:10] <pmxbot> http://chat-logs.dcpython.org/channel/pypa
[14:13:12] <elarson> buck2, dstufft: cachecontrol 0.10.4 is out
[14:19:57] <Alex_Gaynor> buck2: matt mackall was giving them out at pycon
[21:55:13] <buck1> dstufft:
[21:55:16] <buck1> https://github.com/pypa/pip/pull/2033
[21:55:23] <buck1> https://github.com/pypa/pip/pull/2032
[22:12:25] <buck1> dstufft: getting this failure from the pypa/develop branch http://paste.pound-python.org/show/3AvRMOa9TtifAG5AOIC5/
[22:12:59] <buck1> dev became dev_r0
[22:21:07] <buck1> Yasumoto: ^ ?
[22:21:13] <Yasumoto> yo
[22:21:22] <Yasumoto> want me to try to repro?
[22:21:33] <buck1> would be a good start
[22:21:43] <Yasumoto> ;)
[22:22:24] <buck1> there are a bunch of similar tests in taht file, asserting without the _r0
[22:22:27] <buck1> they pass.
[22:27:37] <Yasumoto> buck: hm
[22:27:42] <Yasumoto> https://gist.github.com/Yasumoto/ebb88c8b10d44b331558
[22:27:59] <Yasumoto> is there a branch I should grab?
[22:28:18] <Yasumoto> i'm on OS X 10.9 if that changes anything for my effectiveness
[22:38:43] <buck_> Yasumoto: any luck, my client derpped
[22:40:07] <buck1> Yasumoto: dstufft if i remove the tag_svn_revision=True from its setup.cfg, the test passes again
[22:43:31] <buck1> ... i'm going to check it in like that for now, to unblock myself
[23:32:45] <Yasumoto> ah
[23:32:49] <Yasumoto> buck1: I said
[23:32:52] <Yasumoto> 15:24 Yasumoto: https://gist.github.com/Yasumoto/ebb88c8b10d44b331558
[23:32:55] <Yasumoto> 15:24 Yasumoto: is there a branch I should grab?
[23:32:58] <Yasumoto> 15:25 Yasumoto: i'm on OS X 10.9 if that changes anything for my effectiveness
[23:39:43] <buck1> Yasumoto: naw it was failing for me on develop
[23:39:47] <buck1> i'm on osx...
[23:39:55] <buck1> 10.9
[23:39:58] <Yasumoto> hmm
[23:40:06] <buck1> our homebrew is slightly different prob
[23:40:15] <Yasumoto> you said svn? I pulled it from github
[23:40:34] <buck1> the test suite uses svn
[23:41:08] <buck1> Yasumoto: for discussion https://github.com/pypa/pip/pull/2045
[23:41:18] <buck1> Yasumoto: you have svn available on your box?
[23:51:19] <Yasumoto> ah, gotcha
[23:51:23] <Yasumoto> yah, I do
[23:51:40] <Yasumoto> buck1: I gotta run out, but I'll check back either later tonight or tomorrow
[23:53:08] <buck1> Yasumoto: thanks
[23:57:26] <tdsmith> Hi #pypa (and buck1)! What is my range of expectations for a binary C extension built against a UCS2 Python 2.7 and imported from a UCS4 Python 2.7? Does this always work? Always fail? Sometimes fail mysteriously?