PMXBOT Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Sunday the 6th of November, 2016

(Back to #pypa-dev overview) (Back to channel listing) (Animate logs)
[17:46:13] <dstufft> xafer: I'm going to cut a 9.0.1 release unless there's something you want me to wait on
[17:48:46] <xafer> dstufft: nope, it's good for me. Btw, did you see my comments on setuptools 27.3.0 ?
[17:49:19] <dstufft> xafer: um, don't think I did
[17:49:48] <xafer> https://github.com/pypa/pip/commit/d6a41466c8eec86ef403ac891df9528dde77e245#commitcomment-19711284
[17:50:27] <dstufft> erm
[17:50:31] <dstufft> not sure how I got the 27 number
[17:54:32] <dstufft> xafer: ^ should fix it either way :]
[17:55:45] <xafer> And concerning the automated vendoring, I'm (honestly) interested on why you prefer "invoke generate.authors/vendor" vs python update_authors.py/re-vendor.py, it's not like we'll be passing complex args/configuration to our tasks ?
[17:55:45] <xafer> But if you feel strongly about it, I can certainly add a task decorator :o
[17:59:24] <dstufft> xafer: well one thing is I don't want a whole bunch of different python scripts to execute, it's kind of clogging up the directory structure, and once you move them all to a single file if you want to allow calling only some of them and not all of them, or paraemtrize them at all (e.g. for a --dry option) you start basically re-inventing invoke
[18:01:54] <xafer> dstufft: I've never used it so I'll give it a go. Am I supposed to use invoke.run instead of subprocess.check_output then ? Any other things I need to change ?
[18:02:13] <dstufft> xafer: ctx.run now I think with the latest invoke
[18:02:37] <dstufft> ctx being the first argument apssed into the task function
[18:41:21] <xafer> dstufft: https://github.com/pypa/pip/pull/4093 now uses invoke, no idea if it is idiomatic invoke though.
[18:52:24] <dstufft> xafer: will take a look in a bit, thanks :]
[19:28:17] <dstufft> https://s.caremad.io/LFJNjAdPsq/ what version of pip (X.Y only) is hitting PyPI in the last 2 days, not counting today yet
[19:56:11] <xafer> dstufft: did you already use the difflib.restore ? I'm not sure it can do what we want. I thought about using https://pypi.python.org/pypi/patch also but maybe `git` might be simpler
[19:56:38] <dstufft> xafer: Nope never used difflib.restore, just looked like it could apply a patch
[19:57:25] <dstufft> just using `git apply` or whatever seems like a pretty sane idea
[19:57:38] <dstufft> I think it's reasonable to say you need git installed to work on pip :)
[19:58:59] <dstufft> I was originally going to suggest `patch`, the CLI tool but then I remembered windows doesn't have that :)
[20:02:13] <xafer> Yup :) But I agree that having patches in a separate directory would be much cleaner than my SPECIAL_CASES tuple :o
[20:16:43] <xafer> dstufft: https://github.com/pypa/pip/commit/d86d1713647f791979b9267ffc5773479d0ef469 contained _vendor files :-p
[20:17:07] <dstufft> er
[20:17:09] <dstufft> I am a moron
[22:37:27] <xafer> dstufft: when you have time for a last review, I'd love to see https://github.com/pypa/pip/pull/4093 merged, to check it off my list.
[22:47:14] <dstufft> heh
[22:47:25] <dstufft> 9.0.1 is #6 most used pip on PyPI so far today