[00:00:05] <glyph> dstufft: metasetup should have a command to _generate_ it, but then you don't edit it
[00:00:09] <dstufft> glyph: even if it looked like this http://bpaste.net/show/Ek5PD4tBBzsY7oTpCizh/
[00:00:41] <glyph> dstufft: don't lead with the solution, lead with the problem. why would I do this?
[00:01:41] <dstufft> glyph: So you're doing a bunch of hacky stuff to serialize modules and then reconstitute them inside of a setup.py, what if we just don't do that, and use the setuptools functionality (sort of) to do this
[00:02:06] <dstufft> I say sort of, because it requires a little bit of monkeypatch
[00:02:16] <dstufft> This is what d2to1, and pbr do
[00:03:08] <glyph> dstufft: that would be a good idea if setuptools worked
[00:03:12] <glyph> dstufft: but, too bad, it doesn't
[00:03:37] <glyph> dstufft: also, I don't see how that helps me at all with the plugin-generation case
[00:04:30] <dstufft> glyph: afaik this part of setuptools works
[00:06:31] <dstufft> monkeypatching distutils.core.Distribution or w/e the hell the location is, makes me feel far less janky than trying to serialize + reconstitute a bunch of modules
[00:06:56] <dstufft> it would work for setup.py develop && pip install -e
[00:07:51] <dstufft> hrm, the hard part would be hooking in for at_setup_time after setup_requires, but before setup()
[02:55:10] <Ivo> glyph: you assume I don't know that? This being considerate of people using project they're not maintainers of that have PIL as a dependency
[02:56:43] <Ivo> moving to pip 1.5 has made stuff that worked before quite a pain for a few people, while sometimes its judged to be necessary for improving things, its always nice to minimize that wherever possible
[02:57:08] <Ivo> I was hoping to track down contact details for PIL's maintainer, turns out he's a hard man to get an email address for
[03:39:55] <Ivo> still made a release for it in 2009 tho
[03:40:55] <dstufft> and he's maintained the download server
[03:41:12] <dstufft> so its not like he peaced out and left everyone to rot or anything
[03:41:20] <Ivo> how long does an 'important' package maintainer need to be completely afk for before we can think about updating their pypi page for them...