Log file Viewer

[19:53:03] <pf_moore> dstufft: ping

[19:53:21] <dstufft> pf_moore: hi

[19:53:43] <pf_moore> hi - have you ever seen these sort of weird errors on travis? "No module named 'six'; 'pip' is a package and cannot be directly executed"

[19:54:27] <dstufft> pf_moore: normally it means there's an import error somewhere

[19:55:10] <pf_moore> Hmm, as simple as that? Wish it'd say that :-) Might be how I've vendored the retrying module, I'll take a look. Thanks

[19:55:22] <dstufft> pf_moore: sadly the import error is probably getting masked, it's hard to tell without throwing some debug crap in there

[19:55:30] <pf_moore> Yeah, figures

[19:56:00] <dstufft> pf_moore: it doesn't just say it because of some of the hacks that have gotten hidden over time, well pip/_vendor/_init__.py hides some import errors, and then python -m pip hides some too

[19:56:26] <dstufft> commenting out https://github.com/pypa/pip/blob/develop/pip/_vendor/__init__.py#L93 might get you the real error

[19:56:50] <pf_moore> Does Unix barf on CRLF line endings in modules? Because it seems to work OK for me...

[19:56:55] <dstufft> um

[19:57:01] <dstufft> I don't honestly know

[19:57:12] <dstufft> I don't even know how to type a CRLF line ending

[19:57:18] <pf_moore> OK, let me have a fight with git over CRLF translation, see if that helps

[19:57:39] <pf_moore> lol be glad - I wish I never had to see CRLF

[19:58:36] <dstufft> even when I'm on windows I configure my editors to just use \n

[19:59:47] <pf_moore> Yeah, I avoid CRLF, but git likes to add them and I never worked out how to tell it to stop (it's core.autocrlf or something, need to check the docs)

[20:01:53] <dstufft> pf_moore: it seems to work on my OSX box fwiw

[20:01:57] <dstufft> I checked out your PR

[20:03:27] <dstufft> there doesn't seem to be CRLF's in that file

[20:06:24] <pf_moore> dstufft: Ta, good to have that confirmed

[20:09:18] <dstufft> pf_moore: which means it may be something with how travis is setup

[20:10:01] <pf_moore> No, I'm getting it locally now (on the Python 3 Flake8 run, of all places). Looks like it probably is something I've done, leave it with me

[20:12:19] <dstufft> okay

[20:12:41] <pf_moore> dstufft: found it - retrying imports six, which was there on my local test (because I have it installed in my system python) but not in a "proper" test

[20:13:00] <dstufft> pf_moore: ah

[20:13:08] <dstufft> yea rewrite it to use from pip._vendor import six

[20:15:02] <pf_moore> Hah, it's obvious that we *need* to do this, but I was taking the "we don't change vendored code" to heart, and never thought of it!

[20:15:14] <pf_moore> New build running now, here's hoping :-)