[12:21:18] <d0c_s4vage> is it ok to use rst formatted docstrings in pip's source? I didn't see any mention of preferred documentation styles in the pip development documentation
[15:27:19] <gsnedders> pip _vendor doesn't have any magic to replace references in vendored packages to other vendored packages, does it? just done manually on update? i.e., there's always a diff from what the script/Makefile produces?
[16:03:14] <ronny> dstufft: any oppinion on creating a module that creates a import context for vendored packages where they do not have to be patched (related to my distutils-sig discussion)
[16:06:02] <dstufft> I guess I'd have to see the mechanism that does it. We had some magic that attempted to do it before and had problems with it, particularly when Debian and co would unvendor us
[16:06:19] <ronny> (it would also make the de-vendoring by distros less lethal
[16:07:37] <ronny> dstufft: (i have the same problem as pip in pylib and pytest, and potentially some mroe things in future)
[16:09:46] <dstufft> ronny: I guess my opinion is basically, if it works without polluting sys.path then I'd say we're definately open to something that reduces the effort of vendoring. I wouldn't say it's _critical_ since the manual patching is really not very difficult to do, but as a nice to have, sure.
[16:11:08] <ronny> dstufft: i wouldnt mess with sys.path, but put there may need for something in sys.meta_path and some kind of context-manager
[16:11:32] <dstufft> yea we munge with sys.meta_path some now as it is
[16:12:03] <ronny> (the idea is, that when vendored packages are imported, they will get to see a different __import__ that maps from within the vendor controlled space onto the vendored packages first
[16:35:55] <gsnedders> ronny: where would you use the context manager? anywhere you import something from a vendored package?
[16:44:51] <ronny> gsnedders: no, i think not at all actually