Log file Viewer

[17:06:48] <simon_weber> I need to distribute c code in my python package that's _not_ an extension, just a related binary dependency. Overriding 'install' to do run make works fine, but the result doesn't get copied out of pip's tmp build directory. Any ideas?

[17:20:47] <dstufft> simon_weber: you probaly want package_data

[17:21:57] <simon_weber> dstufft: yup, looks like it. thanks!

[18:45:53] <ronny> sup

[18:46:17] <ronny> dstufft: ping?

[18:47:06] <ronny> dstufft: are there any libs i can use for constructing/emmiting the new metadata formats?

[18:47:23] <dstufft> not yet, primarily because they aren't fully accepted yet

[18:48:28] <ronny> so im in a bad spot :/

[18:48:53] <ronny> dstufft: whats the minimal metadata i need to add so a clean uninstall is possible

[18:49:28] <ronny> my recovery from the sinus op is at a stage where im up for doing more work on the gumby elf again

[18:51:13] <dstufft> ronny: inside of a dist-info you need a RECORD file IIRC

[18:51:21] <ronny> ok

[18:52:13] <ronny> dstufft: i want to do a evil develop install mode first that propperly supports uninstall and namespaced packages

[18:52:28] <ronny> dstufft: recrod needs to incldie the dist-info items, right?

[18:52:40] <dstufft> I'm not sure offhand

[18:53:28] <ronny> acccording to a whl i opened it does ^^

[18:54:03] <ronny> dstufft: btw, is it valid to have a dist info folder without a version inside of its basename?

[18:54:36] <dstufft> hrm

[18:54:52] <dstufft> I don't know, I'm not sure if anyone thought about that

[18:55:09] <ronny> develop installs dont necessaryly have a version attached

[18:56:05] <ronny> i'll experiment

[18:56:25] <ronny> dstufft: random other topic, will there stillb e attempts to being packaging stuff into the stdlib?

[18:57:43] <dstufft> ronny: Not really, I'm somewhat opposed to it in general. Maybe some of the low level pieces but high level tools I don't think really belong there

[18:57:51] <ronny> (my current impression is that no matter what one will put into the stdlib it will end in a distutils alike situation)

[18:59:48] <ronny> stdlib additions tendto be a meess ^^

[19:00:03] <dstufft> there's a reason it's ensurepip and not pip in the stdlib

[19:00:35] <ronny> yes, that was fabulous

[19:00:58] <ronny> the stdlib should get something to painlessly gather and use wheels

[19:01:19] <ronny> hmm

[19:01:27] <ronny> im running amok with ideas it seems

[19:02:01] <ronny> (im still in hospital, i just recovered enough to starthacking, my brain is a happy excited puppy)

[19:03:11] <ronny> dstufft: what would i need to examine to figure valid dist-info names

[19:04:05] <dstufft> ronny: um, that's going to be in pkg_resources I think

[19:15:19] <ronny> dstufft: its utterly confusing to walk trough the codebase to the distribution finders

[19:15:27] <dstufft> yea it is

[19:15:32] <dstufft> I don't know that code path well

[19:17:13] <ronny> ah, found it

[19:17:19] <ronny> seems like anything goes

[19:17:56] <ronny> oh, and for extra fun dist-info is no supported inside of eggs

[19:17:58] <ronny> fml

[19:18:34] <ronny> dstufft: hmm, not supporting wheels on sys.path is a serious impediment for some details

[19:19:11] <dstufft> gotta go

[19:19:21] <dstufft> daughter'sd figutre skating

[19:21:07] <ronny> oh, enjoy