Log file Viewer
#pypa-dev logs for Monday the 23rd of February, 2015
[00:07:19] <r1chardj0n3s> ionelmc: yep, one of the exercises is "make a package and submit it to pypi"
[00:11:11] <dstufft> Obviously PyPI is just a really good place to when you want to print nested lists
[00:41:16] <r1chardj0n3s> Ivo: if you'd like I can forward you the list of names that I found, and you can wander through to find other gems :)
[00:41:37] <r1chardj0n3s> to be honest, I feel I should do something with this list, but I'm not sure what that should be. maybe I'll just post it to twitter
[00:42:20] <r1chardj0n3s> thinking about setting up a twitter bot that posts lev-distance-1 project registrations
[00:59:21] <r1chardj0n3s> oh... dstufft, store.gen_file_path doesn't appear to be generating absolute paths any longer...
[00:59:48] <dstufft> r1chardj0n3s: No, PyPI doesn't actually directly manipulate the filesystem anymore, it uses an abstraction now
[01:02:28] <dstufft> http://docs.pyfilesystem.org/en/latest/interface.html package_fs should have that interface
[01:14:13] <r1chardj0n3s> dstufft: this code works, but could you eyeball it for sanity from your perspective pls? https://bitbucket.org/pypa/pypi/pull-request/73/use-the-fs-abstraction-layer/diff
[07:23:01] <Ivo> pip has had some "fun" straddling stringio & bytesio between py2 & 3 in the past, IIRC
[07:23:54] <r1chardj0n3s_afk> (I probably shouldn't have said "never, ever" 'cos warehouse isn't deployed yet...)
[07:24:47] <Ivo> pls dont suggest possibility of warehouse going the way of distutils2, you'll give me nightmares
[07:27:02] <Ivo> sure, but it is big project, only part-time funding for one developer to actually work on it, rest volunteer, to do a painless switch-over
[07:28:21] <Ivo> I wonder if, <Python.org> would ever consider something like funding bountysources for issues related to it
[07:28:53] <r1chardj0n3s_afk> I remember *last pycon* saying that if psf needs one thing, it needs a project manager
[07:29:29] <Ivo> Well, after whatever beaurocracy approved x funding, I'd guess to Nick C to either directly manage or delegate someone to do
[07:31:27] <Ivo> I see in the last like.. 5 years, crowd funding has gotten ridiculously effective for quite a few creative things on the internet
[07:32:44] <r1chardj0n3s_afk> I don't doubt that. I think you either under-estimate the amount of effort that would be required to manage that, or over-estimate the volunteer resource pool to draw managers from
[07:33:38] <Ivo> well my specific idea only needs very-part-time management from one person, mostly being trusted to effectivel dole out imaginary source of money
[07:34:28] <Ivo> because I'm totally reasonably imagining PSF has oodles of money it can't figure out how to spend :D
[07:35:06] <r1chardj0n3s_afk> and liase with other volunteers to determine tasks, scope/spec them, manage listings on funding site(s), manage applicants, handle book-keeping, manage on-going task assignments, ...
[07:36:15] <r1chardj0n3s> I've had sprint rooms at pycon with 8 people wanting to help; the big problem was finding something they could usefully work on.
[07:36:28] <r1chardj0n3s> (I'm really, really not making this a dig at dstufft, please don't take it that way)
[07:37:37] <r1chardj0n3s> the people *I* should be working for so I don't have to fight so much to get funding to go to pycon ;)
[07:38:00] <r1chardj0n3s> also, I could work on warehouse and not feel bad about abandoning my other responsibilities
[07:38:20] <dstufft> I'm doing a massive big refactor of warehouse so that it's not "sit and wait for dstufft to figure out how we should do X in WArehouse since Warehouse is basically a custom framework written by dstufft"
[07:39:00] <r1chardj0n3s> please to be done with that by pycon. also, having a good list of sprint tasks by pycon would be *really* good
[07:39:18] <dstufft> there's a flaskify branch, but I'm not a huge fan of flask for anytihng more than small things
[07:41:54] <r1chardj0n3s> dstufft: fwiw the Horizon folk have finally pulled their fingers out of their collective arses and we're finally making progress on a bunch of stuff. Which unfortunately means my predicted Warehouse time has whittled away to nada :(
[07:42:47] <r1chardj0n3s> I didn't get much further with anything to do with pypi name collisions today because Horizon, and I'm not likely to tomorrow either :/
[07:44:53] <r1chardj0n3s> on the one hand, progress in Horizon is good (hey, it's what I'm paid to do) but on the other hand...
[07:45:24] <r1chardj0n3s> ok, I gotta AFK again, most likely for the rest of the evening. catch y'all tomorrow :)
[07:46:03] <dstufft> r1chardj0n3s: as far as progress goes, there's user login and logout, user profile (no templating or anything done for them), the detial pages for projects are mostly done (needs some themeing, but layout wise is ok), still needs simple API and search to match where we were at with warehouse
[07:48:16] <dstufft> Warehouse is focusing primarily on replicating PyPI's functionality and APIs so we can replace the old code base, once WArehouse *is* PyPI then further enhancement is on the table
[17:20:03] <msabramo> I thought of sending a PR that checks for "nester" in the name on upload and then says "Are you sure you don't want to use https://testpypi.python.org/pypi?"
[19:06:18] <msabramo> so I'm playing with a package that overrides a stdlib module (pdb); when installed not as a wheel, the egg ends up at the head of sys.path and it overrides; when installed as a wheel, it cannot override; any way to make a wheel take precedence?
[19:12:43] <qwcode> msabramo, if you don't know, it's helpful to know that the egg override is happening due to last line of easy-intall.pth (in site-packages)