PMXBOT Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Wednesday the 8th of May, 2019

(Back to #pypa-dev overview) (Back to channel listing) (Animate logs)
[13:58:10] <phildini> Hey asmacdo are you around?
[13:58:24] <asmacdo> phildini: yeah hey
[13:59:10] <phildini> Cool! There are some of us at PyCon this year interested in getting https://github.com/pypa/warehouse/pull/4078 to completion. Are you around? Are you still poking at it?
[13:59:17] <phildini> It looks like tests are the major thing needed.
[13:59:24] <phildini> Also cc cooperlees on ^^
[14:00:00] <cooperlees> yeah, asmacdo worked on it last year with me :) He couldn't make it this year.
[14:00:14] <phildini> cooperlees: Steve and I are chatting in the hall.
[14:00:22] <asmacdo> phildini: yeah tests are the main thing.
[14:00:44] <cooperlees> We didn't want to spend the time with the tests unless it was going to get accepted in general.
[14:00:53] <cooperlees> Like the design. That's the main thing we need
[14:01:00] <cooperlees> Then a way to introduce security for write APIs
[14:01:22] <asmacdo> there are a pile of other things that aren't great (pagination, serialization can be improved, subdomain doesn't work as expected)
[14:02:00] <phildini> cooperlees: that makes a ton of sense, does this mean posting to the mailing list with a "Hey I'd like to push on this, is this ok?" Or does it need a major rewrite to get accepted? Who would I ask about this, because Steve and I (and you) all want something like this for our projects.
[14:02:18] <phildini> And to be clear, I'm willing to do the work.
[14:02:34] <cooperlees> Basically, if Donald, Dustin and Ernest (all here) accept we could probably just get to work on it
[14:02:52] <cooperlees> They are the main 3 warehouse maintainers / owners
[14:02:56] <cooperlees> in 26C
[14:03:01] <phildini> Cool. Are all those folks in the room right now? We could come swing by and get a pre-flight checklist from them.
[14:03:13] <cooperlees> Yes, everyone but Ernest.
[14:03:43] <asmacdo> phildini: there is 1 thing that really needs to be figured out before it gets merged, but the rest of it feels fine for an initial implementation
[14:03:54] <phildini> asmacdo: what is that?
[14:04:07] <asmacdo> i put the new api on the subdomain api.pypi.org
[14:04:17] <asmacdo> but it is visible on pypi.org as well
[14:04:36] <asmacdo> not sure if that's a deployment issue, or if something is wrong in the routing/config
[14:05:12] <phildini> That feels like something that could also be checked for in unit tests, based on whatever we decide the final decision is?
[14:06:38] <phildini> Anyway. Steve and I will swing by the room Soon™
[14:07:05] <asmacdo> Sure. That would be fine.
[14:10:03] <asmacdo> phildini: projects/{name}/releases/{version}/files endpoint is broken due to a change that earnest made, probably an easy fix
[14:10:21] <asmacdo> EWDurbin ^
[14:12:17] <asmacdo> thanks for working on this phildini :)
[14:39:16] <phildini> Summary of in-person chat between myself, dstufft, di_codes, cooperlees: People are going to start looking at https://github.com/pypa/warehouse/pull/4078 (JSON API for PyPI). My immediate goal is going to be getting it running, resolving conflicts, and starting to write some tests. Anyone who wants to help with this effort, much obliged. The major concerns from the group were around the shape of the responses on the packages
[14:39:16] <phildini> list endpoint for cache-ability, but I think we can manage those concerns as we move along.
[14:39:50] <cooperlees> +1 - agree
[14:40:50] <asmacdo> awesome. phildini i've poked at it recently, so if you have any trouble, let me know
[14:52:46] <phildini> asmacdo: I'm going to start by getting things running with your PR applied, and see if I run into issues. Do you have any uncommitted test changes or anything?
[14:54:36] <asmacdo> phildini: yeah, but its in a messy state. i recommend using the PR as it is. youll have some merge conflicts in the requirements, but othewise should apply cleanly
[14:54:45] <phildini> 👍
[15:02:38] <EWDurbin> asmacdo: broken in what way?
[15:05:25] <EWDurbin> asmacdo: not seeing any issues.
[15:05:32] <EWDurbin> asmacdo: can you send steps to reproduce?
[15:12:21] <EWDurbin> ah, i see not URLs broken
[15:12:31] <EWDurbin> stale PRs gonna stale :)
[15:13:41] <cooperlees> :)
[15:15:49] <asmacdo> EWDurbin: yeah, its not a real problem, it just needs to be updated to match whats current
[15:34:16] <phildini> Rebased branch that I'm working off of, to add tests, etc: https://github.com/phildini/warehouse/tree/hypermedia-take-2
[15:59:29] <cooperlees> phildini: Nice :) I don't know a good way to split the test writing tho - Ideas?
[16:00:08] <phildini> cooperlees: I'm getting my env set up slowly. I'm pretty happy to give you edit perms on my fork if you want to make commits against that branch?
[16:01:13] <cooperlees> Yeah - Please do - @cooperlees - Then lets talk for 5 mins on parts you want to cover and I'll cover some other files etc. right?
[16:14:04] <phildini> cooperlees: 👍
[16:17:12] <arti> When enabling 2fa on test.pypi.org i dont see any backup codes generation option. Is this currently not implemented?
[16:21:03] <arti> nvm, found same issue in github https://github.com/pypa/warehouse/issues/5800
[18:29:12] <cooperlees> arti: Yes, not implemented: https://github.com/pypa/warehouse/issues/5800
[18:50:10] <ronny> aloha
[18:51:01] <ronny> im wondering - when will pip support pep-517 editable installers in the backend - i'd like to put gumby eld back to life
[18:51:11] <ronny> *gumby elf
[18:51:28] <ronny> (gumby elf is a package bundler targetting wheel/sdist)
[19:23:47] <cooperlees> ronny: You at the US PyCOn sprints in Cleveland? There is talk about this right now in room 26C if you are.
[19:24:37] <ronny> cooperlees: unforztunately not, im in europe
[19:25:00] <ronny> cooperlees: if your are closeby them, shoot them a greeting from me however
[19:25:02] <cooperlees> But there is talk right now between setuptools maintainers and pip maintainers.
[19:25:21] <cooperlees> I'm working on other stuff but been listening to pieces
[19:26:04] <ronny> i see
[19:26:20] <ronny> im mainly on setuptools_scm annd im looking to replace it if i can handle ^^
[19:36:39] <cooperlees> yeah, I can't help much here sorry.
[19:36:49] <cooperlees> pradyunsg is in the discussion. Lets see if he has notifications on
[19:38:04] <cooperlees> asmacdo: Did you push any of the unit tests you wrote anywhere? Just trying to work on it now
[19:38:33] <pradyunsg> ronny: heyo!
[19:38:45] <cooperlees> Go toad_polo !
[19:38:56] <ronny> pradyunsg: yohe :)
[19:40:09] <ronny> pradyunsg: currently gumby elf has this really nasty hack where it generates a wheel with a shim python file and a local version to install on "setup.py develop" via pip, i'd like to get that sane instead
[19:42:24] <asmacdo> cooperlees: i don't see any, but they were not good tests anyway
[19:45:08] <cooperlees> asmacdo: Rightio
[20:00:48] <pradyunsg> ronny: any links for me to look at?
[20:02:16] <ronny> pradyunsg: https://github.com/RonnyPfannschmidt/gumby_elf has the code from 4 years ago, i didnt yet start on it again (back then there was no PEP-517/518 and extra hacks where needed)
[20:04:33] <pradyunsg> I'll post the current draft we have for the editable interface in a bit.
[20:05:02] <ronny> the nasty hack is https://github.com/RonnyPfannschmidt/gumby_elf/blob/master/src/bootstrap_template.py.txt - being incooperated via https://github.com/RonnyPfannschmidt/gumby_elf/blob/03391bdf0cc66faaa2a5e034769e557c31974fd9/src/packing.py#L26-L43
[20:06:51] <pradyunsg> https://www.irccloud.com/pastebin/wmDkkSH0/
[20:09:13] <ronny> pradyunsg: that one looks interesting, the "virtual wheel" is a strange concept with no more context
[20:09:30] <pradyunsg> brb.
[20:09:44] <pradyunsg> in about 2 hours actually.
[20:10:05] <ronny> thats midnight for me ^^
[20:10:50] <ronny> pradyunsg: but let me say that - this virtual wheel with an actual exact fie map sounds a lot better than my local funk wheel with a extra p´pair of execs and maybe a file too much
[23:17:08] <njs> dstufft: "it occurs to me that I could make the argument that basically every committer to a PyPA project should be a gardener" <-- fwiw this was my first thought when I saw the post on discourse
[23:23:13] <ehashman> I do have a lot of nice plants