Log file Viewer
#pypa-dev logs for Monday the 30th of March, 2020
[04:45:28] <devesh> pradyunsg: So does that mean I should keep 'return None" as is, and build asserts and conditionals around them to make sure mypy annotations works regardless?
[07:36:08] <gutsytechster> pradyunsg: I read the logs. There is no issue. I see you've got lots to do. :)
[07:36:36] <gutsytechster> Though I think you haven't reviewed the proposal yet. Let me know when you do.
[13:37:13] <sumanah> My mom is so worried for me because I'm in New York City..... there is nothing I can do to adequately reassure her. Which is hard for both of us.
[13:38:22] <sumanah> also, people with anxiety: https://twitter.com/rokwon/status/1243695810329509888 "who else has the type of anxiety that tells you that, if you stay hypervigilant & informed & if you, in detail, imagine every possible bad outcome, you will thereby ward off harm to yourself & your whole community like some kind of powerful, jumpy magician
[13:47:04] <pradyunsg> I'm going to try adding colors to 'pip uninstall' output today because... colors are nice and there's no reason pip uninstall can't look nicer. And this isn't a mind-bendlingly difficult problem like discussing dependency resolution/designing abstractions for it; which is nice and relaxing in a way. :)
[15:02:41] <McSinyx[m]> pradyunsg: I'm so sorry for the late reply, I've seen the messages in the morning but other work for school blew me away
[15:06:23] <McSinyx[m]> the original plan is only to download in parallel the packages in the same *tier* or resolution, which might already be an enhancement, since packages usually have more than one dependencies
[15:10:06] <pradyunsg> McSinyx[m]: I understand. I do think it's do-able, which is why I'm not saying "push the abort button" for this.
[15:10:55] <pradyunsg> McSinyx[m]: At the same time, the fact that there is a fallback in the proposal, in case these things don't work out helps.
[15:13:20] <pradyunsg> McSinyx[m]: BTW, it'd be nice to have some more detail on what the "backup project" would entail (like "Change option definition to be a more declarative, consistent, static data-structure, replacing the current ``partial(Option, ...)`` form" :P)
[15:14:28] <pradyunsg> McSinyx[m]: And to add references to the relevant issue for it; possibly including a summary of the discussion as well.
[15:15:47] <pradyunsg> McSinyx[m]: In summary, other than the "backup plan" section, the proposal seems pretty great! :)
[15:25:09] <techalchemy> pradyunsg, did you actually mean for the new warning to pop up in distro packaging?
[15:25:46] <techalchemy> <pradyunsg> Just kidding, we figured that adding back those endpoints and making it loudly warn the user would be a good thing to do, since a *lot* of traffic on our issue tracker was to that issue of broken wrappers
[15:26:03] <pradyunsg> there are a lot of things in distro packaging that I'd love to change if I'd inf time.
[15:26:03] <McSinyx[m]> I got that, I was just trying to make sure that there were no misunderstanding between us that I'm heading to something too big
[15:26:24] <McSinyx[m]> btw if there's anything came to mind that might cause you inconvenience, don't hesitate to tell us
[15:26:28] <McSinyx[m]> we are here to help you and other maintainers during the summer, and I hate that if I become some sort of burden
[15:28:28] <pradyunsg> techalchemy: ah. not really; more along let's print a warning to users, instead of refusing to work for them -- because their distro does something we tell them to not do (call pip's internals directly in their wrappers).
[15:28:59] <pradyunsg> plus, it's not like those distros changed what they were doing anyway. *cough* debian *cough*
[15:31:09] <pradyunsg> techalchemy: yea, but only on the one they directly distribute. Anyone getting it via *any other* mechanism gets the message.
[15:31:39] <pradyunsg> because that wrapper makes assumptions that we explicitly told people to stop making when we made the _internal move in pip 10.
[15:31:42] <techalchemy> pradyunsg, pip3 won't invoke any other pip besides the one that debian distributes
[15:34:17] <techalchemy> sumanah: ============================================================= 1 failed, 231 passed, 21 skipped in 442.72 seconds
[15:34:50] <pradyunsg> techalchemy: the traffic numbers on pip's issue tracker suggest that it is *very much* still an issue. :( https://usercontent.irccloud-cdn.com/file/cuWPMjdd/Screenshot%202020-03-30%20at%209.03.48%20PM.png
[15:37:58] <sumanah> there may be some creative things where we can help inform people about their options ..... so, ping me
[15:38:08] <pradyunsg> Some days I feel like ripping this band aid (warning and not breaking) out at some point, and file an issue on debian's issue tracker and then start pointing users at that.
[15:38:32] <sumanah> but right now I know both of you have other things you want to concentrate on? I think?
[15:39:27] <pradyunsg> the colors in uninstall are... progressing. turns out logging + terminal colors + spinner + various logging levels... still has complexity. :)
[15:41:27] <sumanah> techalchemy: have you pushed a new version of https://github.com/pypa/pipenv/commits/feature/vendor-update ? That improves the test failures to only 1 failure?
[15:42:02] <sumanah> maybe you have been --amend ing https://github.com/pypa/pipenv/commit/c44c514f99fc23c0b0667c8187f13f5f4f12d2f0 and pushing?
[17:20:44] <sumanah> devesh: as I understand it, if pmxbot was disconnected during some period, it just misses those messages
[17:21:22] <sumanah> techalchemy: https://github.com/pypa/pipenv/issues/3369#issuecomment-606082540 you're getting cheers on GitHub :-
[18:29:43] <sumanah> techalchemy: apology unnecessary but accepted :-) just wanted to let you know people are rooting for you
[20:03:27] <sumanah> techalchemy: lemme know if you have some time to talk today? it would be good to manage expectations in case it's unlikely you can put out a prerelease tomorrow (end of March)
[20:43:51] <devesh> This might be an odd question, but what are the other repos, perhaps in the pypa org which need folks to fix bugs, create features etc.
[20:44:56] <techalchemy> though i'd like for things to be lower complexity, and i think it's not needed to be as complex as it is
[20:45:16] <devesh> Haha, I have spend just around 2 years with python, hence the "lower complexity" bit techalchemy
[20:45:43] <pradyunsg> techalchemy: we'll release something that has the shape of a resolver, but still working on the finer details. :)
[20:46:28] <devesh> Well everyone sits together and work for specified features which are planned lol
[20:48:52] <devesh> But I did see that most of the projects revolve around sub components of pip in pypa
[20:49:03] <techalchemy> devesh, i think most OSS projects are too busy trying to put out fires to plan features
[20:49:37] <techalchemy> they might give off the illusion they are planning coherently but even within the pypa new releases usually break components of other PyPA software in surprising ways
[20:52:56] <techalchemy> pradyunsg, my last test failure is something that succeeds when i dont run it on CI :|
[20:54:51] <sumanah> Hi techalchemy devesh pradyunsg - so devesh I think your question is a good one
[20:55:02] <sumanah> "what are the other repos, perhaps in the pypa org which need folks to fix bugs, create features etc...and perhaps the complexity is similar to or maybe slightly lower than pip"
[20:56:16] <sumanah> To the larger question regarding planning: in some software projects (both closed and open source) there's a role called an "Architect" and/or an overlapping role called "Product manager"
[20:57:04] <sumanah> Many open source projects do have central planning of new features, prioritizing work, etc.
[20:57:53] <sumanah> https://opentechstrategies.com/files/pub/MZ+OTS_OS_Archetypes_report_ext_scr.pdf is a useful guide to the many different types of open source projects that exist
[20:58:41] <sumanah> So, years ago, dstufft and others formed a very loose coalition called the PyPA
[20:59:33] <sumanah> devesh: the PyPA deliberately has very little hierarchy. This has recently caused problems and many of its members would like more hierarchy
[21:00:04] <sumanah> devesh: for background you can, for instance, read https://discuss.python.org/t/closing-the-loop-on-pypa-governance-bdfrn/1776/ and related/linked discussions
[21:00:50] <sumanah> devesh: I want to be clear that the Python Software Foundation is a nonprofit that pays, like, 5-7 people to keep PyPI, PyCon, and the financial/legal/etc. infrastructure of Python going
[21:01:23] <sumanah> devesh: there are many different open source projects written in Python that benefit from the PSF existing, but in most of them, the PSF has no _direct_ involvement.
[21:01:52] <sumanah> Right. So you understand.... PSF is sort of a scaffolding that many volunteer initiatives can grow on, like vines
[21:03:24] <sumanah> Zulip, Beautiful Soup, and the various projects that (for instance) Will Kahn-Greene works on at Mozilla https://bluesock.org/~willkg/blog/mozilla/socorro_2019.html are all examples of Python projects where there is some clear planning and prioritization and coordination
[21:04:11] <devesh> BTW sumanah: are there a list of github orgs related to python somewhere under which a list of projects like, pypa, psf etc>
[21:04:16] <sumanah> devesh: you also mentioned that you'd like to work on something where "Well everyone sits together and work for specified features which are planned lol" .... when you say "sits together" do you mean literally, in person?
[21:05:38] <devesh> sumanah: I actually meant having discussion over various means, documenting things etc, but in a more informal manner
[21:05:43] <sumanah> someday, after the pandemic passes, you can come to an in-person "sprint" (like a hack week). I participated in one at a PyCon North America many years ago and helped Mailman 3.0 get released https://wiki.list.org/DEV/Home
[21:06:39] <devesh> True "sits together", which ironically is not possible now :) But yes I participated in an Online DevSprint by BangPypers
[21:06:50] <sumanah> devesh: Right. So, within individual PyPA projects, there are efforts like that -- and sometimes across PyPA projects too. For example, I helped the PSF apply for funding so we could pay some people to do 1 big project with pip
[21:08:04] <sumanah> devesh: I think Twine https://github.com/pypa/twine/ might be a good repository for you to look at.... Brian Rutledge has been making steady progress as a maintainer and is open to collaborating on initiatives to improve the project
[21:08:33] <sumanah> devesh: you asked: "are there a list of github orgs related to python somewhere under which a list of projects like, pypa, psf etc" Sort of. Let me explain.
[21:08:51] <devesh> And how do you set time aside from your regular work for open source projects? With all this wfh, work in my company is slow, so I can contribute a lot more
[21:09:23] <sumanah> Anyone can create an open source project that involves/is written in Python. So you can use GitLab, GitHub, Launchpad, and so on, and search for "Python" or look using their tags and so on.
[21:09:42] <sumanah> If you're looking for projects that are officially affiliated with PyPA, use this list https://packaging.python.org/key_projects/ .
[21:10:11] <sumanah> (not all of them are on GitHub! Some are on BitBucket, for instance. So be careful)
[21:10:39] <sumanah> The Python Software Foundation has started acting as a GitHub organization https://github.com/psf/ for a few projects
[21:11:10] <sumanah> There's also a Python organization https://github.com/python/ for repositories related specifically to the Python programming language itself
[21:12:03] <sumanah> the wonderful thing about a hyperlink on the web is that it can go to any other webpage
[21:12:39] <sumanah> and new sites can go up all the time! and you can link to them! but the very flexibility inherent in that means also that sites can go away, move, etc. and then old links get "linkrot"
[21:13:14] <sumanah> similarly: it's great that anyone can start a new open source repository! but also the lack of centralization can make it hard for new folks like you to discover the stuff you want to find.
[21:14:49] <sumanah> The PyPA only "owns" a few repositories related specifically to packaging. The https://github.com/python/ is similarly limited in its topic. And https://github.com/psf/ .... I think (and I am not sure, I am not officially saying this, I may be wrong)
[21:15:17] <sumanah> ... I think it's for a few projects that may need to apply for funding under the PSF umbrella, or where the original "owner" wanted it to be under the PSF umbrella to make it more official and less personal, or as part of a governance .... thing. I don't know!
[21:16:20] <techalchemy> (RE: PSF projects, I know for the ones that were kenneths they were just too important and he didn't want the responsibility of owning them anymore so the PSF stepped in to steward the project(s))
[21:16:24] <sumanah> devesh: If you are looking for a list of Python-related code "orgs" because you want to analyze them systematically in some way, then I can try to give you some guidance. If you are looking for projects to contribute to, I would give you different guidance. And so on
[21:17:06] <sumanah> devesh: and as for time management ..... https://changeset.nyc/ is my firm. I run a consultancy dedicated to open source work
[21:17:35] <devesh> Thanks sumanah for that great overview, I learned quite a lot, I also remembered about https://github.com/pallets
[21:18:02] <devesh> My question about Python-related code "orgs" was just out of curiosity, I was looking more for projects to contribute to
[21:18:13] <techalchemy> yeah for splitting time, my company pays me to spend some time on open source, and currently sumanah has been helping me make sure i do that in a productive way but I don't think she is scalable
[21:18:20] <sumanah> other people in this channel who have more standard employment can speak with more experience on how to make time outside of that for personal FLOSS contribution
[21:19:18] <devesh> sumanah: wow so your firm is specifically for helping people out for floss projects, that's great
[21:19:37] <sumanah> devesh: yup! and if you decide you want to compete with me, go ahead, there is plenty of room in this market in my opinion
[21:19:57] <sumanah> devesh: more projects to contribute to: take a look at https://github.com/virtual-biohackathons/covid-19-bh20
[21:22:08] <sumanah> techalchemy: I am attempting to scale Changeset! I have subcontractors and someday I aim to have other employees even!
[21:22:17] <devesh> thanks for the link, I will surely look into it, I was also interested in projects which I can contribute to more long term as well
[21:23:00] <techalchemy> sumanah, the testcase that fails seems to fail only when it runs under test, it works fine when i just try it by hand
[21:23:26] <techalchemy> it has to do with pep517 backends not installing properly in nested virtualenvs
[21:24:30] <travis-ci> pypa/pip#15344 (master - d79632a : Xavier Fernandez): The build is still failing.
[21:24:30] <travis-ci> Change view : https://github.com/pypa/pip/compare/57cb9416452c...d79632a7a9aa
[21:24:32] <sumanah> also have you filed a bug already about the test failure? (does it make sense to? is there someone else who could help in any way?)
[21:24:39] <techalchemy> it's kind of up for debate i guess, but it feels like a serious rabbit hole
[21:25:16] <techalchemy> well the test that's failing doesn't actuall claim to be testing pep517 backends
[21:25:20] <sumanah> techalchemy: so my take might be: make a prerelease now but with a warning in the release notes saying "known bug: PEP517 backends and nested virtualenvs"
[21:25:42] <techalchemy> well that's the thing... nested virtualenvs are only a thing that happen this way in the test environment
[21:26:56] <techalchemy> since pipenv gets installed in a virtualenv for testing, and then creates more virtualenvs to run the test suite -- the inner virtualenv then spawns pep517 subprocesses to build in isolation etc
[21:27:33] <sumanah> we can get some people to manually test on Windows once the prerelease is out, and, simultaneously, we can get some folks to help fix this particular automated test
[21:28:38] <techalchemy> tbh this test claims to only test fixing a toml issue it just happens to install something that has custom backend
[21:29:21] <techalchemy> this test suite is crap anyways, i dont even know how much of pipenv is actually under test
[21:33:36] <sumanah> ok! techalchemy it's past 5:30pm ET here in NYC .... do you want to try to keep working on this stuff today or quit for the day?
[21:39:14] <techalchemy> same time here, i will work a little bit longer, try to push this up and then likely wrap up for the day
[21:40:08] <sumanah> It feels like you can in fact do this by 5pm tomorrow as long as you have like 3 hours or so
[21:40:36] <techalchemy> i am in meetings at 10 and 2, and everyone is aware this is getting released
[21:41:25] <sumanah> ok, techalchemy then I'll ping you around 10:30 or 11am .... but of course there is no need for you to wait for me
[22:27:33] <travis-ci> Change view : https://github.com/pypa/twine/compare/5482d6877df0cd90545c4708e5abe13b359f8a08...f7402e0d2e1c1e6ecde61dc8bcba515b807a48a0
[22:29:46] <travis-ci> Change view : https://github.com/pypa/pip/compare/57cb9416452cc81686e34f1a61ca955f5c0c2e91...d79632a7a9aaec8d75d312ec43ad991fcc833c50