Log file Viewer
#pypa-dev logs for Friday the 7th of August, 2020
[05:05:15] <travis-ci> Change view : https://github.com/pypa/pip/compare/f17c1d6d3bf4...595aa7b294f8
[09:46:27] <travis-ci> Change view : https://github.com/pypa/pip/compare/595aa7b294f8...d6d84e7c91da
[10:17:38] <travis-ci> Change view : https://github.com/pypa/pip/compare/d6d84e7c91da...e8f52198a94f
[15:43:26] <sumanah> cooperlees: about https://twitter.com/ChangesetLLC/status/1291062082574704649 you said (I just saw in backlog) "I don't know if twitter is the place for such a thread :\ " ... could you talk about that a little more?
[15:44:00] <sumanah> you said "Maybe a tweet linked to discourse? I take it you get more engagement doing it this way?" .... so, I'm trying to get the word out to people who are more likely to look at tweets and Twitter threads than to click through to another site
[15:44:31] <cooperlees> I just hate reading insightful things that way. Personal preference here tho.
[15:44:32] <sumanah> this is one of several things I'm trying to publicize what's happening to people who don't usually pay attention to Python-specific developer info sources
[15:45:00] <sumanah> cooperlees: I hear you! I have usually done this sort of thing as a blog post that I also tweet out a few sentences at a time
[15:45:41] <sumanah> Nitter is faster to load in case you want to see tweets without all the tracking JS https://nitter.net/ChangesetLLC/status/1291062082574704649#m
[15:47:01] <sumanah> https://nitter.net/AlexandraErin/status/1274172971192659968 is a thread by Alexandra Erin that talks about how to get things to get wider publicity on Twitter.... threads are a big part of it
[15:47:57] <sumanah> cooperlees: oh did you feel I was abusing your feed? what would have made my thread politer?
[15:48:00] <cooperlees> I get people are lazy tho and I bet engagement drops by at least 1/2 if all the good facts are in the linked blog
[15:48:50] <cooperlees> "pip big change. resolver coming September in 20.X - Techy details here - blog.changeset.llc/pip_september
[15:50:18] <sumanah> cooperlees: and, separately from Twitter, if you have some tips on how to get more programmers (who use Python but don't pay attention to our internal discussions) aware of what's coming, I'd love some advice
[15:51:08] <cooperlees> sumanah: I am no guru there. I love your goals and intent. I do not have great suggestions.
[15:51:32] <cooperlees> sumanah: a nerdy was I've always thought is a python module - python3 -m latest_python_news
[15:54:01] <cooperlees> Or introduce a once a day / week stdout headline, fact and URL in pip output itself when run on a terminal by a human only
[16:05:47] <sumanah> cooperlees: so that is a cool and fun idea! but how would you get people to install it?
[16:06:35] <sumanah> oh interesting. Just by default, have that come with every install of pip, starting with, like, pip 20.2.2?
[16:07:13] <cooperlees> But yeah, adding a subcommand is the polite way. But we could sneak in a once the mount announcement when something breaking / big change is coming
[16:07:34] <cooperlees> then touch a file in ~/.pipannouncement and use mtime of that file to ensure we're not to spammy
[16:10:31] <fungi> i'll readily admit i only ever look at tweets when someone posts a url to one. i don't even have a twitter account so most stuff there may as well not exist for me
[16:10:51] <sumanah> fungi: as a data point: do you subscribe to https://mail.python.org/mailman3/lists/pypi-announce.python.org/ ?
[16:11:54] <sumanah> I don't have a Facebook account, nor am I on Instagram, Telegram, Reddit, Hacker News, or Stack Overflow. So I rely on other folks to spread word and help me understand what's going to be helpful in those places
[16:12:26] <fungi> i did subscribe to the old pypa-dev ml on google groups, but that was a bit more e-mail-friendly than discourse (even if it did break pgp-mime and stuff)
[16:12:42] <sumanah> and of course there are innumerable Discord, Slack, IRC, and other groups where I am not, where some people get news
[16:12:59] <fungi> and yeah, same here for fb and similar social media. i really don't have an account on any of the popular sm platforms
[16:13:44] <sumanah> fungi: as a data point -- where do you get news that affects you as a programmer on stuff other than Python? like, other languages, or on the OS level?
[16:15:21] <fungi> oss-security ml, debian-devel/debian-announce, openbsd misc, also some usenet newsgroups (though less and less any more)
[16:16:32] <fungi> granted, my interests straddle software development, systems administration, infosec and crypto
[16:18:21] <fungi> oh, and of course more than a hundred irc channels across half a dozen irc networks
[16:22:43] <sumanah> also fungi would it be ok for you to forward the pypi-announce email ("pip 20.2 release, plus changes coming in 20.3") to openbsd-misc? I did a little searching there and it looks like pip comes up every once in a while
[16:25:17] <fungi> weechat for irc, mutt for e-mail/mailing lists and slrn for usenet (mostly because i miss the vms days) are how i mostly deal with the volume of discussion. i don't really listen to podcasts or read lwn articles unless someone suggests them specifically (though i have contributed articles to some sites like opensource.com when prodded)
[16:26:04] <fungi> i'll admit i haven't really used pip much in an openbsd context so i'd feel slightly weird bringing it up in a thread on misc myself
[16:35:51] <fungi> my openbsd deployments are usually tightly dependent on verifiable software, and pypi really only provides transport security (https) and checksums right now. maybe once tuf gets further along...
[17:02:25] <sumanah> ok, I've just done a bit more work, pinging contacts who work on TensorFlow, who have a colleague at JFrog (hope the Artifactory folks get the message and let us know ASAP of any problems), and other CZI grantees
[17:28:13] <sumanah> fungi: you may be interested in https://testandcode.com/124 or https://www.pythonpodcast.com/pip-resolver-dependency-management-episode-264/ which are about the dependency resolver work
[17:29:02] <sumanah> if you search for "podcast" on https://wiki.python.org/psf/PackagingWG you'll find a few others about packaging work .... suggesting podcast interviews about our work is one of the things I've done to try to get the word out, on this project & past ones
[17:29:42] <fungi> thanks! and yeah, we've already tested all of openstack's several hundred entry requirements with the new resolver (i would have replied to the discussion on distutils-sig if we hit problems with it). also we didn't see any appreciable performance degredation
[17:30:39] <fungi> i've been following the resolver work closely since lifeless's first attempt on it ages ago
[17:31:39] <fungi> we've basically been masking all our deps with a massive constraints list for years so that we can properly pre-resolve versions
[17:34:37] <fungi> but yeah, i have fond memories of all the debates on whether to use sat or backtracking solvers
[17:35:56] <sumanah> oh wow, I just thought "huh, I wonder if I can still find my Slashdot login from like 1998?" (I cannot, it's linked to an email that died many years ago) and I found https://news.slashdot.org/story/15/05/25/160236/building-hospitable-open-source-communities-video which has vile comments about me. wheeeeeee
[17:51:46] <sumanah> ok, made a new login and posted https://slashdot.org/submission/12143615/breaking-changes-coming-to-pip-203-pythons-install-tool
[17:52:09] <fungi> oh sheesh. well, hopefully you know that the hard work you and changeset have put into the python community infrastructure is much appreciated (at least by me! but by others too)
[17:52:20] <sumanah> fungi: I'm glad that you have been ... like ... defensively driving regarding your constraints and dependencies!
[17:52:49] <sumanah> fungi: much thanks! yeah it was a particular "WHOA" to see that comment thread because my experience in the last few years has been a lot less toxic
[17:55:11] <fungi> a project the scale of openstack has to handle dependencies with care or else all development would grind to a halt. the complexity increases (roughly) geometrically with dependency count, and any one of them updating can basically block all our pre-merge testing/gating, but we want to run as close to latest of all of them as we can to not in turn block downstream adoption of new dependency versions
[18:17:16] <sumanah> cooperlees: I'm rolling your suggestion around in my head.... Nicole also mentioned an idea of having a module people could install, and interact with on the command line, *to send feedback*
[18:18:29] <sumanah> I think we may not be able to make time to make this happen with the current team before October, but it's something to consider for later, or for volunteers to take up
[18:19:36] <cooperlees> Now if we had a new extensible non xmlrpc api it would be easy to extend for this use case :D
[23:48:30] <travis-ci> Change view : https://github.com/pypa/pip/compare/d6d84e7c91da255084bc6c8cdc909e3e70b8fea7...e8f52198a94f9f8203afcfe34111f3c21dd1bf35