14:31:58 <fao89> #startmeeting Pulp Triage 2020-09-25
14:31:58 <fao89> !start
14:31:58 <fao89> #info fao89 has joined triage
14:31:58 <pulpbot> Meeting started Fri Sep 25 14:31:58 2020 UTC.  The chair is fao89. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:31:58 <pulpbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:31:58 <pulpbot> The meeting name has been set to 'pulp_triage_2020-09-25'
14:31:58 <pulpbot> fao89: fao89 has joined triage
14:32:20 <ppicka> #info ppicka has joined triage
14:32:20 <ppicka> !here
14:32:20 <pulpbot> ppicka: ppicka has joined triage
14:32:22 <fao89> topic: Enabling docstring linting with pydocstyle?
14:32:27 <ggainey> #info ggainey has joined triage
14:32:27 <ggainey> !here
14:32:27 <pulpbot> ggainey: ggainey has joined triage
14:32:28 <GregorSamsa> #info GregorSamsa has joined triage
14:32:28 <GregorSamsa> !here
14:32:28 <pulpbot> GregorSamsa: GregorSamsa has joined triage
14:32:38 <ppicka> yes please +1
14:32:45 <GregorSamsa> I think we already decided to do this but the problem was just finding the time
14:32:51 <x9c4> #info x9c4 has joined triage
14:32:51 <x9c4> !start
14:32:51 <pulpbot> x9c4: Error: Can't start another meeting, one is in progress.
14:32:52 <pulpbot> x9c4: x9c4 has joined triage
14:33:06 <x9c4> hah
14:33:34 <dkliban> #info dkliban has joined triage
14:33:34 <dkliban> !here
14:33:34 <pulpbot> dkliban: dkliban has joined triage
14:33:41 <bmbouter> #info bmbouter has joined triage
14:33:41 <bmbouter> !here
14:33:41 <pulpbot> bmbouter: bmbouter has joined triage
14:33:41 <ipanova> #info ipanova has joined triage
14:33:41 <ipanova> !here
14:33:42 <pulpbot> ipanova: ipanova has joined triage
14:33:44 <bmbouter> !friday
14:33:44 <pulpbot> ♪ It's Friday, Friday, gotta get down on Friday ♪
14:33:48 <fao89> topic: Enabling docstring linting with pydocstyle?
14:33:54 <x9c4> +1
14:34:03 <ggainey> we have an issue for "write script that runs exactly the checks that travis does" - if that existed, I'd have no probs at all w/this
14:34:06 <bmbouter> oh there is open floor on this
14:34:08 <bmbouter> there is
14:34:15 <bmbouter> this only shows up here b/c at the last meeting we converted the type
14:34:29 <ggainey> but I keep tripping over things because travis does something a little different than what happens if I run black/flake in my dev-env
14:34:40 <bmbouter> yup we gotta talk about dat too
14:34:47 <bmbouter> that's a problem to resolve as well
14:34:54 <ggainey> yes please :)
14:35:00 <bmbouter> I propose skip so we can open floor this
14:35:09 <x9c4> I keep tripping because what flake lints depends on what plugins are installed. And that differs from plugin to plugin.
14:35:11 <ggainey> ah- I thought we were already in open-floor, my bad
14:35:15 <GregorSamsa> yup
14:35:30 <fao89> we are
14:35:42 <bmbouter> ah we switched it
14:35:45 <GregorSamsa> if we standardize the flake8 settings across projects, it would help but probably not solve the problem
14:35:48 <x9c4> _This_ is openfloor, because we swithched the order
14:35:59 * bmbouter gets with the problem :)
14:36:05 <ggainey> ahhhhh rightright, I remember that discussion, coolio
14:36:08 <bmbouter> s/problem/program/
14:36:16 <bmbouter> ok so here we are at open floor
14:36:18 <GregorSamsa> in https://github.com/pulp/pups/blob/master/pup-0008.md we agreed to adopt pydocstyle
14:36:25 <GregorSamsa> we just haven't done it yet
14:36:30 <ipanova> ggainey: we have not announced that because this is a "trial" one
14:36:33 <GregorSamsa> but I am fine to revisit on pulp-dev
14:36:40 <ggainey> anyway - I am +1 for doclinting, but would like it to be available in some more-automatic way than "go read the plugin docs" if we could
14:36:45 <bmbouter> agreed, and I'd like us to make good on that
14:37:00 <bmbouter> the key to it is getting Travis to do it really
14:37:08 <bmbouter> the local linting is pretty much a separate issue
14:37:14 <bmbouter> an important, but separate issue
14:37:17 <GregorSamsa> yea, I had a proposal for local linting that I wanted to discuss at pulpcon but didn't have time to prepare it
14:37:46 <ggainey> bmbouter: it's really frustrating to regularly submit a PR and then find out an hour later that it failed 30 secs in because you missed a lint-issue that you didn't know about
14:37:59 <bmbouter> ggainey: I agree, so let's resolve that with high prio
14:38:04 <ggainey> +1
14:38:09 <bmbouter> but should it hold up the forward progress on this linting improvement?
14:38:12 * ggainey cheers
14:38:32 <ggainey> bmbouter: I lean towards 'yes'
14:38:34 <bmbouter> maybe it should, I'm interested in perspectives on it
14:39:01 <ggainey> I have a personal bias, but I suspect it's really frustrating for occasional committers to get snagged on this
14:39:07 <fao89> we have an issue for that, I'm trying to find it
14:39:24 <GregorSamsa> https://pulp.plan.io/issues/6938
14:39:24 <ggainey> fao89: yeah, I tried and failed as well, godspeed sir! :)
14:39:38 <fao89> GregorSamsa++
14:39:38 <pulpbot> fao89: GregorSamsa's karma is now 1
14:39:40 <ggainey> there we go
14:40:06 <bmbouter> in terms of a specific plan on when and how I think we can probably handle that via pulp-dev maybe?
14:40:12 <ggainey> +1
14:40:14 <fao89> I would like to work on this
14:40:15 <ggainey> sounds good2me
14:40:22 <ggainey> fao89++
14:40:22 <pulpbot> ggainey: fao89's karma is now 110
14:40:23 <x9c4> fao89++
14:40:24 <pulpbot> x9c4: fao89's karma is now 111
14:40:27 <bmbouter> but I'm reading a clear desire to handle both local pre-commit checks and enabling pydocstyle on travis as a "together" goal
14:40:42 <GregorSamsa> that's fine with me. I am happy to help.
14:40:51 <ggainey> bmbouter: that's my preference, but I'm open to other options
14:40:53 <fao89> we have a request from galaxy to enable to run the CI locally, I think 6938 is a good first step
14:41:01 <ggainey> ah cool
14:41:15 <bmbouter> I agree
14:41:33 <ppicka> +1
14:41:35 <bmbouter> so when/where do we talk about the "my local env is different from travis so pre-commit stuff doesn't work well" problem ?
14:42:16 <bmbouter> perhaps that's here?
14:42:20 <ggainey> bmbouter: isn't that the pulp-dev@ discussion?
14:42:31 <ggainey> I may be missing something tho
14:42:44 <bmbouter> it can be there, my main thing to fao89 is that it's not planned to a level to be worked on without that convo
14:43:03 <bmbouter> if you read this ticket it reads like "let's do this!"
14:43:23 <GregorSamsa> we definitely need a discussion on pulp-dev first
14:43:35 <ggainey> +1
14:43:37 <bmbouter> +1
14:43:58 <ppicka> +1
14:44:00 <bmbouter> we need a crisp problem statement to seed that convo
14:44:03 <GregorSamsa> one of the problems we need to solve is that flake8 runs whatever plugins are installed
14:44:18 <bmbouter> yup
14:44:19 <ipanova> +1 agree
14:44:30 <bmbouter> also the "local version" and the version on travis may differ
14:44:56 <GregorSamsa> yup
14:45:01 <bmbouter> also not all plugins may be pydoctest compatible, e.g. so there is a "must enable everywhere for my local tests to pass anywhere" problem
14:45:22 <bmbouter> I think if we got all ^ into 1 email to start that convo we would do well
14:45:37 <GregorSamsa> yea, I need to do some research too to see how we could do it
14:45:47 <GregorSamsa> fao89 and I can work on it
14:45:58 <GregorSamsa> research into what the options are
14:46:07 <bmbouter> who can send the pulp-dev email w/ problem statements?
14:46:18 <GregorSamsa> I can do that
14:46:23 <fao89> GregorSamsa, are you daviddavis?
14:46:37 <GregorSamsa> ha yea, I can't log into daviddavis
14:46:38 <ggainey> heh
14:46:53 <GregorSamsa> I'm on a different computer
14:47:03 <ipanova> heh
14:47:28 <fao89> I was thinking how GregorSamsa knows all these things, and I never saw him on IRC before
14:47:32 <GregorSamsa> lol
14:48:11 <bmbouter> this is great
14:48:27 <dkliban> lol
14:48:28 <GregorSamsa> just put me and fao89 down as having the AI here
14:48:34 <ipanova> i was hoping this is community user lol
14:48:42 <GregorSamsa> :D
14:50:13 <fao89> GregorSamsa--
14:50:13 <pulpbot> fao89: GregorSamsa's karma is now 0
14:50:21 <fao89> daviddavis++
14:50:21 <pulpbot> fao89: daviddavis's karma is now 367
14:50:25 <GregorSamsa> hehe
14:50:28 <fao89> making it right
14:50:36 <ggainey> heheh
14:50:40 <ggainey> next topic?
14:50:51 <fao89> Propose: Dropping the (unwritten) rule to format docs sources in block layout with 100 chars width.
14:51:37 <x9c4> +1
14:51:54 <GregorSamsa> heh you can't +1 your own proposal
14:52:19 <ggainey> interesting. I do like the idea of "one sentence per line" - I've always tried to do that working in anything that auto-formats the text
14:52:22 <x9c4> In the spirit of black formatting, we should add line breaks at language features not 100 chars.
14:52:40 <fao89> x9c4, you have to create another user for doing that, you have to learn with daviddavis =P
14:52:48 <ggainey> heh
14:52:50 <GregorSamsa> lol
14:52:56 <ggainey> cheaters, all of you :)
14:53:09 <x9c4_the_other_o> like this?
14:53:13 <GregorSamsa> perfect
14:53:18 <ggainey> might have been just a *little* obvious
14:54:09 <x9c4> Anyway, when you edit one word, you need to reformat the whole paragraph a t the moment.
14:54:24 <x9c4> This is tedious and makes for unneccessary big diffs.
14:54:31 <ggainey> hrm, yeah
14:54:34 <ipanova> yeah
14:54:45 <bmbouter> it's true but humans do read in paragraph form and the unrendered docs are read often
14:54:48 <x9c4> When you add one phrase the diff-stat should be +1
14:55:07 <x9c4> GregorSamsa, see i plugged it another way. ;)
14:55:23 <bmbouter> for example, python as a language's whole focus is readability, I suspect docs would do well to have a similar goal
14:55:35 <bmbouter> this approach makees the main focus the diff
14:56:01 <bmbouter> x9c4: as you can see I've thought more about this from our convo yesterday
14:56:20 <ggainey> bmbouter: reading our tags with all their embedded tags and links is not all the useful, in my experience
14:56:59 <ggainey> ^the useful^that useful
14:57:11 <GregorSamsa> I often read the unrendered docs
14:57:21 <bmbouter> agreed, but if we separate every sentence on its own line are we going closer or further from readability
14:58:03 <ggainey> for me, we're already past the liune where it's a difference - I 'make html' in my dev-env and read the results in a browser much of the time already
14:58:10 <x9c4> We should still strive to have lines under 100 chars. You can break at comma if needed.
14:58:22 <bmbouter> I am big +1 on the 100 chars limit
14:59:10 <ipanova> +1 to 100 char limit
14:59:11 <bmbouter> also fwiw our docs readability is inhibited by the use of RST, markdown I believe is widely seen as more readable
14:59:26 <GregorSamsa> +1
14:59:33 <ipanova> and i think we are following the 100char limit, are not we?
14:59:42 <bmbouter> we are not because travis does not enforce
14:59:56 <bmbouter> I used to ask for it, but I think I was being annoying so I kind of stopped
15:00:31 <ipanova> i see
15:00:40 <ipanova> well i do it localy
15:01:37 <bmbouter> I also do cuz pycharm tells me
15:01:39 <x9c4> I'm not against the 100 char limit. A partial sentence of that length might not be fit for docs anyway.
15:01:40 <ggainey> I don't think I've ever seen it mentioned anywhere
15:02:03 <bmbouter> I think it mostly comes from contributors who have line_wrap editors
15:02:08 <ggainey> can a travis-rule like this tell the diff between discussion and code-examples?
15:02:30 <bmbouter> I think deciding we want an improvement to limit docs to 100 chars would be a good goal
15:02:41 <bmbouter> how and to what extent we can achieve that we can solution it up lata
15:03:55 <bmbouter> is there a 100 chars doc checking ticket? and if so can we start putting comments on it? if not, can a stub be made as a next step?
15:05:34 <travis_ci> I cannot do that
15:05:51 <dkliban> lol
15:05:51 <ipanova> travis said no :/
15:05:57 <ggainey> hehehe
15:06:18 <GregorSamsa> !friday
15:06:18 <pulpbot> ♪ It's Friday, Friday, gotta get down on Friday ♪
15:06:35 <GregorSamsa> I don't think there is a ticket and I think creating a ticket is a good next step
15:07:04 <GregorSamsa> also, I gotta go. see you all monday.
15:07:11 <ggainey> seeya sir!
15:08:11 <x9c4> I can create a ticket.
15:08:15 <ggainey> kk
15:08:21 <x9c4> What does it contain?
15:08:50 <travis_ci> The build failed.
15:08:51 <ipanova> that we want to limit docs to 100 chars and enable this in travis
15:08:51 <x9c4> line breaks at ",.:" and no more than 100 chars?
15:09:19 <ggainey> works4me
15:09:27 <bmbouter> lolol thanks travis_ci
15:09:32 <bmbouter> x9c4: +1
15:09:42 <bmbouter> travis_ci: can you maybe speed up your pipelines some?
15:09:43 <fao89> next topic?
15:09:51 <ggainey> yus
15:10:24 <travis_ci> Error: 'can you maybe speed up your pipelines some?' is not a valid command
15:10:29 <ggainey> :P
15:10:45 <fao89> topic: Pulp2 and makedirs
15:10:49 <ggainey> that's me
15:10:54 <ggainey> lemme explain what I'm seeing
15:11:04 <ggainey> this is in the pulp2-to-3-migration context
15:11:40 <ggainey> you have p2 installed. p3 install creates a pulp-group, adds apache to it, sets the gid bit on all the pulp-home dirs, and sets dir-perms to 775
15:11:54 <ggainey> it needs to do this so links can be created inside various directories
15:12:09 <ggainey> then, pulp2 syncs new content - and ends up with 755 dirs
15:12:48 <ggainey> the reason tjhis is happening, is because we use os.makedirs() in a number of places - and os.makedirs() pushes perms, even when explcitly specified as 775, thru umask
15:14:07 <ggainey> so, my question: do we try to address this by trying to insure pulpp2 is running with a umask of 002 (instead of 022)? or do I fix all the pulp2 makedirs() uses, to use instead our itil.mkdir call, and make sure it dplays the right game w/changing/resetting umask?
15:14:10 <ggainey> thoughts?
15:14:14 <fao89> I think ipanova filed a similar issue
15:14:21 <ggainey> fao89: it's the one I'm working on
15:14:32 <ggainey> https://pulp.plan.io/issues/7445
15:14:34 <ipanova> fao89: that's the issue ggainey is trying to fix actually
15:14:38 <ggainey> lemme add to agenda, I'm an idiot
15:14:52 <ggainey> sorry about that
15:15:06 <ggainey> anyway - I'm fishing for ideas here, what do folk think?
15:16:10 <ggainey> (also, I am unsure which process' umask would need to change, to affect the pulp2 code - httpd?)
15:16:24 <ipanova> i think fixing the mkdir calls might be the best
15:16:36 <fao89> this is complicated
15:16:52 <ggainey> ipanova: that's the direction I'm leaning. We have a number of them, but it's a pretty straightforward change
15:17:18 <fao89> how this can affect pulp2 users?
15:17:22 <ggainey> fao89: Fun With Umasks - you *always* end up with something other than what you expected :)
15:17:30 * bmbouter puts on his umask
15:17:37 <ipanova> we can also ask for input on the list, i think we have enough sysadmins who could weigh in
15:17:45 <ggainey> fao89: zacly. current-behavior is, content/published dirs end up 755
15:17:52 <ggainey> ipanova: oo, that's a fine idea
15:18:04 <bmbouter> I think we need more input for sure
15:18:08 <ggainey> bmbouter: it keeps out the PermissionCovidBits :)
15:18:15 <bmbouter> lol
15:18:22 <bmbouter> but this is an important issue, let's figure it out!
15:18:30 * bmbouter puts on his tin foil no-covid hat
15:18:34 <ggainey> I'm good with that - how about I add a bit of explanation, and then point pulp-dev@ to the issue?
15:19:00 <fao89> I'm wondering why we need 775 at pulp3
15:19:22 <ggainey> fao89: ttereshc ipanova can either of you talk to migration's need to create links?
15:19:40 <ggainey> (because I 8think* that's what's driving this, it breaks migrating)
15:20:08 <bmbouter> fao89: some other services may need writer perms and they can't run as the pulp user is my understanding
15:20:11 <bmbouter> s/writer/write/
15:20:13 <ggainey> specifically, if you pulp2-sync a new repo after installing pulp3 and before you're 'done' with your 2-to-3 migration
15:20:14 <ipanova> it does break migration, i need to remember the details
15:20:24 <ggainey> yeah
15:20:36 <ipanova> from the issue:
15:20:37 <ipanova> drwxrwsr-x. 173 apache pulp 4096 Sep  2 08:34 iso    <------------------    existing content by the time pulp3 installed
15:20:42 <ggainey> the pulp2 dirs, after pulp3 install, are owner/apache, group/pulp
15:20:43 <ipanova> drwxr-sr-x.  12 apache pulp  106 Sep  2 08:35 modulemd  <----------------------   new content after pulp3 installed
15:20:53 <ggainey> yup
15:20:56 <ipanova> pulp2 removed w perms forthe grups
15:20:59 <ipanova> groups
15:21:22 <ggainey> ipanova: it doesn't remove perms from existing dirs - but newly-created ones will have 755 instead of 775
15:21:27 <fao89> I think the problem is not on 755 pulp2, I believe it is the 775 on pulp3
15:21:29 <ipanova> and this makes it impossible to create a hard link
15:21:36 <ipanova> ( i remembered)
15:21:40 <fao89> but it would require a lot of changes at the installer
15:22:16 <ggainey> fao89: the installer?
15:22:55 <fao89> wait, 755 is on pulp2 or 3? I think I misread
15:22:59 <ipanova> fao89: it wil lrequire fix in pulp2... immage you run installer on pulp2. Installer fixes everything. then new pulp2 content comes in, it has screwed up perms and pulp3 is doomed
15:23:08 <ggainey> exactly
15:23:16 <bmbouter> I agree I think it will require pulp2 fix
15:23:23 <ggainey> pulp2 creates new direectories as 755, no matter what perms you pass to makedirs
15:23:27 <ggainey> it def does
15:23:32 <bmbouter> and that is ok to me
15:23:36 <ipanova> ggainey: i think it will be good to draft an email with how to fix this perm issue
15:23:47 <bmbouter> I agree, this is sufficiently complicated that we need that next step
15:23:49 <ggainey> ipanova: concur, I have that as my action item
15:23:57 <ipanova> sweet
15:24:26 <ipanova> are we ready to move on?
15:24:35 <ipanova> we have 6 mins left ^_^
15:25:14 <ggainey> +1
15:25:21 <ipanova> i see no other topics
15:25:25 <ipanova> triage?
15:25:28 <ggainey> yus
15:25:36 <fao89> oh yeah
15:25:44 <fao89> I forgot the triage
15:25:49 <ipanova> heh
15:25:49 <fao89> I was about to end
15:26:01 <fao89> !next
15:26:10 <ipanova> #info ipanova has joined triage
15:26:10 <ipanova> !start
15:26:10 <pulpbot> ipanova: Error: Can't start another meeting, one is in progress.
15:26:11 <pulpbot> ipanova: ipanova has joined triage
15:26:15 <ipanova> hm
15:26:35 <fao89> !end
15:26:43 <fao89> this is crazy
15:26:52 <ipanova> how do we start the triage?
15:27:05 <dkliban> !next
15:27:52 <ggainey> hrm
15:28:04 <x9c4> we didn't tell pulpbot that we switched? Maybe he's sulky.
15:28:27 <dkliban> fao89: try to end it again
15:28:35 <fao89> I tried
15:28:42 <fao89> it does not allow me to end
15:28:50 <dkliban> who started the meeting?
15:28:53 <ipanova> #endmeeting
15:28:53 <ipanova> !end
15:29:01 <fao89> I started
15:29:17 <fao89> then x9c4 tried to start, then ipanova
15:29:22 <dkliban> oh
15:29:23 <x9c4> !end
15:29:26 <fao89> and pulpbot is confused now
15:29:30 <bmbouter> lol
15:29:31 <bmbouter> !friday
15:29:31 <pulpbot> ♪ It's Friday, Friday, gotta get down on Friday ♪
15:29:36 <ggainey> !dance
15:29:36 <pulpbot> ♪┏(°.°)┛┗(°.°)┓┗(°.°)┛┏(°.°)┓ ♪
15:29:37 <bmbouter> well it knows what's important at least
15:29:39 <x9c4> Sorry, i didn't mean to.
15:29:44 <ipanova> can we do a manual triage?
15:29:46 <fao89> !chair
15:29:46 <pulpbot> fao89: Error: "chair" is not a valid command.
15:29:57 <fao89> is there a way to list the chair?
15:30:12 <x9c4> !help
15:30:12 <pulpbot> x9c4: (help [<plugin>] [<command>]) -- This command gives a useful description of what <command> does. <plugin> is only necessary if the command is in more than one plugin. You may also want to use the 'list' command to list all available plugins and commands.
15:30:34 <fao89> !list
15:30:34 <pulpbot> fao89: Admin, Aka, AutoMode, Channel, Config, Ctcp, Filter, Format, Karma, Later, MeetBot, Misc, Network, NickAuth, Owner, Plugin, PluginDownloader, PulpTriage, RSS, Redmine, Services, Topic, User, and Utilities
15:30:41 <dkliban> i am going to restart pulpbot
15:30:56 <ggainey> sigh - bots are supposed to DM you responses to help
15:30:56 <fao89> !issue 7568
15:31:01 <ipanova> https://pulp.plan.io/issues/7568 this is in post, let's add to the sprint
15:31:08 <ggainey> +1
15:31:14 <fao89> https://pulp.plan.io/issues?query_id=143
15:31:43 <dkliban> fao89: should i restart pulpbot?
15:32:09 <fao89> we have only 3 issues, I think we can do it manually
15:32:31 <fao89> issue: https://pulp.plan.io/issues/7568 propose accept and add to sprint
15:32:31 <dkliban> ok
15:32:36 <ipanova> +1
15:32:55 <dkliban> +1
15:32:59 <ggainey> +1
15:33:10 <fao89> issue: https://pulp.plan.io/issues/7549
15:33:23 <fao89> propose accept and add to sprint
15:33:30 <ipanova> accept?
15:33:38 <ggainey> katello is asking for this - and in fact we've said since pulpimport/export beginning that it needed a dry-run option
15:33:38 <ipanova> will this be worked on this sprint?
15:33:41 <ggainey> I'd accept, not add to sprint
15:33:47 <ggainey> unlikely
15:33:54 <ipanova> ack let's accept
15:33:55 <ggainey> prob next, so we can hit 3.8
15:34:03 <fao89> accept
15:34:06 <ggainey> +1
15:34:22 <fao89> issue: https://pulp.plan.io/issues/7548
15:35:05 <fao89> propose accept
15:35:21 <ipanova> seems like it needs more discussion? wdyt
15:35:47 <ipanova> good candidate for the open floor dalley?
15:36:09 <fao89> we can skip for next triage/open floor
15:36:13 <dkliban> let's skip
15:36:16 <ggainey> ok
15:36:20 <ipanova> ok
15:36:24 <fao89> !end
15:36:37 <fao89> !friday
15:36:37 <pulpbot> ♪ It's Friday, Friday, gotta get down on Friday ♪
15:36:44 <ipanova> !pulp
15:36:44 <pulpbot> 🍊 Yay, Pulp! 🍊 Go team go! 🍊
15:36:54 <ggainey> :)