16:30:41 <djmitche> #startmeeting weekly
16:30:41 <bb-supy> Meeting started Tue May 17 16:30:41 2016 UTC and is due to finish in 60 minutes.  The chair is djmitche. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:30:41 <bb-supy> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:30:41 <bb-supy> The meeting name has been set to 'weekly'
16:30:46 <djmitche> #topic Introduction
16:30:53 <djmitche> #link https://titanpad.com/buildbot-agenda
16:31:19 <djmitche> looking at backlog, lots of traffic on github today, but not much here in #buildbot
16:32:15 <bb-trac> [trac] #3551/enhancement (new) updated by dustin (empty comment) http://trac.buildbot.net/ticket/3551
16:32:54 <bb-trac> [trac] #3552/defect (new) updated by dustin (Can you give a little more detail for how to reproduce this?) http://trac.buildbot.net/ticket/3552
16:33:12 <djmitche> #topic Development Week in Review
16:33:40 <djmitche> A bunch of requests around APIs closed
16:33:49 <tomprince> Latent workers should start in parallel now.
16:34:11 <djmitche> #info latent workers can now start in parallel, rather than being serialized
16:34:14 <djmitche> awesome :)
16:34:30 <skelly> that is a welcome change
16:34:59 <tomprince> That was some sponsored work.
16:35:33 <djmitche> looks like a lot of good fixes to protocol handling (PINGING, NoSuchMethod, etc.)
16:36:04 <djmitche> oh, "Start latent workers as part of the build, rather than before the build" -- that's a big change!
16:36:10 <djmitche> very cool
16:36:19 <djmitche> anything else to highlight this week?
16:36:34 <tomprince> That is the bit that makes them parallel.
16:36:54 <djmitche> yeah I just didn't realize what a major refactor that was
16:37:02 <djmitche> very nice
16:37:15 <djmitche> rutsky: you around for a 2340 update?
16:37:19 <djmitche> #topic Bug 2340 Update
16:39:04 <djmitche> guess not
16:39:24 <djmitche> #topic Turning on protected branches in Github
16:39:46 <djmitche> https://github.com/blog/2051-protected-branches-and-required-status-checks
16:39:54 <djmitche> not sure who suggested this
16:39:57 <tomprince> I did.
16:40:06 <djmitche> oh, heh
16:40:10 <djmitche> small meeting today ;)
16:40:19 <djmitche> so you'd protect master?
16:40:29 <tomprince> I think circleci/appveyor/travis should all be protected.
16:40:32 <djmitche> with required status checks, I think?
16:40:33 <djmitche> ok
16:40:33 <tomprince> master an eight.
16:40:36 <djmitche> right
16:40:41 * sa2ajj nods
16:40:43 <tomprince> s/protected/required/
16:40:44 <gracinet> djmitche: I'm there, not much to say for now
16:40:55 <djmitche> ok I feel less alone :)
16:41:07 <sa2ajj> +1 for protected branches
16:41:09 <tomprince> I probably wouldn't require things be merged forward, given the long time travis takes.
16:41:15 <tomprince> (sadly)
16:41:20 <sa2ajj> we seem to ignore some issues...
16:41:20 <djmitche> tomprince: I guess the downside would be it's difficult to merge PRs when we have a non-code travis failure
16:41:42 <djmitche> of which we've had a few, I think
16:41:44 <tomprince> One can push rebuild in travis, in that case.
16:41:45 * sa2ajj is not sure what's the ratio...
16:42:09 <tomprince> I haven't run into it since I've been back, yet.
16:42:11 <djmitche> yeah, I suppose that's just more pressure to get the travis issue solved
16:42:20 * sa2ajj nods
16:42:22 <djmitche> good datapoint, ok
16:42:30 <djmitche> anyone opposed?
16:42:49 * sa2ajj . o O (yay!)
16:43:10 <djmitche> #info proposal is to protect the 'master' and 'eight' branches, requiring that travis / appveyor / circle all succeed for a PR to be merged
16:43:20 <djmitche> #agreed everyone present seems to agree this is a good idea
16:43:29 <djmitche> tomprince: do you want to set that up? I think you have perms..
16:43:54 <tomprince> Yep. Doing it now.
16:44:42 <djmitche> cool
16:44:50 <djmitche> #topic mainstreaming buildbot_travis
16:45:08 <djmitche> gracinet: tell me more :)
16:45:14 * sa2ajj . o O (i need to have a vacation and look at meetbot...)
16:46:17 <gracinet> ok djmitche, the think is we were discussing last thursday I think of buildbot_travis currently at https://github.com/tardyp
16:46:44 <djmitche> so "mainstreaming" meaning moving it to the bulidbot org?
16:47:00 <gracinet> tardyp and the people around agreed the topic should be brought up to bring it within buildbot
16:47:05 <gracinet> yes that's what it means
16:47:45 <skelly> we will be using it soon too
16:48:04 <skelly> and, hopefully, I'll have time to work on it
16:48:07 <gracinet> from what tardyp explained to me last week, using buildbot_travis for buildbot would also be interesting
16:48:16 <djmitche> indeed
16:48:27 <djmitche> I don't see a reason not to move it
16:48:34 <gracinet> I am interested as well, did some testing… and then had to jump onto unrelated work
16:49:30 <djmitche> #info proposal is to move https://github.com/tardyp/buildbot_travis to the buildbot org, and make it a buildbot product
16:49:31 <gracinet> didn't go that far, but Pierre says it'd open the road for check-by-check reporting
16:49:39 <djmitche> yes, that'd be great
16:49:56 <gracinet> instead of waiting for all jobs
16:49:56 <djmitche> so I think there's general agreement we can move it into the org
16:50:04 <djmitche> using it to test buildbot itself is probably more work
16:50:07 * sa2ajj seconds the motion
16:50:08 <djmitche> but also a good idea
16:50:09 <djmitche> cool
16:50:17 <djmitche> #agreed will move buildbot_travis into the buildbot org
16:50:18 <gracinet> sure, hope I can help on the meta travis buildbot
16:50:24 <djmitche> cool!
16:50:33 <djmitche> #action tardyp to move buildbot_travis
16:50:46 <djmitche> OK
16:50:48 <djmitche> #topic PyCon US
16:51:00 <sa2ajj> i'm not sure that we should assign that action to tardyp :)
16:51:10 <djmitche> well, it's his repo
16:51:14 <tomprince> Well, he is the only one able to do it.
16:51:25 <sa2ajj> without his consensus
16:51:50 <djmitche> gracinet: did tardyp have reservations, or was he in favor of this plan?
16:51:50 * sa2ajj agrees on the points
16:52:20 <tomprince> sa2ajj: He can certainly decline at this point. But he is the only one that can take further action.
16:52:25 <gracinet> I think he was in favor, even suggested this would be one of the topics today, but I didn't watch on friday and monday
16:52:30 <tomprince> Either moving it or declining to move it.
16:52:39 <djmitche> ok, cool
16:52:42 <djmitche> yeah, nothing's written in stones
16:52:45 <djmitche> it's all 0's and 1's
16:52:56 <djmitche> anyway
16:53:06 <djmitche> #info PyCon US is in 2 weeks
16:53:18 <djmitche> I will be there, including for the sprints, although I am not planning any Buildbot sprints
16:53:19 <gracinet> of course, if at this point he disagrees or wants to polish, that'd be his call (actually buildbot_travis is a fork of someone else originally)
16:53:33 <djmitche> right, Jc2k_ started it
16:53:38 <djmitche> I tihnk
16:53:46 <djmitche> memory hazy
16:54:11 <djmitche> #info if anyone wants to meet up at PyCon, organize that on the mailing list
16:54:19 <djmitche> that's all I've got
16:54:45 <djmitche> anything else?
16:55:09 <tomprince> I'm pondering trying to hide workers from builds.
16:55:21 <tomprince> So the onlything builds talk to is the workerforbuilder.
16:55:54 <tomprince> Doing that would allow decoupling having exactly one workerforbuilder per (builder, worker) pair.
16:56:00 <djmitche> I like that
16:56:25 <djmitche> there's some principle popular with Rubyists that involves never using more than one dot (so no self.workerforbuilder.worker)
16:56:36 <djmitche> I can't thin of the name
16:56:47 <tomprince> principle of demeter, I think?
16:57:00 <djmitche> yes
16:57:17 <tomprince> http://c2.com/cgi/wiki/LawOfDemeter?LawOfDemeter
16:58:03 <tomprince> At least ... if I can get some funding to do it.
16:58:04 * tomprince grins
16:58:29 <djmitche> you're getting pretty good at that :)
16:58:50 <tomprince> Well, I've had one contract, so far.
16:59:04 * sa2ajj wonders if there's a book "how to grin in 21 simple steps"..
16:59:11 <djmitche> haha
16:59:37 <djmitche> #endmeeting