16:30:19 <djmitche> #startmeeting weekly
16:30:19 <bb-supy> Meeting started Tue Feb 23 16:30:19 2016 UTC and is due to finish in 60 minutes.  The chair is djmitche. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:30:19 <bb-supy> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:30:19 <bb-supy> The meeting name has been set to 'weekly'
16:30:38 <djmitche> #topic Introduction
16:30:45 <djmitche> #link https://titanpad.com/buildbot-agenda
16:30:46 <infobob> http://paste.pound-python.org/show/T8kKjnaQepkVVJRqnm0I/ (repasted for djmitche)
16:30:51 <djmitche> damnit infobob
16:30:55 <djmitche> you mess up my meeting notes
16:31:18 <djmitche> hopefully a quick meeting today, but please add any items to the agenda as we go
16:31:27 <djmitche> #topic GSoC 2016
16:31:33 <tardyp> hello
16:31:46 <djmitche> #info Dustin and Pierre submitted the application well ahead of schedule - still waiting to hear if we're accepted
16:31:59 <bb-github> [13buildbot] 15rutsky commented on issue #2013: @tardyp ... 02https://git.io/v2Geh
16:32:00 <tardyp> and pierre?
16:32:07 <djmitche> well, as a co-admin
16:32:23 <djmitche> I'll admit I did the bulk of the work ;)
16:32:27 <tardyp> I only clicked on one link. you did all the job :)
16:32:51 <djmitche> #info We have four mentors (two of whom are the admins) signed up -- we'll need a few more!
16:32:54 <rutsky> according to schedule, we're waiting for GSoC application approvat atm?
16:32:58 <djmitche> that's right
16:33:25 <djmitche> https://developers.google.com/open-source/gsoc/timeline
16:33:37 <djmitche> #info Accepted orgs will be announced on February 29
16:34:57 <djmitche> #info we've had a few interested potential students as well
16:35:10 <djmitche> which is great -- very early!
16:35:39 <djmitche> verm__: you around to talk about your tool?
16:36:35 <djmitche> #topic Bug 2340
16:36:53 <djmitche> #info The *big* refactor PR has landed!
16:36:58 <rutsky> yay! :)
16:37:05 <djmitche> #info for the master/ subdirectory, anyway
16:37:10 <djmitche> rutsky: so what's next?
16:37:12 <skelly> how mnay commits and lines changed?
16:37:17 <rutsky> I'm slowly working towards slave part
16:37:17 * djmitche looks
16:37:37 <rutsky> I have bunch of small related independant fixes to land and working with at atm
16:38:19 <djmitche> #info Next steps: buildbot-slave -> buildbot-worker, small fixes to land
16:38:29 <tardyp> oh, I did not realize the slave part did not land..
16:38:37 <djmitche> #info Change freeze has been lifted -- PRs can now be merged by anyone with permission, per the policy
16:38:49 <rutsky> atm there are following PR waiting for review related to 2340: https://github.com/buildbot/buildbot/pull/2012
16:38:49 <tardyp> I even remember that there were chunks about slave part..
16:39:31 <rutsky> yes, some of those slave-chunks will go away (e.g. in local build worker support)
16:39:39 <djmitche> oh, looks like the landing was a fast-forward so I don't have a great answer for you skelly
16:39:40 <skelly> to answer my own question: 694 commits, 329 files changed, +3267, -3404
16:39:40 <djmitche> "a lot"
16:39:45 <djmitche> ah, great
16:39:47 <skelly> per GH
16:39:51 <rutsky> most of them will stay, to support old buildslave
16:40:23 <rutsky> I think GH lies
16:40:42 <rutsky> it's more like +12K, -12K
16:40:55 <djmitche> tardyp: just slave/buildslave/null.py
16:42:11 <tardyp> ok
16:44:01 <skelly> rutsky: you're right, checking the merge commit and your two commits after gives: 340 files changed, 17451 insertions(+), 11724 deletions(-)
16:44:23 <bb-github> [13buildbot] 15rutsky commented on issue #1995: @adityadivekar03 ... 02https://git.io/v2GTY
16:44:35 <rutsky> Hm... + 5K lines... I'm good at writing tests! :)
16:45:10 <djmitche> #info merge statistics: 340 files changed, 17451 insertions(+), 11724 deletions(-)
16:45:20 <djmitche> #topic Development week-in-review
16:45:52 <cmouse> what
16:45:57 <cmouse> 's the best way to suggest features?
16:46:06 <djmitche> cmouse: bugs on trac.buildbot.net
16:46:08 <cmouse> given that I can't make a trac account for some reason  =)
16:46:13 <bb-github> [13buildbot] 15tardyp opened pull request #2016: fix problem where parent_changeids might be corrupted (06master...06bug3461) 02https://git.io/v2Gkr
16:46:13 <cmouse> it considers me spam =)
16:46:18 <djmitche> oh, ugh
16:46:26 <cmouse> even after captcha
16:46:31 <rutsky> :angry:
16:46:37 <djmitche> can you email me? dustin@buildbot.net?  I'll get that straightened out
16:46:43 <djmitche> Trac's spam-fighting stuff is AWFUL
16:46:52 <rutsky> is cmouse is NOSPAM group?
16:46:54 <djmitche> I think if you just added "not" to it somewhere, it might work better
16:47:02 <djmitche> not if the account doesn't exist yet
16:47:06 <tardyp> cmouse: I created a bug for the problem you reported: http://trac.buildbot.net/ticket/3461
16:47:11 <cmouse> tardyp: thanks
16:47:14 <tardyp> I just submitted a potencial fix
16:47:14 <djmitche> #info a few minor PRs merged after beta7
16:47:40 <cmouse> my feeble suggestion is to have Git step to use repository url from changes if not given
16:47:45 <rutsky> oh, cmouse can't be in NOSPAM since it's can't even register...
16:47:46 <tardyp> please let me know if this fixes your issue. For me your database has some kind of corruption
16:47:57 <djmitche> #info lots of open PRs to get merged this week
16:48:03 <cmouse> tardyp: i can still rm -rf the setup
16:48:06 <cmouse> tardyp: and see if the database gets fixed
16:48:18 <djmitche> #topic PR 1994: install service_identity in Travis to silence Twisted's warnings
16:48:56 <cmouse> i wrote an almost oneline GitlabGit step
16:49:03 <cmouse> but I suppose someone else might find it useful, too
16:49:30 <cmouse> it helps doing CI in gitlab for forks
16:49:34 <djmitche> rutsky: per the PR it looks like the plan is to add service_identity to the test requirements and mention it in the docs?
16:49:49 <rutsky> in 1994 I propose to install service_identity in Travis to silence warnings
16:49:56 <tardyp> djmitche: looks fine for me
16:50:03 <cmouse> rutsky: in 2016 it's finally going there? =)
16:50:19 <djmitche> haha
16:50:32 <cmouse> rutsky: only 22 years of patient waiting...
16:50:34 <djmitche> rutsky: so are there open questions on this PR?
16:50:47 <rutsky> he-he)
16:51:18 <rutsky> initially I though only about silencinf warning
16:51:30 <rutsky> should be document this?
16:51:48 <djmitche> that it's required for silencing the warning? yes, I think so
16:51:51 <rutsky> this _should_ into requirements.txt when we will create ont
16:51:57 <djmitche> yes
16:53:13 <djmitche> ok
16:53:18 <bb-github> [13buildbot] 15rutsky commented on issue #1994: According to discussion in IRC I need to document this issue too (add note to the docs about how to fix this issue if it occurs). 02https://git.io/v2GLa
16:53:18 <rutsky> OK, I'll think about documenting this
16:53:21 <djmitche> #topic PyCon Europe, US
16:53:31 <djmitche> #info PyCon Europe 2016: event is July CFP deadline is march 6th
16:53:39 <djmitche> #info PyCon US May 28-June 5, Portland, OR -- https://us.pycon.org/2016/
16:53:40 <rutsky> when we will create requirements.txt I think we will move it there
16:53:50 <djmitche> I think that sounds like a good plan
16:54:59 <djmitche> so I'm planning to go to PyCon US
16:55:00 <tardyp> So I was planning to hit the CFP this year for pyconeu
16:55:07 <djmitche> proposing what?
16:55:16 <tardyp> at present the buidbot nine developmenets
16:55:21 <djmitche> great!
16:55:41 <tardyp> I think this is a great time, and was hoping we can release before July
16:55:51 <djmitche> If there are enough buildbotters at either conf, it would be good to have a birds-of-a-feather session
16:55:55 <djmitche> I hope so too!!
16:56:27 <djmitche> so I'll send an email to devel@, but is anyone on the channel going to either PyCon?
16:57:31 <djmitche> heh, don't all yell at once you'll crash freenode
16:57:54 <tardyp> I think I'll go pyconeu
16:58:01 <djmitche> #action djmitche to send email about both PyCons to the devel@ and user@ lists
16:58:08 <djmitche> I figured if you're proposing you were planning to go ;)
16:58:24 <tardyp> I was trying to end the silence.
16:58:36 <djmitche> #topic PR 3459 - using CircleCI to build and upload docs
16:58:52 <djmitche> rutsky: to be honest I think this is a lot of work for a relatively minor feature
16:59:01 <tardyp> I agree
16:59:10 <rutsky> djmitche: I don't think this is a minor feature
16:59:21 <tardyp> the setup time is done now, but we'll have to maintain it
16:59:22 <djmitche> how so?
16:59:35 <rutsky> proper docs are priority #2 (after priority #1 that BB works :) )
17:00:06 <djmitche> anyway, I'm happy to see you do it if you feel it's important
17:00:12 <tardyp> I think the docs can be fixed post commit if there are some things that we did not see with review
17:00:18 <rutsky> I'm quite disappointed about some parts of docs, especially when docs not correspond to the code
17:00:48 <rutsky> and I hope to do something with Sphinx configuration to make it less error-prone
17:00:50 <tardyp> well, this part will not be fixed by publishing the docs for the PRs
17:01:18 <tardyp> we verify that the docs build pre-commit, for me this is enough
17:01:24 <myheadhurts> skelly: ansible force color is set. That's what I meant by we force it
17:01:31 <rutsky> with ability to see how docs are actually seen after built we would be able to see final result of changes
17:01:49 <tardyp> you can argue the same for UI patches
17:01:55 <rutsky> it's a shame that we have trivial formatting errors
17:02:00 <tardyp> we usually submit screen shots of the results
17:02:11 <tardyp> which github supports well
17:02:27 <rutsky> tardyp: exactly! I would like to see some e2e working UI for each PR too, but this is significantly harder
17:02:41 <tardyp> we can ask people to submit screenshot of the docs generated
17:03:01 <tardyp> its not about e2e tests, they wont tell you how nice the UI feel
17:03:08 <djmitche> the docs divergence is bad, sorry about that :(
17:03:19 * djmitche in another meeting now.. should we wrap this up?
17:03:25 <djmitche> we can keep talking obvi
17:03:28 <rutsky> let me conclude :)
17:03:45 <rutsky> IMO it's work a day or two of work if it will work
17:03:58 <rutsky> with Travis + S3 it doesn't work
17:04:06 <tardyp> we can enable it right now, and if it breaks too much we disable
17:04:06 <rutsky> and looks like it should with CircleCI
17:04:13 <rutsky> tardyp: agree
17:04:27 <rutsky> is there any objections for enabling CircleCI
17:04:28 <rutsky> ?
17:04:33 <djmitche> no
17:04:40 <rutsky> e.g. is someone don't trust permissions they ask?
17:04:57 <tardyp> #info we will enable circleci for docs generation and see how it behave
17:05:18 <rutsky> or is the fact that they are propriatary service bothers someone? (Travis is Open Source AFAIK)
17:05:35 <tardyp> travis is not *really* open source
17:05:52 <tardyp> I dont think there are any instance of travis outside of the official
17:06:03 <rutsky> tardyp: they haven;t open sources some of their components?
17:06:13 <tardyp> it is very dependent of their infra
17:06:29 <rutsky> I don't really investigated this topic, just saw that at least some of their tools are on GitHub
17:06:31 <tardyp> and there is not really an easy step by step installation doc afaik
17:07:02 <tardyp> anyway, I dont have objection for circleCI
17:07:08 <rutsky> ok
17:07:17 <rutsky> I have another topic: do we have GitHub bot?
17:07:29 <tardyp> yes we do
17:07:37 <tardyp> bb-github
17:07:57 <rutsky> if CircleCI will work I would like to add rule to publish direct link on built docs to PR status using GitHub status API
17:08:25 <rutsky> like Travis marks commit in PR as "OK" or "FAIL" with link on build results, but with link on built docs
17:08:34 <tardyp> it shall be possible
17:08:40 <rutsky> is bb-github capable to do this?
17:08:43 <tardyp> no
17:08:51 <tardyp> bb-github is the ircbot
17:09:13 <tardyp> you should look at the github api, create a client credential
17:09:16 <rutsky> oh, and do we have bot like Hubot, that listens for Github events?
17:09:30 <tardyp> there is then a status api that you can call from circleCI
17:09:48 <tardyp> in buildbot, I think we have github changesource
17:10:11 <rutsky> is github changesource works for pull request?
17:10:14 <tardyp> for listening to events, and I think we had a status report, but needs to be ported to nine reporter api
17:10:25 <tardyp> I think so
17:10:42 <rutsky> hm, I think we can make it work for PRs and update commit status on Github!
17:10:42 <tardyp> the problem for that is security
17:10:57 <tardyp> you need well isolated workers
17:11:19 <tardyp> because the PR owner can take control of your workers
17:11:37 <rutsky> why? I want bot that waits for "build" finish event from unsecure worker (e.g. CircleCI) and updates status on Github
17:11:52 <tardyp> travis and circle CI did take care of that, but metabb does not have isolated servers for testing the PR
17:11:57 <rutsky> I don't want to update Github status from inside of worker
17:12:39 <tardyp> well for that, I think it is best to do that as a small post process script in the end of circleCI
17:12:40 <rutsky> I would run on metabb only task with fixed steps, no PR source content will be used
17:12:46 <tardyp> rather than doing complex event processing
17:13:21 <rutsky> if post process script will require secutity tokens to update github status, than I can't use them
17:14:06 <rutsky> I think "wait for Github event and do some work with Github API" is ideal use case for Github bots like Hubot
17:14:20 <rutsky> however I didn't worked with them, so I may be wrong.
17:14:48 <rutsky> OK, let's continue this discussion after meeting :)
17:14:59 <tardyp> right
17:16:14 <tardyp> #endmeeting
17:16:29 <cmouse> do these tags actually mean something here? =)
17:16:34 <cmouse> like, cause action or something?
17:16:36 <rutsky> looks like only djmitche can do this :)
17:16:46 <djmitche> #endmeeting