17:00:14 <djmitche> #startmeeting weekly
17:00:14 <bb-supy`> Meeting started Tue Jan  3 17:00:14 2017 UTC and is due to finish in 60 minutes.  The chair is djmitche. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:14 <bb-supy`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:14 <bb-supy`> The meeting name has been set to 'weekly'
17:00:36 <djmitche> #topic Introductions
17:00:41 <rutsky> hi!
17:00:41 <tardyp> hello
17:00:43 <djmitche> Happy new year :)
17:00:51 <djmitche> https://titanpad.com/buildbot-agenda
17:00:53 <tardyp> best wishes for everyone!
17:01:08 <rutsky> happy new year!)
17:01:56 <dol-sen> happy new year to all here :)
17:02:00 <djmitche> :)
17:02:14 <djmitche> no meeting last week, so we're spanning two weeks this time
17:02:16 <djmitche> but still pretty quiet
17:02:33 <dol-sen> tis the season :)
17:02:55 <djmitche> looks like most PRs are docs-related
17:02:57 <djmitche> that's a good sign
17:03:19 <djmitche> #topic Two Weeks in Review
17:03:28 <tardyp> Indeed, nothing fancy.
17:03:29 <djmitche> Any other updates?
17:03:39 <djmitche> tardyp: looks like we got new Hyper tokens -- anything to know about there?
17:03:41 <tardyp> we had not even one trac bug this week
17:04:13 <tardyp> well I am not sure exactly what happen, but there is a script at hyper that ban account without credit cards after a while
17:04:38 <djmitche> oh :(
17:04:44 <tardyp> we were one of those, and they had to apply a special coupon code in order to make our account an exception
17:04:51 <djmitche> haha, oops
17:04:59 <tardyp> anyway, this is now resolved, and the CI is back on track
17:05:01 <djmitche> awesome
17:05:11 <djmitche> #info very light two weeks
17:05:17 <djmitche> #info PRs for docs, mostly, and no bug updates
17:05:27 <djmitche> #topic Releases
17:05:35 <tardyp> and buildbot was pretty nice at not spamming the API retrying with the new quarantine code :)
17:05:39 <djmitche> 0.9.2 is still the latest in the nine series
17:05:41 <djmitche> awesome :)
17:06:02 <tardyp> its beg of month, so I'm planning to make 0.9.3
17:06:06 <djmitche> We should figure out what's going on with 0.8.x
17:06:08 <djmitche> ok
17:06:13 <djmitche> #info 0.9.3 coming soon
17:06:21 <djmitche> what are the headlines for 0.9.3?
17:06:33 <tardyp> not much more than the weekinreview
17:06:56 <tardyp> but a couple of bugfixes and small improvement
17:07:07 <djmitche> great
17:07:12 <djmitche> a calm release is a good release :
17:07:12 <djmitche> :)
17:07:17 <tardyp> there is the SetProperties also that is being discussed right now
17:07:21 <djmitche> #info incremental bugfixes and improvements
17:07:27 <rutsky> I'm want to complete bug 2340 (slave->worker renaming) during this week, though it's not tied to BB releases
17:07:40 <tardyp> rutsky: marvelous!
17:08:33 <djmitche> rutsky: yeah, it's sooo close to the end, let's get it done :)
17:08:48 <djmitche> #topic Mail issues
17:09:11 <djmitche> I checked the bounce logs on the mailman server today, and there have been a few bounces that very well may be bouncing, but nobody has been removed since last check
17:09:26 <tardyp> verm__: did you have a chance to look at the lab network instabilities?
17:09:29 <djmitche> #info no new removals, but nothing has changed to "solve" the issue
17:09:58 <tardyp> I realize that I wasn't in the sysadmin list yesterday..
17:11:54 <djmitche> :(
17:12:00 <djmitche> I think I'm off that list too.. should check
17:12:18 <djmitche> #topic irc logs
17:12:28 <djmitche> rutsky: how do you think we should proceed/
17:13:01 <rutsky> I think we should 1) investigate  hosted alternatives to botbot.me
17:13:29 <djmitche> ok
17:13:32 <rutsky> 2) if there is no any, either stick with Jack Grigg solution, or investigate how to host own solution (and which one)
17:13:38 <djmitche> is jackgrigg so bad?
17:13:50 <rutsky> IMO no, jackgrigg is good enough
17:14:04 <rutsky> it logs everything and it's searchable using Google etc
17:14:16 <djmitche> yeah
17:14:27 <rutsky> it's not new and shiny --- the only "problem" with it :)
17:14:27 <tardyp> jackrigg has been restored. nice
17:14:28 <djmitche> maybe we should just stick with that?
17:14:30 <djmitche> haha
17:14:38 <djmitche> I don't think we need to make extra work for ourselves
17:14:55 <tardyp> it was down for a while, and I though we didn't have anything anymore
17:15:00 <tardyp> so +1 lets keep it that way
17:15:37 <rutsky> agree, at least until hosted alternative will be found which doesn't require much work
17:16:19 <djmitche> ok
17:16:28 <djmitche> do you want to close the ticket up, then, rutsky?
17:16:58 <rutsky> let me check alternatives and I will close it if I will not find anything
17:17:26 <rutsky> #action rutsky check botbot.me hosted alternatives and close ticket if there is nothing suitable there
17:17:34 <djmitche> ok
17:17:42 <djmitche> #topic trac -> github issues
17:17:53 <djmitche> I don't think anything's changed here..
17:18:01 <rutsky> ping sa2ajj ?
17:18:32 <djmitche> I'll ask in the bug
17:18:42 <tardyp> he has been silent during the holiday season
17:18:45 <djmitche> yeah
17:18:49 <djmitche> that's OK, most of us have :)
17:18:53 <tardyp> yep
17:20:59 <bb-trac> [trac] #3623/task (assigned) updated by dustin (@sa2ajj, what do you think?    I think we should not let the perfect be the enemy of ...) http://trac.buildbot.net/ticket/3623
17:21:06 <djmitche> #topic     Distribute single-file Buildbot-worker using something like pyinstaller
17:21:11 <rutsky> docker-compose is written in Python and distributed in single file - I knew that there are projects for this, but always thought that this is *very* hackish and unusable in practice solution.
17:21:13 <djmitche> rutsky: ^^ do tell..
17:21:59 <tardyp> you mean with all the deps in it?
17:22:05 <rutsky> it would be nice to have builbot-worker installable just by copying single executable (worker binary)
17:22:08 <rutsky> tardyp: yes
17:22:20 <tardyp> for the worker I agree indeed
17:22:40 <rutsky> this is "new and shiny" approach actively popularizing by Go programs now
17:22:40 <tardyp> its just that we have been using pypi for all our releases
17:23:08 <tardyp> so it would mean having another release area
17:23:23 <rutsky> tardyp: yes, I don't think it's pypi friendly
17:23:46 <djmitche> does that one file run directly, or does it "unpack" to install a whole virtualenv?
17:23:59 <tardyp> also we have a bunch of deps that need some c extensions (e.g twisted)
17:24:20 <tardyp> so this means we have to take care of making one release per platform
17:24:23 <rutsky> djmitche: in case of docker-compose I think it behaves as it runs directly, not sure is it unpacking in background anything
17:24:30 <tardyp> like the go guys need to do
17:24:53 <tardyp> I have some experience with python packers
17:25:00 <bdbaddog> I've seen a bunch of packages which wrap all dependencies into a single runnable file.. Though as you said.. twisted brings in native binary needs.
17:25:16 <djmitche> I think those deps are optional for twisted?
17:25:19 <rutsky> tardyp: yes, since this is single executable it should be built for i686/amd64/windows etc, probably PyInstaller have something to help with building all this
17:25:20 <tardyp> it does put a zip of everything in that file, and use the file that python can run within zipfiles
17:25:44 <tardyp> djmitche: twisted itself has cext for optimization
17:25:51 <bdbaddog> pex is one of the solutions in this space I beleive.
17:26:09 <tardyp> yes, I used pex already, but only for linux
17:26:18 <djmitche> yeah, I think if those extensions aren't available, twisted falls back to something slower
17:26:22 <tardyp> and only for ubuntu, so this was pretty simple
17:27:03 <tardyp> djmitche: I know that I was never able to install twisted without python-dev installed. so I am not sure we can
17:27:40 <tardyp> I think windows installer makes sense indeed. but for linux I doubt it
17:28:07 <tardyp> and osx/bsd as well, they all have decent pip installation
17:28:10 <bdbaddog> If you build python rather than install it, you won't need python-dev installed..
17:28:29 <bdbaddog> isnt' there a pypi wheel for twisted bits on windows?
17:28:45 <tardyp> bdbaddog: yes, I was just using that arguement as a clue that twisted cannot run without its c-ext
17:29:02 <tardyp> twisted does not release wheels afaik
17:30:40 <djmitche> I see
17:30:46 <tardyp> somebody wants to make some experiments?
17:30:49 <rutsky> so, is there interest in trying this?
17:31:01 <rutsky> I can create ticket for this and try to experiment
17:31:02 <djmitche> yes, I think this would help adoption substantially
17:31:17 <djmitche> #info it would be nice to have builbot-worker installable just by copying single executable (worker binary)
17:31:17 <tardyp> I think this could make sense if this is not too much burden
17:31:44 <rutsky> tardyp: yes, only if this doesn't require too much efforts
17:32:01 <djmitche> OK, we're a little over time -- anything else to discuss?
17:32:18 <rutsky> nothing from me
17:32:23 <tardyp> I remember trying to do a windows installer of a proprietary python program(before pyinstaller) and this was not a good experience
17:32:33 <tardyp> I made some progress on the benchmarks
17:32:52 <tardyp> buildbot nine is 4x slower thatn eight at the bench of storing logs
17:33:19 <tardyp> 10k lines/s compared to 40k lines/s
17:33:48 <tardyp> which is not that bad regarding the much hight complexity of the logchunk handling
17:34:17 <djmitche> #topic Efficiency benchmarks
17:34:33 <djmitche> #info buildbot 9 can store 10k lines/s compared to 40k lines/s
17:35:00 <djmitche> #info but brings a host of new features including storing them in a database, so not bad
17:35:11 <djmitche> #endmeeting