11:18:52 <lbt> #startmeeting Mer Release Management Meeting 15 May 2012
11:18:52 <MerBot> Meeting started Tue May 15 11:18:52 2012 UTC.  The chair is lbt. Information about MeetBot at http://wiki.merproject.org/wiki/Meetings.
11:18:52 <MerBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
11:19:42 <lbt> So our release process still has a few manual steps which have tripped us up a bit
11:20:21 <lbt> however we put out Mer PRE-RELEASE 0.20120517.0.1 on sunday
11:21:09 <lbt> that's in cobs and seems to be working acceptably
11:21:36 <Stskeeps> :nod:
11:21:40 <Stskeeps> qt5 is partially merged as well
11:21:58 <lbt> Ideally we'd get vendor reports from Sage, mdfe and others if there were problems
11:22:36 <lbt> I saw qt5 changes yesterday which I guess will go into the release?
11:22:56 <Stskeeps> yes
11:22:59 <Stskeeps> some modules missing
11:23:01 <Stskeeps> like no qt quick 2 yet
11:24:02 <lbt> hey mdfe_ ... just saying the pre-release is out and wondering if you (or other vendors) had any comments/issues
11:24:23 <mdfe_> hi all,
11:24:50 <mdfe_> yesterday I build a mer-next image for plasma active
11:25:27 <mdfe_> It works for me
11:25:37 <lbt> good
11:25:41 <Sage> I'm going to build nemo today against -next
11:26:40 <Stskeeps> ok
11:26:45 <Stskeeps> so we can probably do go/no go tomorrow
11:27:35 <lbt> yep - anything else wrt this release then?
11:27:49 <mdfe_> not from my side
11:28:13 <Stskeeps> lbt, can you walk through how you attempted copy project?
11:28:19 <lbt> #info Mer PRE-RELEASE 0.20120517.0.1 seems good with PA and Nemo
11:28:31 <lbt> OK
11:28:56 <lbt> The main thing I started with was the test environment
11:29:20 <lbt> so that's the OBS deployment thing - and getting some test projects setup
11:29:42 <lbt> The code was in the backend and had a passthrough in the api server too
11:30:02 <lbt> the main work has been in writing a small osc command for it
11:30:10 * Sage would like to hear more background. Copyproject for vendors or for mer core?
11:30:14 <lbt> that invokes the api now
11:30:34 <lbt> yes - it uses phaeron's patch from way back
11:30:50 <Stskeeps> Sage: both
11:31:20 <lbt> Sage: we'll apply it to the Mer OBS packages and push it upstream ASAP
11:31:39 <lbt> currently it "does stuff" but it's not "working"
11:31:51 <Sage> ah, ok
11:32:13 <lbt> eg the project ends up needing a manual kick to actually rebuild - and the point is also to avoid rebuilds that are not needed
11:32:15 <phaeron> lbt: so which part worked and which didn't
11:32:31 <lbt> phaeron: I think it's racy
11:32:38 <lbt> so it changes
11:32:45 <Stskeeps> can you walk step by step through, by chance?
11:32:56 <phaeron> lbt: you mean kick it to rebuild because it is empty ?
11:33:02 <Stskeeps> copy-project is pretty much the blocker for a lot of things, so i think we can take the time in the meeting to sort it out
11:33:11 <phaeron> agreed
11:33:20 <phaeron> and I can invest some more time in it
11:33:48 <Stskeeps> on my side i'm working on sb2-b1-obs as usual inbetween all the other stuff i do
11:34:08 <lbt> phaeron: so it can do a rebuild or it can sit in "unresolveable" or "broken"
11:34:39 <phaeron> lbt: ah broken state I haven't seen. I need to try locally.
11:35:29 <lbt> Stskeeps: so phaeron's patch added "withbinaries" to the existing copyproj method
11:36:04 <lbt> looking for link
11:36:47 <lbt> https://github.com/Merproject/open-build-service/commit/c2c4731638d16a67ac0b3432207d78188b31f91f
11:37:05 <phaeron> yes
11:37:08 <Stskeeps> :nod:
11:37:12 <phaeron> lbt: did you try without it
11:37:52 <lbt> can't recall - pretty sure I did
11:38:06 <phaeron> ok
11:38:12 <Stskeeps> okay, and you set up a project that is meant to be copied at a later point, how?
11:38:44 <lbt> scripted as part of the VM setup - split out into a testing operation
11:38:58 <Stskeeps> ok, got a meta prj for the project?
11:40:20 <lbt> here's the basic setup ... https://gist.github.com/2701072
11:40:41 <lbt> so I setup MDS and a prj building against it and then copypac acl
11:40:56 <Stskeeps> ok, and then you copyprj MyMerUx?
11:41:21 <Stskeeps> (how?)
11:41:39 <lbt> ./osc-wrapper.py --debug -t -A https://obsfe.dgreaves.com:444 copyprj -b -o MyMerUx MyMerUx4
11:42:31 <Stskeeps> and then what happens?
11:43:28 <lbt> https://gist.github.com/2701096
11:43:46 <Stskeeps> ok
11:43:50 <Stskeeps> and that causes..?
11:43:53 <Stskeeps> just walking through it :)
11:43:58 <lbt> my VMs are down so I can't grab the logs :/
11:44:16 <lbt> essentially it's inconsistent
11:44:22 <lbt> the copy works
11:44:29 <lbt> but it does seem to rebuild
11:44:47 <lbt> although it gets stuck in a bad state
11:44:50 <lbt> um
11:45:26 <lbt> so it has 2 targets... and I've seen one target build OK and the other be 'stuck'
11:45:46 <phaeron> maybe it rebuilds because you asked it to rebuild. Did you try fixing the bad post copyprj state by doing obs_admin check_project on the admin
11:45:53 <phaeron> *on the backend
11:45:53 <lbt> obs_admin  --deep-check-project MyMerUx3 armv8el
11:46:16 <lbt> so yes
11:46:17 <phaeron> yeah what happend when you do that
11:46:21 <lbt> it builds
11:46:53 <Stskeeps> ok, so two things: it shouldn't rebuild upon copy, and it ends up in a bad state
11:47:48 <lbt> yes
11:48:01 <phaeron> bad state is fixed by the check so I think that function is missing a "ping" to the scheduler to notice the new project contents
11:48:13 <Stskeeps> phaeron: enough for you to debug with?
11:48:29 <phaeron> Stskeeps: I believe so , I'll start working on it again tonight
11:48:33 <Stskeeps> ok, cool
11:48:44 <phaeron> I'll try to sync with lbt
11:48:47 <lbt> so I'm working on the backend code too atm ... sync with me
11:48:50 <lbt> :)
11:48:57 <phaeron> lbt: :) same area =
11:49:01 <phaeron> ?
11:49:19 <phaeron> I mean are you working on that same feature ?
11:49:23 <lbt> I think it picks up the changes too quickly ... almost needs a lock
11:49:25 <lbt> yes
11:49:40 <phaeron> ok
11:50:28 <lbt> anyhow... this feature is currently my main OBS work
11:50:50 <phaeron> it does do a ping at the end of the function ..
11:50:51 <lbt> once it's done we can do another Mer OBS release based on 2.3.0
11:51:05 <phaeron> https://github.com/Merproject/open-build-service/blob/c2c4731638d16a67ac0b3432207d78188b31f91f/src/backend/bs_repserver#L1493
11:51:40 <lbt> where does it get the binaries from ?
11:52:02 <lbt> I expected a copy of :full
11:52:21 <lbt> which would appear in a publish
11:52:48 <phaeron> lbt: https://github.com/Merproject/open-build-service/blob/c2c4731638d16a67ac0b3432207d78188b31f91f/src/backend/bs_repserver#L1449
11:52:54 <phaeron> it links specific parts
11:53:09 <phaeron> and then asks the scheduler to look at this project
11:53:12 <Stskeeps> once we're past this hurdle i'd like to see if we can do copyprj of a <link project="X"> project to project B, then change to <link project="Y"> in project B and have it do what we hope it does
11:53:32 <lbt> phaeron: ah, got it
11:53:52 <lbt> Stskeeps: yep
11:55:39 <phaeron> ok
11:56:00 <lbt> #info OBS copyprj command with no-rebuild is progressing. osc interface present. Some debugging to do to resolve state inconsistencies and the fact that it does still rebuild.
11:56:18 <Stskeeps> any more things?
11:56:28 <lbt> Not on that topic
11:56:36 <phaeron> I still think it rebuilds because you ask it to rebuild :D
11:56:55 <lbt> I'll check - it was pre-SF so I need to revalidate
11:57:09 <lbt> I want to do a rebase to 2.3.0 first
11:57:16 <lbt> we should review things from Tizen conf though
11:57:25 <lbt> eg spectacle
11:57:44 <lbt> OBSLight
11:57:45 <Stskeeps> yeah, spectacle's dead in upstream, we can modify it to our own needs
11:58:05 <phaeron> like making it generate debian packaging ?
11:58:05 <Stskeeps> Anas was working on some kind of 'make' like build system for distributions as well
11:58:33 <phaeron> Stskeeps: reinventing openembedded bitbake ?
11:59:05 <Stskeeps> not entirely sure
11:59:17 <Stskeeps> but yes, everything's just a reinvention of gentoo and OE these days
11:59:18 <Stskeeps> :P
11:59:43 <phaeron> ahh gentoo :)
11:59:53 <phaeron> those were the days ..
12:00:03 * phaeron hopes no one noticed :D
12:00:27 <Stskeeps> hehe
12:00:49 <Stskeeps> they use a bit different git like system for tizen but the intention is the same
12:01:03 <phaeron> seriously though debian packaging using spectacle would be cool
12:01:05 <Stskeeps> make OBS be more of a builder and less like a full does-everything system
12:02:08 <lbt> I spoke to Adrian about our areas of work for OBS too and he seemed pretty happy with them
12:02:08 <Stskeeps> they had problems with pristine-tar and gerrit
12:02:46 <lbt> Stskeeps: yeah - I'm worried about gerrit handling complex multi-commit changesets
12:02:55 <lbt> I just don't think it will
12:03:20 <Stskeeps> it can with topic branches, which is tied into MDS2 and CI again..
12:03:21 <Stskeeps> :P
12:03:26 <lbt> what I want is a "pull request review"
12:03:55 <lbt> *nod*
12:04:38 <lbt> so : my git/packaging support - Adrian liked the idea and pointed me at some preliminary work which may help
12:06:00 <lbt> the copyprj idea was less well received - but I don't think I explained the use case too well
12:06:38 <lbt> sb2 he knows about of course
12:06:40 <phaeron> the copyprj code is their own to begin with :D
12:06:53 <lbt> phaeron: yeah - the reason we need it for images
12:07:19 <Stskeeps> and any kind of sane CI as well
12:07:28 <lbt> he's also OK with us continuing to work on BOSS too
12:07:28 <phaeron> lbt: we need it for images ?
12:08:04 <Stskeeps> phaeron: we cannot do full-repo builds otherwise
12:08:06 <lbt> yeah - Stskeeps says mic has issues - II believe him so I've not tried it
12:08:47 <phaeron> mm.. maybe we need to write a wiki page describing what we aim to achieve with this feature
12:08:51 <lbt> Stskeeps: I got the feeling there was more interest in an OBS technical mini-conference too
12:09:05 <lbt> phaeron: I clarified it
12:09:22 <Stskeeps> phaeron: for me the most important thing is the copy-prj+link switch thing, as it takes a lot of the crap out of MDS daily usage
12:10:02 <lbt> http://wiki.merproject.org/wiki/Process#New_process_2
12:10:33 <Stskeeps> i -manually- linkpac atm
12:10:34 <Stskeeps> :P
12:11:28 <mdfe_> sounds very interesting
12:12:01 <lbt> mdfe_: needed to permit better automation of changeset checks
12:12:44 <phaeron> lbt: ok thanks that's similar to what I was looking foe
12:12:45 <phaeron> for
12:13:00 <phaeron> sometimes I forget why I am working on a certain feature
12:13:37 <lbt> #info Post Tizen conference: Spectacle will be maintained by Mer now
12:14:01 <lbt> Stskeeps: did you get any details on the tizen git packaging thing ?
12:14:12 <Stskeeps> no, but review 'gbs' and it makes more sense
12:14:17 <lbt> ok
12:14:46 <lbt> ah yes
12:15:32 <lbt> Tizen will be adopting the version numbering I proposed for MeeGo almost 1.5 years ago
12:15:41 * lbt looks smug
12:18:55 <mdfe_> afk
12:19:11 <lbt> Other stuff...
12:19:46 <lbt> #info We've had donations for another machine (Thanks to vgrade / Martin Brook)
12:20:39 <Stskeeps> ah, good
12:20:42 <lbt> osuosl finally responded to requests too : "We are definitely interested in hosting the Mer project"
12:21:35 <lbt> They could host 2-3 non-build boxes at OSL to start with ... I'll take this up with them
12:21:43 <Sage> nice
12:22:19 <lbt> I spoke to Quim and I think asking Qt for support would be worthwhile too
12:22:37 <lbt> I need to make a case for that
12:23:01 <Stskeeps> could be good to have Mer reference in similar places like ubuntu and others are
12:24:17 <lbt> *nod* I like the idea of hosting the phosts at osuosl - security+mgmt is now rather tight :)
12:25:29 <lbt> I think that's it from me - this week should be focused on copyprj ... then maybe some hacking on git and versioning and CI
12:25:41 <Stskeeps> yeah
12:25:54 <lbt> I still need to sort out a release process for Tools, MINT and SDK too
12:26:01 <lbt> the SDK release is a bit painful
12:26:09 <lbt> as are .ks
12:26:15 <lbt> they're too inter-related
12:26:25 <lbt> but.... that's a future task
12:27:24 <phaeron> cue end music ?
12:27:30 <Stskeeps> zzz?
12:27:30 <Stskeeps> :P
12:27:35 <phaeron> *I will survive*
12:27:43 <lbt> oh ... I am using MerProject github, not MeeGoIntegration
12:27:55 <lbt> we should merge them
12:28:13 <lbt> but yes... anyone else?
12:28:16 <lbt> 3
12:28:19 <lbt> 2
12:28:24 <lbt> 1
12:28:33 <lbt> thank you all for coming :)
12:28:34 <lbt> #endmeeting