11:01:07 #startmeeting Mer release management meeting 12/6/2012 11:01:07 Meeting started Tue Jun 12 11:01:07 2012 UTC. The chair is Stskeeps. Information about MeetBot at http://wiki.merproject.org/wiki/Meetings. 11:01:07 Useful Commands: #action #agreed #help #info #idea #link #topic. 11:01:18 o/ 11:01:21 o/ 11:01:48 Sage ------|= 11:01:54 phaeron ------|= 11:01:55 not around, vacation 11:02:09 ah yes ... so stabbing him with a sword won't help 11:02:47 #info ARMv6 had been broken since 0405, fixed by disabling Cortex string routines in glibc for armv6 11:03:15 #info connman 1.1 merged, locale archives fixed (295), libjpeg-turbo-devel (225) 11:04:07 #info one testing round before release probably good idea, snapshot tomorrow? 11:04:11 prerelease 11:04:18 was on phone 11:04:21 'sorry 11:04:24 OK 11:04:26 phaeron: :) 11:04:32 sage is on vacation 11:04:40 ouch stabbed 11:06:50 so what's new from your side? i was gone for a week 11:07:19 I was looking at mds to see if it can run without git 11:07:26 can it wrap a bare repo ? 11:07:39 it doesn't need it as such for non-source 11:07:48 for source it does, atm 11:07:59 but i think the intel guys patched it to not use git 11:08:04 yeah never mind source 11:08:29 so would be worth looking into 11:08:52 as kind of a dynamic download on demand 11:09:26 we have one proposed release blocker, https://bugs.merproject.org/show_bug.cgi?id=289 11:09:51 i think i'll not grant that for this release but it should be in next cycle s 11:09:54 tart 11:11:38 lbt, please what you just said in #mer in here please 11:12:04 d'oh 11:12:15 289 is a release blocker since a release shouldn't happen if it's half-accepted 11:12:35 but it's not merged yet so we can ignore it for this release 11:12:58 would be good if you could grant or decline the release_blocker status 11:13:13 for this one, decline 11:13:18 but we will review it for next 11:13:18 mmm 11:13:23 can you leave it pendin 11:13:31 https://bugs.merproject.org/buglist.cgi?query_format=advanced&field0-0-0=flagtypes.name&bug_status=ASSIGNED&bug_status=TRIAGEDUPSTREAM&bug_status=REOPENED&bug_status=RESOLVED&bug_status=RELEASED&bug_status=VERIFIED&bug_status=CLOSED&type0-0-0=substring&value0-0-0=Release_Blocker 11:13:36 that's what was reported 11:13:51 i fixed 349, 377 so maybe that should have been under my assignee 11:16:37 so, what about lbt's status/prjcopy? 11:16:46 OK 11:17:01 # info Mer release tools rewrite - using them to do SDK, Tools and OBS : http://gitweb.merproject.org/gitweb?p=mer/release-tools.git;a=shortlog;h=refs/heads/lbt 11:17:28 so I wanted to be able to do releases of Tools to combine with Core to make SDK 11:17:45 #info Mer release tools rewrite - using them to do SDK, Tools and OBS : http://gitweb.merproject.org/gitweb?p=mer/release-tools.git;a=shortlog;h=refs/heads/lbt 11:18:00 having done that... 11:18:04 #info SDK : now release based, new tools (needs a new release with some updates as mentioned in http://www.mail-archive.com/mer-general@lists.merproject.org/msg00548.html ) 11:18:36 This provides a trivial 'update' script : raises an issue: how should we (and oss-style vendors) handle updates? 11:19:23 :nod: 11:19:42 a proper implementation of this kinda relies on the ability to do builds and branches 11:19:52 yes 11:19:58 #info Mer OBS setup-obs.sh script has been updated to use new OBS release (which isn't done yet) 11:20:05 sonach reported success with it 11:20:24 good 11:20:32 unsurprisingly the old version which points to an uncontrolled Testing OBS repo was broken 11:20:59 we should soon be able to release/publish any OBS project - not sure we're "doing it right" though 11:21:16 i think it's in the right direction for sure 11:21:21 OBS 2.3 has capabilities in this area too 11:21:36 we should investigate and share if possible 11:22:07 on the copyproj 11:22:17 what capabilities 11:22:22 OK 11:22:38 not 100% sure - Adrian mentioned it and it's to do with making releases 11:22:51 read the release notes and marked "must read up" 11:23:09 ok 11:23:15 maybe the patchinfo stuff 11:23:32 copyprj: I spent some time on that but had to break for other stuff (and sanity) ... phaeron has had a go and it works for him but not for me :) 11:23:34 and maintenance projects 11:23:47 yes, I think it could be those 11:24:25 last test I did with copyprj was quite big and it worked. I will move on to copyprj + linked prj 11:24:38 lbt: are you doing a copyprj form mds to obs ? 11:24:54 yes 11:24:59 it doesn't work 11:25:01 and you see that working ? 11:25:09 'from mds'? 11:25:15 I'm trying one using cobs/mds 11:25:40 I mean are you doing a project copy from mds directly to an obs ? 11:26:14 https://gist.github.com/2913004 11:26:26 that's my setup script 11:26:34 it makes projects that build against mds 11:26:35 everytime you paste that gist I get fits :D 11:26:43 why? 11:26:50 that's copypac, not copyprj? 11:27:00 "setup script" 11:27:05 ok 11:27:11 it defines starting position pretty clearly 11:27:11 and what is your copyprj command line? 11:27:15 then I do a copyprj 11:27:51 I use : osc-wrapper.py --debug -t -A https://obsfe.dgreaves.com:444 copyprj -n -b MyMerUx MyMerUx44 11:28:14 ok 11:28:23 -n == now == nodelay 11:28:33 -b == binaries == withbinaries 11:28:59 ok so this means you are not doing copyprj from mds to obs 11:29:01 and we're sure it sends the correct url? 11:29:06 POST https://obsfe.dgreaves.com:444/source/MyMerUx44?comment=osc+copyprj+from+project%3AMyMerUx&oproject=MyMerUx&cmd=copy&nodelay=1&withbinaries=1 11:29:27 ok 11:29:40 the only difference in the POST is I don't add a comment 11:29:48 it then "does stuff" 11:29:54 and I get variable results 11:29:58 osc -A appl api -X POST '/source/Project:copy6?cmd=copy&nodelay=1&oproject=home:Admin&withbinaries=1' 11:30:04 so I do it 3 times to 3 target projects 11:30:27 1 succeeds, 1 rebuilds, 1 says "failed" with no-source uploaded 11:30:38 so I have done that 6 times and got same working results. 11:31:01 the only difference between us is that comment , and that you _build_ directly against MDS 11:31:10 I build against mds through cobs 11:31:15 the difference so far is that phaeron uses a build target of cobs:Mer:fake:Core:i586 11:31:36 I use MerDS:Core:i586 11:31:56 right, MerDS directly is more bound to be reliable i would think 11:32:02 and a typical setup 11:32:06 yes 11:32:18 but if his works I should reproduce to verify 11:32:22 then we can isolate 11:33:09 I don't agree. if merds breaks because it is not implementing some corner case of obs-obs link then it is merds's faulty 11:33:46 anyway I will setup mds here and test that scenario 11:33:58 what don't you agree with? I'm going to try your approach 11:34:09 I don't agree with what Stskeeps said 11:34:34 mmm 11:34:46 i just meant that MDS linked directly is the 'typical' mer setup 11:35:03 yeah ... local LAN = more reliable, not less buggy :) 11:35:11 MDS could have bugs 11:35:12 and MDS-OBS-OBS might be prone to failures, or sometimes slowness reducing chance of races 11:35:15 but yes 11:35:17 MDS might have bugs 11:35:26 ok we are all saying same thing 11:35:54 yes 11:35:54 :P 11:36:30 sounds like a plan forward then 11:36:56 yup 11:37:04 i vote for postponing release until monday in order to get proper testing done 11:37:08 with a prerelease tomorrow 11:37:26 OK 11:37:29 k 11:37:59 alright, anything else to discuss? 11:38:05 no 11:38:58 "releases" 11:38:58 thank you all for coming then 11:39:05 .. mm? 11:39:38 well, as I said, there's an issue about releases 11:39:43 it affects Nemo too 11:39:58 getting info into a release about what release it is is hard 11:40:31 asking a Mer device what release it has is not defined 11:40:46 update vs upgrade 11:41:01 that is correct 11:41:20 kinds of "release" ... static and updated 11:41:35 Mer core releases are static so people can rely on them 11:41:45 Nemo ones are dynamic so customers see updates 11:42:06 err, kinds of release repo 11:42:26 OS updates are a hard problem :) 11:42:36 hence the need to discuss... 11:42:37 i think it's worth starting a discussion thread on 11:42:41 on mailing list 11:42:56 since I've been looking at createrelease scripts with an eye on what vendors want 11:43:00 will you start it? 11:43:04 OK 11:43:25 any thoughts here to prime it - or are we busy? 11:43:38 preparing birthday treats perhaps? 11:43:46 well, some research into existing SSU methods at least 11:43:50 and old plans for meego 11:44:00 i need to go have some food before i collapse on the keyboard:P 11:44:02 also suse approaches 11:44:12 since OBS may well assist 11:44:17 OK 11:44:29 thanks all for coming 11:44:38 #action lbt to start release update discussions 11:44:39 #endmeeting