[Releasing] Meeting minutes: Qt 5.2 release team meeting 16.09.2013

[Heikkinen Jani]

Meeting minutes from Qt 5.2.0 release team meeting 16.09.2013:
- Problems with v4 on QNX (most of the unit tests are failing, QtGraphicaleffects won't work etc)
- As planned Qt 5.2 feature freeze on Friday 20th, merge from dev to stable on Monday (23rd) morning. Target: Alpha release 26th September
  - After merge from dev to stable starts, staging is disabled for dev (except admins)
  - In case of CI system failure on new feature merging to dev there is a possibility grace period after feature freeze. What will be taken in will be decided case by case.
- MinGW version will be 4.8.0 for Qt 5.2. Update to newer version will happen after 'official' release of Mingw-w64 to upstream (considered for Qt 5.3).
- Next meeting: 23.09.2013 16:00 CET

IRC log below:

(5:00:59 PM) jaheikki3: akseli: iieklund: kkoehne_: sahumada: thiago: fkleint: ZapB_: tronical: ramotyka_: wolfgang-b: vladimirM: aholza: peter-h: ping
(5:01:08 PM) sahumada: jaheikki3: pong
(5:01:09 PM) peter-h: jaheikki3: pong
(5:01:12 PM) akseli1: jaheikki3: pong
(5:01:13 PM) ZapB: jaheikki3: pong
(5:02:52 PM) jaheikki3: Time to start Qt5.2 release team meeting
(5:03:03 PM) jaheikki3: On agenda today:
(5:03:05 PM) wolfgang-b: jaheikki3, pong
(5:03:22 PM) jaheikki3: Feature freeze & Alpha release
(5:03:22 PM) jaheikki3: mingw version
(5:03:22 PM) jaheikki3: QTBUG-32582: Debug symbols missing for Qt 5.1.0 SDK under Linux
(5:03:22 PM) jaheikki3: Next meeting
(5:03:22 PM) jaheikki3: Any additional items to agenda?
(5:04:10 PM) wolfgang-b: We are still having problems with v4 on QNX, not sure how important that is for the general Qt 5.2 release
(5:05:06 PM) jaheikki3: wolfgang-b: What kind of problems?
(5:05:30 PM) wolfgang-b: Most unit tests are failing, QtGraphicaleffects won't work etc.
(5:06:04 PM) thiago [~thiago at kde/thiago] entered the room.
(5:06:05 PM) jaheikki3: :(
(5:06:25 PM) kkoehne: wolfgang-b: I guess somebody is investigating?
(5:06:31 PM) jaheikki3: Is there some bug report about that already?
(5:07:38 PM) wolfgang-b: yes, Simon Hausmann, Fabian Bumberger and Petr Nejedly are looking into it.
(5:08:22 PM) wolfgang-b: a qt-project bug report is not yet existing, Fabian will create one
(5:09:16 PM) jaheikki3: OK, great. hoping they can find solution soon. In my opinion that shouldn't affect to feature freeze /alpha but should be fixed before beta
(5:09:39 PM) ZapB: agreed
(5:09:57 PM) vladimirM: I was talking about this with Lars and Simon via email. They are aware of the issue.
(5:10:03 PM) wolfgang-b: I agree. Just wanted to mention this earl, as we are currently working on it for about 4 weeks
(5:10:29 PM) jaheikki3: wolfgang-b: thanks, it is good to know
(5:10:46 PM) vladimirM: It looks like that V4 is not fully tested on ARM yet. Simon suggested to take more time for this after the feature freeze
(5:11:01 PM) lars: vladimirM: wolfgang-b: we have some stuff we need to finish up, but will then look into this directly after
(5:11:18 PM) thiago: Alpha = validation of the technology, not the implementation
(5:11:24 PM) thiago: it's fine to have bugs
(5:11:24 PM) vladimirM: the problem for us that we cannot work with the 5.2 really as long as it does not work well
(5:12:11 PM) lars: vladimirM: see above. We need to finish up some of the refactorings for 5.2, but will look at this directly after.
(5:12:23 PM) lars: vladimirM: for now, you could try using the interpreter. that one should work fine
(5:12:56 PM) lars: vladimirM: just use QV4_FORCE_INTERPRETER=1
(5:13:00 PM) lars: as env variable
(5:13:22 PM) vladimirM: Oh! Ok... Thanks!
(5:13:24 PM) wolfgang-b: lars, thanks. We will try this for the moment
(5:13:29 PM) FabianBu [~quassel at 195.3.173.181] entered the room.
(5:13:42 PM) jaheikki3: vladimirM: As agreed earlier Alpha is just src package delivery
(5:14:08 PM) jaheikki3: To start validating new features etc. like thiago said
(5:15:05 PM) vladimirM: understand. we make our packages on our own so far. thanks for the clarification! :-)
(5:16:38 PM) jaheikki3: As known, feature freeze is this friday and merge from dev to stable will be done on Monday morning
(5:16:47 PM) jaheikki3: Then qt5.git update so we could have src packages available earliest Tuesday 
(5:16:56 PM) jaheikki3: If we are lucky
(5:17:14 PM) jaheikki3: So alpha release can be done earlies on wed
(5:17:17 PM) thiago: btw, we need to assess any features that tried to merge into dev but failed due to CI
(5:17:27 PM) thiago: we should offer a grace period for those (and those only!)
(5:18:19 PM) jaheikki3: thiago: actually there is 2 days for that ;)
(5:18:36 PM) jaheikki3: If Sergio starts merge on monday morning
(5:18:47 PM) jaheikki3: feature freeze is on Friday
(5:19:11 PM) thiago: I know
(5:19:21 PM) thiago: but it might not have finished by Monday
(5:19:48 PM) thiago: I mean only if it's not the submission's fault that the CI failed
(5:20:06 PM) thiago: if it tripped a test that hadn't been found yet, then it's the feature's own fault and it gets bumped to 5.3
(5:20:40 PM) jaheikki3: I understand and agree
(5:20:44 PM) kkoehne: thiago: Can't be such border-line commits then just re-submitted to stable?
(5:21:14 PM) ***ZapB best finish QOpenGLTexture quickly then
(5:21:18 PM) thiago: kkoehne: if they're already in the dev queue, leave them there
(5:21:33 PM) thiago: on Monday, after we start the merge, we should disable staging for dev
(5:21:43 PM) thiago: except for admins, who can do it for those features
(5:21:55 PM) thiago: that will also get people to realise they need to re-target
(5:22:20 PM) kkoehne: thiago: Yes, we have had that in the past already (2 weeks, if I remember right).
(5:22:33 PM) thiago: 2 weeks is too long. I'm thinking 3 days at most.
(5:23:39 PM) jaheikki3: agree
(5:24:36 PM) jaheikki3: sahumada has done already merge from stable to dev so it should help merging from dev to stable
(5:25:13 PM) sahumada: jaheikki3: I am planning to do the last one on Friday .. at least for qtbase and qtdeclarative
(5:25:30 PM) sahumada: there is a new conflict in qtbase already
(5:27:10 PM) jaheikki3: Plan was to do alpha 26th Sep and I think it is still OK
(5:27:53 PM) jaheikki3: But of course if there will be problems with Merge or CI then it might be tight
(5:28:29 PM) thiago: we'll look into it next Monday, I guess
(5:28:53 PM) jaheikki3: yes, I agree. 
(5:29:00 PM) jaheikki3: OK, lets continue with mingw version
(5:29:10 PM) jaheikki3: kakoehne?
(5:29:40 PM) jaheikki3: Any comments about that?
(5:29:49 PM) kkoehne: jaheikki3: mingw-builds gcc 4.8.0 (which we used for 5.1) and newer gcc versions / mingw-builds are not binary compatible.
(5:30:04 PM) kkoehne: jaheikki3: I think we should just stay with 2.8.0 for 5.2
(5:30:41 PM) jaheikki3: kakoehne: agree. I think we cannot break binary compability
(5:30:47 PM) thiago: kkoehne: any chance we can implement the solution I outlined?
(5:30:54 PM) kkoehne: jaheikki3: Good thing is that they're trying to get an 'official' release of Mingw-w64 upstream, so with a bit of luck next time we won't have all the hickups.
(5:30:55 PM) thiago: bring those symbols back into the libs where they were?
(5:31:24 PM) kkoehne: thiago: mmmh....that'll require probably our own patched mingw-builds release.
(5:31:37 PM) kkoehne: thiago: Honestly speaking I'm not sure how much time I want to sink in there :)
(5:32:19 PM) kkoehne: thiago: But it's probably doable.
(5:32:50 PM) thiago: fair enough
(5:32:52 PM) kkoehne: thiago: Anyhow, as I said previously I'm not aware of any bigger problems with the current toolchain.
(5:33:04 PM) thiago: so we ship with the old for now?
(5:33:08 PM) jaheikki3: I think it is better to stay with 4.8.0
(5:34:11 PM) kkoehne: thiago: The symbol is anyway only one problem ... The other is that due to changes in the Mingw-w64 headers, 5.1.1 already doesn't compile any more.
(5:35:26 PM) thiago: ok, so we stay with the old one for now
(5:35:37 PM) thiago: let's update when mingw releases an official, for 5.3
(5:35:44 PM) kkoehne: thiago: Yes.
(5:35:46 PM) jaheikki3: agree
(5:35:50 PM) thiago: leave a note in the changelog that this will happen
(5:36:10 PM) kkoehne: thiago: a note in the changelog that nothing chnaged? :)
(5:36:16 PM) jaheikki3: ;9
(5:36:31 PM) jaheikki3: Ok, then that debug symbol issue: QTBUG-32582: Debug symbols missing for Qt 5.1.0 SDK under Linux
(5:37:01 PM) jaheikki3: We tried to configure & build with -force-debug-info
(5:37:30 PM) jaheikki3: Produced binaries was almost 4 times bigger than current installation package
(5:37:38 PM) kkoehne: jaheikki3: ?
(5:37:42 PM) kkoehne: jaheikki3: Did you pass -release?
(5:37:59 PM) kkoehne: jaheikki3: Ah, sorry, should be -release -separate-debug-info
(5:38:27 PM) jaheikki3: kkoehne: no, just  -force-debug-info
(5:38:50 PM) kkoehne: jaheikki3: That'll probably add debug symbols to the library itself. Use -separate-debug-info
(5:39:23 PM) thiago: -release -force-debug-info -separate-debug-info
(5:39:27 PM) thiago: all three are necessary
(5:39:36 PM) kkoehne: thiago: Really? SOunds like a bug ...
(5:39:39 PM) thiago: note that no one builds with -force-debug-info, it might not work
(5:39:52 PM) thiago: they are three separate actions
(5:39:56 PM) thiago: 1) build in release mode
(5:40:02 PM) thiago: 2) enable -g with release
(5:40:07 PM) ***kkoehne builds regularly with -release -force-debug-info
(5:40:09 PM) thiago: 3) strip out debug symbols into .debug files
(5:40:11 PM) thiago: kkoehne: ah, ok
(5:41:16 PM) jaheikki3: OK, i'll try that later. I'll add comments to the error report.
(5:42:04 PM) jaheikki3: Is there something else than next meeting time & date?
(5:43:32 PM) jaheikki3: OK, then next meeting: Next monday in that same time?
(5:43:49 PM) wolfgang-b left the room ("Leaving").
(5:44:06 PM) wolfgang-b [~wolfgangb at 195.3.173.181] entered the room.
(5:45:11 PM) jaheikki3: Nobody disagree so next meeting will be 23.09.2013 16:00 CET
(5:45:22 PM) ZapB: great - thank you
(5:45:32 PM) jaheikki3: Let's end this meeting now. Thanks for everyone!
(5:45:36 PM) jaheikki3: Bye
(5:45:46 PM) peter-h: thanks, bye
(5:46:18 PM) sahumada: bye
(5:46:29 PM) wolfgang-b: bye