[Releasing] Meeting minutes: Qt 5.1 release team meeting 29.04.2013

[Salovaara Akseli]

Hi,

Meeting minutes from Qt 5.1 release team meeting 29.04.2013:

-          First installers available last Friday and new installers should be available via http://download.qt-project.org/snapshots/qt/5.1/5.1.0-beta1/backups/ daily

o    Win8/MSVC2012 64bit OpenGL package available as new installer

-          For now configure.exe will be available via .7z and .zip source archives like in previous releases. Thiago will study if we can change configure.exe to follow the same rules as the include headers from syncqt.

-          It seems that some cleanup is needed in Bugreports

o    More than 40 error corrections during week 17 according to code review changes, only 8 errors corrected according to Bugreports statistic

-          MinGW 4.8.0 integration: It was integrated in the CI system but that resulted in weird errors and was reverted. Studies ongoing to resolve the issue.

-          Package ICU 5.1 instead of ICU 4.9

o    Current packages still use ICU 4.9. Pre-compiled ICU 5.1 + runtime waiting for installing them on the build machines & change the local configuration

-          Merge from stable to release will happen once stable is good enough (meaning sometime after beta(s) is released and there isn't any major findings)

-          Next meeting  06.05.2013 16:00 CET

IRC log below

Br,
Akseli


(17:00:29) akseli: iieklund: kkoehne: ramotyka: sahumada: thiago: fkleint: ZapB: tronical: janiheikkinen: ping
(17:00:35) sahumada|home: akseli: pong
(17:00:38) janiheikkinen: pong
(17:01:05) fkleint: akseli: pong
(17:01:19) thiago: akseli: pong
(17:01:29) kkoehne: akseli: pong
(17:01:46) ZapB: akseli: pong
(17:01:54) akseli: Time to start Qt 5.1 release team meeting
(17:01:54) akseli: On agenda today:
(17:01:59) akseli: - 5.1 installer \ packaging status
(17:02:00) akseli: - 5.1 error metrics
(17:02:00) akseli: - MinGW version change from 4.7.2 to 4.8.0
(17:02:00) akseli: - Package ICU 5.1 instead of ICU 4.9
(17:02:00) akseli: - Merging stable->release (timing)
(17:02:06) akseli: Other topics we should have in agenda?
(17:02:56) thiago: not from me
(17:03:10) akseli: feel free to raise topics during meeting... if i start from the installers
(17:03:33) akseli: first installers available last friday and new installers should be available via http://download.qt-project.org/snapshots/qt/5.1/5.1.0-beta1/backups/ every day
(17:03:41) akseli: experimental Android offline installers available (Linux and Windows host installers, armv7 and x86 targets included)
(17:03:52) akseli: Win8/MSVC2012 64bit OpenGL package available as new installer
(17:04:23) akseli: deceided on mailing list that no changes into configure.exe (.7z and .zip source archives) will be done compared to previous releases
(17:04:42) akseli: currently testing syncqt if and how it affects builds
(17:05:33) akseli: any questions or comments regarding to installers available?
(17:06:47) thiago: configure.exe in big .zip => agreed
(17:07:04) thiago: big .zip and big .tar.gz differing => not agreed
(17:07:27) thiago: put the file in all packages or none. Don't differentiate based on .7z/.zip vs .tar.gz/.tar.xz
(17:07:47) thiago: is configure.exe in the qtbase*.zip ?
(17:08:01) kkoehne: I agree with Thiago. I think the current association with .zip->windows, tar.gz ->unix etc is bogus.
(17:08:47) sahumada|home: thiago: yes ... qtbase*zip and qtbase*7z
(17:09:30) thiago: I don't like that
(17:09:41) thiago: the discussion on the ML is that qtbase is pristing
(17:09:43) thiago: pristine
(17:09:49) sahumada|home: I sent an email to releasing at qt-project.org<mailto:releasing at qt-project.org> as we agreed on this meeting
(17:09:52) thiago: the big packages aren't verifiable anyway, so we may modify as needed
(17:12:15) akseli: so configure.exe should be in big zip & tar.gz packages and removed from modulare qtbase.* ?
(17:13:38) sahumada|home: no .. configure.exe will stay as it is today .. according to what is was (not)decided in the ML
(17:14:16) thiago: I'd say configure.exe should follow the same rules as the include headers from syncqt
(17:14:39) thiago: the proposal was that syncqt would be run on the big package, but not on qtbase and the other individual files
(17:17:09) akseli: i think we had problems having configure.exe on tar.gz since that package is generated on linux but can't remember more details
(17:17:29) akseli: unfortunately iieklund is away at the moment and cant check details
(17:18:43) thiago: I won't have time to look into mksrc in the next couple of weeks
(17:18:49) thiago: but that shouldn't be hard to do
(17:19:02) akseli: i have feeling that we might run into unnecessary delays with configure.exe & tar.gz if trying to bundle it now
(17:19:06) thiago: we had people downloading the .tar.gz for Windows because it was smaller. We need it there.
(17:19:08) sahumada|home: if we remove configure.exe from qtbase (zip and 7z ) .. then somebody needs to make qtbase to bootstrap configure.exe
(17:19:32) thiago: sahumada|home: it does that today in the repo.
(17:19:40) thiago: which means it has bad code inside that tries to bypass that.
(17:19:49) kkoehne: thiago: only if .gitignore is there, AFAIK
(17:19:59) sahumada|home: thiago: only for git builds .. I think .. not from the qtbase source package
(17:20:05) thiago: thank you for confirming "bad code" :-)
(17:20:37) thiago: what happens if you remove the check?
(17:21:10) thiago: hmm.. no .o are shipped, so it will build again
(17:21:46) thiago: we'd need to change the Makefile to avoid doing anything if the .exe is newer than the sources
(17:22:09) thiago: I'll take that offline
(17:24:07) akseli: as long as we are able to produce builds & installers to continue qt5.1 i don't feel confortable to make changes. this might not be the most important feature of qt5.1 after all
(17:24:30) akseli: i'm not saying that nothing will be changed but now i would like to continue as it is ..
(17:25:22) thiago: ok, so let's put it this way: if I can make it happen, we'll consider the change
(17:25:34) thiago: otherwise, configure.exe stays in qtbase
(17:25:59) akseli: sounds like a plan
(17:26:01) thiago: note that the change might be necessary *anyway* in order to enable the syncqt & headers that was proposed on the ML
(17:26:54) akseli: ok, lets see how that will go
(17:27:03) akseli: next item: 5.1 error metrics, janiheikkinen^
(17:27:11) janiheikkinen: Hi
(17:27:16) janiheikkinen: Snapshot from Jira earlier today:
(17:27:31) janiheikkinen: Open 5.1 cases: 154 ( 28 more than last week)
(17:27:51) janiheikkinen: P0: 0
(17:28:06) janiheikkinen: P1: 20 (4 more than last week)
(17:28:21) janiheikkinen: Not evaluated: 53 (9 more than last week)
(17:28:32) ZapB: do you have a link to the meta bug please?
(17:28:54) akseli: https://bugreports.qt-project.org/browse/QTBUG-30688
(17:28:59) ZapB: thx
(17:29:28) janiheikkinen: And then interesting detail ;)
(17:29:49) janiheikkinen: More than 40 error corrections during week 17 according to gerrit changes, only 8 errors corrected according to Jira statistic
(17:30:36) janiheikkinen: So I think not all Jira cases are up to date
(17:31:04) kkoehne: janiheikkinen: DO the 40 ones contain Task-number: entries?
(17:31:22) janiheikkinen: Yes
(17:32:56) kkoehne: Alright, then we indeed have to do some clean up. Or people are waiting to see their changes in a qt.git update (which I doubt)
(17:33:41) akseli: there ain
(17:34:09) akseli: *there ain't so many bugs linked against 5.1 beta metabug. anyone aware of additional items which belongs there?
(17:34:41) fkleint: akseli: The docu meta bug..
(17:35:00) sahumada|home: the syncqt change should be there .. if we are aiming that for 5.1
(17:35:10) fkleint: akseli: https://bugreports.qt-project.org/browse/QTBUG-30173
(17:36:45) fkleint: akseli: added
(17:36:49) akseli: aware of document meta bug and can see that not all of subtasks are up to date
(17:37:03) akseli: fkleint: thanks
(17:37:44) fkleint: akseli: I created yet another subtask to the doc bug https://bugreports.qt-project.org/browse/QTBUG-30912
(17:38:05) fkleint: akseli: since the examples are a bit messed up
(17:38:09) sahumada|home: I wonder if https://bugreports.qt-project.org/browse/QTBUG-30923 is a blocker .. Charts is not an opensource project .. is it ?
(17:39:01) fkleint: OMG, a graphicsview fix by AAHansen caused this?
(17:39:03) akseli: Not a blocker, Charts is not part of open source but have to be checked if root cause is on Qt itself
(17:39:13) fkleint: we will never get that fixed
(17:39:38) thiago: I have a crash-on-start with subsurface too
(17:40:09) thiago: anyway, the above bug is invalid unless a testcase is posted
(17:40:35) thiago: or "Not enough info"
(17:40:43) thiago: you can't expect bibr to debug Charts
(17:42:59) akseli: I will ask more info besided adding comment already to 30923
(17:44:24) akseli: already a bit worried (based on "we will never get that fixed") but please link against metabug if find additional problems
(17:45:05) akseli: maybe then moving onwards..
(17:45:11) akseli: MinGW version change from 4.7.2 to 4.8.0, kkoehne^
(17:45:32) kkoehne: akseli: Yeah, we tried to get 4.8.0 in the CI system, but that resulted in weird errors, and was reverted.
(17:45:57) kkoehne: 'weird errors" being that gcc claimed that it couldn't find system header files that should be there, int eh default locations
(17:46:18) thiago: as in "headers that GCC installed" ?
(17:46:23) kkoehne: thiago: yes
(17:46:43) kkoehne: I'm right now trying to replicate the exact setup here on my machine, locally, since I don't have access to the CI build machines
(17:47:24) kkoehne: However, without any luck so far :(
(17:47:37) thiago: probably a bad install
(17:48:25) kkoehne: Yeah, although sfalt already verified that the header in place _is_ there. Might be also that we're hitting somehow a limit on the number of include dirs or s.th. I've seen similar issues when trying to do a shadow build of qt where the paths just got too long.
(17:48:31) kkoehne: But that wasn't a 4.8.0 regression :(
(17:48:54) kkoehne: I'll bug sfalt tomorrow morning again to have a second look on the machine itself.
(17:49:38) kkoehne: We could also just try to deploy to the build machines, and see whether it'
(17:49:43) kkoehne: s an issue there, too
(17:50:07) kkoehne: will ask iieklund tomorrow whether we can try that.
(17:50:40) kkoehne: akseli: That's it.
(17:50:42) akseli: ok, thank you for the update
(17:50:51) akseli: next item: Package ICU 5.1 instead of ICU 4.9, kkoehne^
(17:51:45) kkoehne: Current packages still use ICU 4.9. I just uploaded runtime + -devel packages containing pre-compiled icu 5.1 for all platforms to download.qt-project.org
(17:52:03) kkoehne: http://download.qt-project.org/development_releases/prebuilt/icu/ specifically
(17:52:04) thiago: any problems?
(17:52:29) kkoehne: No, just that someone has to put them on the build machines, change the local configuration + we then have to update the installer scripts to use the new versions, too.
(17:52:50) kkoehne: I guess iieklund is most qualified to do this.
(17:53:22) kkoehne: akseli: That's it for ICU, too :)
(17:53:29) thiago: ICU is a subject for QtCS too
(17:54:30) kkoehne: Yes, please :)
(17:54:45) fkleint: yes, it would be fantastic to get rid of it..not sure if it can be fully replaced by WinAPI?
(17:55:09) thiago: let's not get ourselves dragged into this now
(17:55:12) thiago: :-)
(17:55:20) fkleint: oki ;-)
(17:55:30) akseli: thanks for the update and last item from agenda: Merging stable->release, sahumada^
(17:55:42) thiago: post-beta
(17:55:48) thiago: beta is released from stable
(17:56:47) sahumada|home: thiago: my question is .. what if we do two betas ? how do we let people know when we are going to do the merge so they dont stage in stable
(17:57:09) thiago: are the two questions related?
(17:57:21) thiago: if so, I don't see how
(17:57:26) thiago: if we do two betas, we release twice from stable
(17:57:40) sahumada|home: thiago: I'd like to send an email today letting people know when we are aiming to merge stable->release
(17:57:44) thiago: and even if we do just one beta, we need to let people know about the builds
(17:57:53) thiago: sure, do that
(17:58:23) thiago: it should be after the beta cycle is complete and we start the release cycle
(17:58:39) thiago: that is, once stable reaches quality to be merged to releases
(17:58:40) sahumada|home: ok .. will do .. so I'll say we will merge stable->release after the first Beta is out
(17:59:01) thiago: well, technically it should be a few weeks after, when we know whether we have release quality
(17:59:04) ZapB: sahumada|home: after last beta you mean?
(17:59:08) thiago: when we know there won't be a new beta
(17:59:16) sahumada|home: ZapB: yes .. that even
(17:59:24) kkoehne: I don't think the world will end if , for some reason, we release an (unplanned) beta2 from release instead of stable.
(17:59:38) sahumada|home: thiago: thanks .. that's more clear for me now
(17:59:43) kkoehne: But we don't plan for a beta 2, do we? :)
(17:59:59) thiago: no, we don't plan for one
(18:00:08) thiago: we'll know when we are beta1+2weeks
(18:01:03) akseli: i think we should try to produce beta release during next week if possible. just to see where we are.
(18:02:38) akseli: mingw and icu items exists but possibly solved during this week already - i hope
(18:03:07) kkoehne: akseli: Yeah, I'd really love to get the updates in the beta. The idea is that people test the updates with the beta, after all.
(18:06:10) akseli: lets see where we are at the end of the week.. if mingw and icu solved very quickly maybe additional release team meeting will be held during this week even?
(18:06:48) akseli: othewise i would like to propose next meeting on next Monday 16:00 CET
(18:07:26) akseli: or does anyone have additional items for today?
(18:07:35) ZapB: nothing from me
(18:07:38) kkoehne: no.
(18:07:42) janiheikkinen: no
(18:07:42) sahumada|home: no
(18:07:50) fkleint: no
(18:08:23) akseli: next meeting then on Monday 16:00 CET at latest
(18:08:30) akseli: thank you all for the meeting
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/releasing/attachments/20130430/fe5f8b6f/attachment.html>