[Releasing] Meeting minutes: Qt 5.2 release team meeting 02.09.2013

Heikkinen Jani Jani.Heikkinen at digia.com
Mon Sep 2 17:20:51 CEST 2013

Meeting minutes from Qt 5.2.0 release team meeting 02.09.2013:
- Qt 5.1.2 will not be released unless something really critical is found
               -This is to secure Qt5.2.0 schedule
- Qt 5.2.0 planned schedule
    - Feature freeze on 20.9.2013
    - Merge from dev to stable will happen immediately after feature freeze
    - Alpha release (source packages) will be done almost immediately after feature freeze as well, at week 39
    - Beta release (with binary installers etc) will be done 3 weeks after alpha, at week 42
    - Release candidate will be done 4 weeks after beta, at week 46
    - Final release will be done 2 weeks after RC, at week 48, which is last week of November
    - Release plan will be available on Qt Project's wiki
- Qt 5.2 Tools & versions
    - Target is to upgrade
               - MinGW version to 4.8.1
                              - both MinGW 4.8.0 & 4.8.1 needs to be shipped. More discussion coming in ML soon
- OpenSSL version to 1.0.1 (or newer)
               - Qt Creator version to 3.0
                              - Beta, RC and Final aligned with Qt 5.2 releases
- Next meeting: 16.9.2013 16:00 CET

IRC log below:
(5:00:18 PM) jaheikki3: akseli: iieklund: kkoehne_: sahumada: thiago: fkleint: ZapB_: tronical: ramotyka_: wolfgang-b: vladimirM: aholza: peter-h: ping
(5:00:24 PM) iikka_: jaheikki3: pong
(5:00:32 PM) akseli: jaheikki3: pong
(5:00:44 PM) fkleint: jaheikki3: pong
(5:00:56 PM) wolfgang-b: jaheikki3, pong
(5:01:37 PM) peter-h: jaheikki3: pong
(5:01:52 PM) sahumada: jaheikki3: pong
(5:02:06 PM) jaheikki3: Time to start first Qt5.2 release team meeting, welcome all!
(5:02:22 PM) jaheikki3: On agenda today:
(5:02:33 PM) jaheikki3: Qt5.2 planned schedule
(5:02:41 PM) jaheikki3: Qt5.2 tools & versions
(5:02:49 PM) jaheikki3: Next meeting
(5:02:56 PM) jaheikki3: Any additional items to agenda?
(5:04:18 PM) fkleint: What is the plan for 5.1.2?
(5:05:12 PM) jaheikki3: fkleint: Good, lets take it to the agenda as well
(5:05:49 PM) jaheikki3: feel free to bring topics during meeting .. let's start from Qt5.2 planned schedule
(5:06:06 PM) jaheikki3: As Lars already informed target is to get 5.2 out at the end of November
(5:06:25 PM) jaheikki3: He informed also that feature freeze is 20.9.2013
(5:06:38 PM) jaheikki3: This means we have 10 weeks to finalize Qt5.2
(5:06:52 PM) jaheikki3: from feature freeze to final I mean
(5:07:08 PM) fkleint: quite ambitious..when will the merge dev->stable happen?
(5:07:38 PM) jaheikki3: fkleint. Immediately after feature freeze
(5:07:55 PM) fkleint: that means we release 5.1.2. before?
(5:08:22 PM) kkoehne_: fkleint: Do we plan for a 5.1.2? I thought not.
(5:08:35 PM) jaheikki3: fkleint: Well, we wouldn't like to do it...
(5:09:10 PM) jaheikki3: To secure 5.2 schedule it would be really good if we could avoid releasing qt5.1.2
(5:09:24 PM) fkleint: We have a 5.1.2 fix version in Jira?
(5:09:36 PM) fkleint: and many bugs with fix version 5.1.2?
(5:09:40 PM) sahumada: we also had 5.0.3
(5:09:43 PM) fkleint: hehe
(5:09:44 PM) kkoehne_: fkleint: well, that's just in case, I guess.
(5:09:46 PM) jaheikki3: fkleint: Yes, we have
(5:10:19 PM) jaheikki3: because it hasn't been decided if we will release 5.1.2 or not
(5:10:39 PM) jaheikki3: But as I said it would be good if we don't need to do it
(5:11:32 PM) fkleint: fex we have https://bugreports.qt-project.org/browse/QTBUG-33062
(5:11:33 PM) jaheikki3: Does someone know some reason why we have to release qt5.1.2?
(5:11:40 PM) fkleint: which I would like to see in a 5.1.2
(5:11:45 PM) fkleint: a stupid crash
(5:13:19 PM) jaheikki3: it is just P2 (at the moment). For me it doesn't seem critical enough for forcing us to do 5.1.2 release
(5:13:57 PM) kkoehne_: I agree. There'll always be things already fixed in stable, but we really want to focus on keeping the 5.2 schedule.
(5:15:06 PM) fkleint: hm..the thing is, V4 might cause quite some surprises and it would be good to have a rock-solid 5.1.2
(5:15:42 PM) jaheikki3: fkleint: Can you explain a bit more?
(5:16:01 PM) fkleint: ?
(5:16:08 PM) kkoehne_: fkleint: Then let's rather do a 5.2.1 soon afterwards, to fix any remaining 'suprises' :)
(5:16:18 PM) fkleint: hm
(5:16:25 PM) fkleint: I won't argue..
(5:16:53 PM) fkleint: but like the 5.1 renderloop performance regression, V4 might have quite an imp[act on Qt Quick
(5:18:33 PM) kkoehne_: fkleint: True. But doing a 5.1.2 would roughly require three weeks or so where we build 5.1.2 package + test etc. I'm just afraid that the 3 weeks will be missing for 5.2, which we really, really, don't want to do.
(5:18:49 PM) jaheikki3: I would like to make a decision that we won't do 5.1.2 if something really critical found. And if that kind of "blocker" found then we have to make it and that's it. But making it for sure... I don't like the idea.
(5:19:15 PM) kkoehne_: jaheikki3: You mean we won't do 5.1.2 unless somethign really critical is found? :-D
(5:19:28 PM) kkoehne_: jaheikki3: +1
(5:19:31 PM) fkleint: Hm..ok
(5:19:50 PM) jaheikki3: kakoehne: That was my proposal ;)
(5:20:13 PM) sahumada: if something really critical is found .. we make 5.1.2 out of the release branch + the fix ?
(5:21:04 PM) fkleint: or merge stable->release preemptively now?
(5:21:11 PM) fkleint: now that 5.1.1 is done?
(5:21:23 PM) kkoehne_: fkleint: What would that gain us?
(5:21:38 PM) fkleint: We would have it ready, then
(5:22:05 PM) sahumada: mmm
(5:22:21 PM) kkoehne_: I'd say let's concentrate on 5.2 now, and think about how exactly 5.1.2 is done when we need it. It doesn't gain us much to prepare right now ...
(5:22:26 PM) sahumada: fkleint: merging stable->release doesnt go through the CI .. so it doesnt take that much time to do it
(5:22:51 PM) sahumada: kkoehne_: +1
(5:23:09 PM) jaheikki3: kakoehne_: +1. If we need to do 5.1.2 we can then make the decision what is the content
(5:23:14 PM) fkleint: Hm, ok, lets discuss 5.2 then
(5:24:13 PM) jaheikki3: OK, We have made a plan how to proceed from feature freeze to final:
(5:24:45 PM) jaheikki3: As said earlier, merge from dev to stable will happen immediately after feature freeze
(5:24:58 PM) jaheikki3: Alpha release will be done almost immediately after feature freeze as well, at week 39
(5:25:17 PM) jaheikki3: Just source packages, nothing else. Like in 5.1
(5:25:46 PM) jaheikki3: Beta release will be done 3 weeks after alpha, at week 42
(5:26:06 PM) jaheikki3: That will contain binary installers etc, like 5.1
(5:26:16 PM) jaheikki3: Release candidate will be done 4 weeks after beta, at week 46
(5:26:30 PM) jaheikki3: And final release will be done 2 weeks after RC, at week 48, which is last week of November
(5:27:27 PM) jaheikki3: Schedule is quite tight but we trust it is doable
(5:28:01 PM) jaheikki3: But we need to be quite tight with feature freeze etc to secure that schedule
(5:28:33 PM) jaheikki3: Any comments?
(5:29:31 PM) kkoehne_: jaheikki3: Maybe put the plan on the wiki?
(5:30:00 PM) thiago [~thiago at kde/thiago] entered the room.
(5:30:05 PM) anshaw left the room (quit: Quit: No Ping reply in 180 seconds.).
(5:30:24 PM) jaheikki3: kakoehne_: that is under work.
(5:30:26 PM) anshaw [~quassel at] entered the room.
(5:30:48 PM) jaheikki3: I want to get your approval before publishing it
(5:30:58 PM) jaheikki3: Even Lars has already done it ;)
(5:32:17 PM) jaheikki3: Ok, seems that this is ok for everyone...
(5:32:21 PM) kkoehne_: jaheikki3: Good :) I think the plan is okay. It's ambitious, but I think we can make it.
(5:33:07 PM) jaheikki3: OK, let's then discuss about tools & versions for qt5.2
(5:33:29 PM) iikka_: tool chain versions used, I'd like to have decision about them as early as possible
(5:33:47 PM) jaheikki3: iikka_: Agree
(5:33:48 PM) iikka_: meaning we don't change them anymore late in the process
(5:34:17 PM) jaheikki3: And as Lars wrote, he will veto build system changes after branching into stable
(5:34:25 PM) fkleint: hehe ;_)
(5:34:32 PM) kkoehne_: jaheikki3: Well, I geuss that was more about qmake changes :)
(5:34:37 PM) iikka_: mingw and openssl versions
(5:34:57 PM) iikka_: for MinGW do we know any updates worth taking now?
(5:35:18 PM) kkoehne_: iikka_: Well, people always want to have the latest & greatest, which translates to 4.8.1 at the moment :)
(5:35:19 PM) peter-h: so we build packages with openssl 1.* , right?
(5:35:36 PM) iikka_: kkoehne_: ok :)
(5:35:49 PM) iikka_: peter-h: yes, openssl was on my list as well
(5:35:54 PM) peter-h: we are using openssl 1.* on CI machines, so would be good to use it for building packages as well...
(5:36:02 PM) iikka_: as far as I know the CI uses 1.0.0 ?
(5:36:03 PM) peter-h: iikka_: ok good...
(5:36:10 PM) peter-h: iikka_: yes, correct
(5:36:27 PM) iikka_: peter-h: yes, we were planning to upgrade it in packaging as well
(5:36:33 PM) iikka_: to match with CI
(5:36:42 PM) peter-h: a problem with mac is they only ship openssl 0.9.*, so we would need to advise users to ship openssl along with apps, like on windows
(5:36:46 PM) thiago: peter-h: when do you plan on having NPN done?
(5:36:50 PM) peter-h: iikka_: ah good
(5:37:08 PM) peter-h: thiago: for my part it is already done, I am waiting for review :)
(5:37:14 PM) thiago: peter-h: hmm... we need to update the OpenSSL search path then
(5:37:20 PM) kkoehne_: iikka_: The draw back is that MinGW-builds gcc 4.8.1 and 4.8.0 appear to be binary incompatible. So we'd have to ship them both, and somehow mark packages differently (since currently the 4.8.0 is just labeled 4.8)
(5:37:31 PM) thiago: peter-h: your code requires building with OpenSSL 1.1 or 1.2.
(5:37:40 PM) thiago: that means upgrading the package builders
(5:37:47 PM) iikka_: kkoehne_: hmm, ok
(5:37:52 PM) thiago: kkoehne_: what happened?
(5:37:54 PM) peter-h: thiago: 1.0.1 or 1.0.2 you mean?
(5:38:08 PM) thiago: peter-h: sorry, yes. Whichever version added the #defines.
(5:38:20 PM) peter-h: thiago: but yes, right, we would need 1.0.1 at least then...
(5:38:52 PM) thiago: peter-h: any chance of #ifndef #define in Qt the same methods?
(5:38:57 PM) kkoehne_: thiago: Some symbol's got removed ... moment ...
(5:39:01 PM) thiago: or is that just a bad idea?
(5:39:21 PM) peter-h: thiago: actually I wonder whether we need NPN at all in 5.2, since it seems ambitious to get SPDY ready for 5.2...
(5:39:35 PM) thiago: peter-h: ok, let's just hold it off then
(5:39:53 PM) thiago: peter-h: wait, don't websockets require them?
(5:39:54 PM) peter-h: thiago: but if we could build with 1.0.1 for packages, the problem would be solved, right?
(5:40:21 PM) peter-h: thiago: apparently not, since the websockets author had the patch ready before the NPN patch...
(5:40:28 PM) thiago: peter-h: ok
(5:40:33 PM) thiago: peter-h: let us continue in the review then
(5:40:45 PM) peter-h: ok...
(5:40:48 PM) thiago: if the CI system can be upgraded, it should be, regardless of NPN
(5:41:00 PM) thiago: but we'll offer a recommendation soon
(5:41:02 PM) peter-h: iikka_: so openssl 1.0.1 would be preferrable for the packages, and CI...
(5:41:11 PM) peter-h: 1.0.1 or later I mean
(5:41:17 PM) iikka_: peter-h: ok
(5:41:51 PM) iikka_: need to sync with CI guys about the openssl version tomorrow
(5:42:12 PM) kkoehne_: thiago: I'll try it once more. Last time I had issues with _INterlockedCompareExchange symbol missing, but it looks like they fixed it in latest headers already.
(5:42:25 PM) thiago: kkoehne_: symbol missing in the headers?
(5:42:38 PM) jaheikki3: OK, but decision is to use openSSL 1.0.1 or newer...
(5:42:42 PM) thiago: kkoehne_: or symbol missing when running?
(5:43:01 PM) peter-h: jaheikki3: yes
(5:43:20 PM) kkoehne_: thiago: missing when running. But as I said, I'll try once more :)
(5:43:22 PM) thiago: kkoehne_: this is really freaky. We don't use InterlockedCompareExchange. We use inline asm with GCC.
(5:43:35 PM) thiago: kkoehne_: let's take this after the meeting or when you have it done.
(5:43:36 PM) kkoehne_: thiago: I think it was coming from ICU
(5:44:15 PM) thiago: do we build ICU or do we get it pre-built from icu-project.org?
(5:44:21 PM) kkoehne_: thiago: We build it ourselves.
(5:44:30 PM) iikka_: thiagO: we build it
(5:44:38 PM) thiago: then the problem is there
(5:44:50 PM) thiago: InterlockedCompareExchange is an MSVC intrinsic
(5:44:55 PM) thiago: it's not a function
(5:45:14 PM) jaheikki3: kakoehne_: If there is some problem with newest mingw can we decide to use same than in 5.1?
(5:45:25 PM) kkoehne_: thiago: There has been long discussions on the mingw-w64 mailing list about it ...
(5:45:51 PM) kkoehne_: thiago: Also, 5.1.1 already doesn't compile anymore with latest Mingw-w64 headers.
(5:46:20 PM) kkoehne_: jaheikki3: Well, I think if we ship 4.8.0, 4.8.1 etc side by side, there is no problem.
(5:46:22 PM) iikka_: mingw 64bit?
(5:46:27 PM) jaheikki3: Because if we need offer both 4.8 & 4.8.1 it is again some extra work for us...
(5:46:31 PM) kkoehne_: iikka_: no, 32 bit.
(5:46:47 PM) iikka_: kkoehne_: ok
(5:47:06 PM) kkoehne_: jaheikki3: I agree that the name of the current package is then a bit misleading ...
(5:47:37 PM) jaheikki3: iikka_: Do you see any problem with this ?
(5:47:56 PM) kkoehne_: jaheikki3, iikka_, thiago: Alright, I'll sort out the compatibility issues, and write a mail about the choices we have on releasing@ tomorrow or so.
(5:48:07 PM) iikka_: well, if the 5.1.1 does not even compile with latest mingw version, yes it sounds a problem
(5:48:19 PM) jaheikki3: ah, true ;)
(5:48:29 PM) kkoehne_: iikka_: Well, fixes are already in stable :)
(5:48:36 PM) iikka_: ah, ok :)
(5:48:38 PM) kkoehne_: iikka_: ... or rather, on it's way to stable.
(5:49:02 PM) iikka_: but let's try to have the mingw version sorted out asap, if we'll upgrade it into 5.2
(5:49:13 PM) jaheikki3: kakoehne_:OK. Let's decide that mingw issue later in ML
(5:49:13 PM) kkoehne_: iikka_: It's just that we might end up replacing 4.8.0 with 4.8.1, and qt 5.1.1 binaries will still run ... but you won't be able to compile them any more :)
(5:49:35 PM) kkoehne_: iikka_: Safest choice is therefore to ship both, or to stick to gcc 4.8.0 package.
(5:49:50 PM) kkoehne_: iikka_: But people will complain then ...
(5:50:07 PM) iikka_: kkoehne_: I think we can not break 5.1.1 mingw builds
(5:50:24 PM) jaheikki3: iikka_: I agree...
(5:50:38 PM) iikka_: if people update with online installer, they would not be able to build 5.1.x anymore with mingw
(5:51:07 PM) kkoehne_: iikka_: Yeah. That's another argument for shipping both toolchains.
(5:51:22 PM) anshaw left the room (quit: Quit: No Ping reply in 180 seconds.).
(5:51:31 PM) kkoehne_: jaheikki3: Next topic? ;)
(5:51:43 PM) jaheikki3: ;)
(5:51:48 PM) iikka_: did we mention creator version for 5.2?
(5:51:49 PM) anshaw [~quassel at] entered the room.
(5:52:08 PM) kkoehne_: iikka_: THat's simple. We need 3.0, and 3.0 will require 5.2.
(5:52:23 PM) jaheikki3: Great, that was easy one ;)
(5:52:27 PM) thiago: sounds risky
(5:53:37 PM) jaheikki3: iikka_:Anything else?
(5:53:49 PM) iikka_: kkoehne_: at which point of time we should have the 3.0 ready? before beta?
(5:54:10 PM) thiago: binary packages are released at beta
(5:54:16 PM) thiago: so, yes, it has to happen before beta
(5:54:17 PM) thiago: or not at all
(5:54:18 PM) kkoehne_: iikka_: We can't have it ready before 5.2 is ready. We have to march together, or fail together.
(5:54:46 PM) kkoehne_: iikka_: we can't really release a creator 3.0 based on an unreleased 5.2.
(5:54:58 PM) kkoehne_: iikka_: So beta, rc, final have to be aligned.
(5:55:14 PM) iikka_: ok
(5:55:40 PM) anshaw left the room (quit: Read error: Connection reset by peer).
(5:55:48 PM) iikka_: that's all from my side....
(5:56:16 PM) jaheikki3: Does someone else has someting on his mind?
(5:56:27 PM) thiago: only that aligning is risky
(5:56:30 PM) thiago: do we have a Plan B?
(5:56:57 PM) jaheikki3: kakoehne_:?
(5:57:05 PM) peter-h: I am working on a new network test server, but that is more a CI thing than release I guess...
(5:57:16 PM) kkoehne_: thiago, jaheikki3: I don't see one.
(5:57:42 PM) kkoehne_: thiago, jaheikki3: But on the positive side, we've already released creator together with latest packages for every Qt 5 version.
(5:58:00 PM) wolfgang-b: jaheikki3, we are having problems with v4 related changes and I am not sure how fast we can resolve them. Currently, our qtdeclarative is not really usable.
(5:58:23 PM) kkoehne_: thiago, jaheikki3: The difference is now that we'll leave Qt 4 behind also for the standalone packages, and will most likely require 5.2 fixes .
(5:58:50 PM) hanne left the room (quit: Quit: hanne).
(5:59:02 PM) thiago: in the past, if creator slipped, we could always just release the previous version
(5:59:38 PM) jaheikki3: wolfgang-b: Ok, sounds a risk for the schedule...
(5:59:58 PM) kkoehne_: thiago: Yes. But the Android support will require latest creator, and the QML language changes too ...
(6:00:09 PM) thiago: what would happen if we released with 2.8?
(6:00:28 PM) thiago: Android would suffer a little, people using QML would not get syntax highlighting for the latest features
(6:00:31 PM) thiago: correct?
(6:00:52 PM) wolfgang-b: jaheikki3, at least for BlackBerry10. I am confused that we are the only ones who are having problems
(6:01:05 PM) kkoehne_: thiago: That's what I know, yes.
(6:01:33 PM) kkoehne_: thiago: Whether Android would suffer a little, or not work at all, I can't tell ...
(6:02:09 PM) kkoehne_: thiago: But honestly speaking, Qt Creator is a very good track record of meeting deadlines, much more than Qt.
(6:02:13 PM) jaheikki3: woldgang-b: Hoping those can be solved soon...
(6:02:34 PM) jaheikki3: kakoehne_:+1, I don't see so big risk with that
(6:02:49 PM) thiago: fair enough
(6:03:15 PM) jaheikki3: OK, maybe then last topic: Next meeting?
(6:03:19 PM) thiago: let's proceed then. the risk is low and there is an alternative if all should fail.
(6:04:26 PM) jaheikki3: I think i would be good to have meeting before feature freeze, 16.9?
(6:05:05 PM) kkoehne_: jaheikki3: Lars wrote 20th in his mail :)
(6:05:40 PM) thiago: 16 is before 20 :-)
(6:05:57 PM) kkoehne_: thiago, jaheikki3: Ah, sorry. was reading it the wrong way.
(6:06:02 PM) jaheikki3: kakoehne: Right ;)
(6:06:23 PM) jaheikki3: and that same time?
(6:06:23 PM) kkoehne_: jaheikki3: 16.9. is fine with me.
(6:06:48 PM) iikka_: ok for me as well
(6:06:59 PM) jaheikki3: Ok, let's have next meeting 16.9.2013 16:00 CET
(6:07:28 PM) jaheikki3: That's all this time. Thank's for everyone!
(6:07:32 PM) jaheikki3: Bye
(6:07:37 PM) iikka_: bye
(6:07:39 PM) peter-h: bye
(6:08:18 PM) wolfgang-b: bye
(6:08:22 PM) kkoehne_: bye
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/releasing/attachments/20130902/77ba38d2/attachment.html>

More information about the Releasing mailing list