[Releasing] State of Qt 5.0.0 wk 49 & Meeting minutes: release team meeting 04.12.2012

[Tanilkan Sinan]

Hi,

Thank you for the meeting today. Here are the minutes (which is also a summary of the state of Qt 5 for week 49):

_Update on the big remaining issues for the latest packages_
The big remaining issues for Qt 5 release candidate are all related to documentation and examples in the packages:
- No win installers available due to compilation issue, qtactiveqt - patch has been cherrypicked
- Documentation missing on Mac and Windows
- https://bugreports.qt-project.org/browse/QTBUG-28336

_No release candidate release candidate_
As Thiago mentioned on  releasing at qt-project.org (Subject: Updating the package releasing proposal), the release team agreed not to have 48 hours from a package is announced until it can be released.

For the release candidate this means that we will release the rc once the packages are ready and tested. (Earliest possible time is after the release team meeting tomorrow).

Next release team meeting: Wednesday 05.12.2012 at 16.00 (CET).

Best regards,
Sinan S. Tanilkan
on behalf of the Qt 5 releasing team

Irc log from the meeting:
[16:01:16] <SinanTanilkan> thiago, steveire, tjenssen, mauricek, treinio, lars, joaijala, iieklund__, sahumada, hanne: Ping
[16:01:21] <thiago> SinanTanilkan: pong
[16:01:23] <sahumada> pong
[16:01:24] <steveire> pong
[16:01:27] <treinio> pong
[16:01:28] <mauricek> SinanTanilkan: pong
[16:01:31] <hanne> pong
[16:01:32] <mauricek> andre_: ping
[16:02:07] <SinanTanilkan> (sorry for not adding you in the list andre_, that was not intentional).
[16:02:07] <johanna> pong
[16:02:26] <andre_> pong
[16:02:44] <SinanTanilkan> Any comments to the proposed agenda?
[16:03:11] <SinanTanilkan> Ok. Let's start then.
[16:03:11] <iikka> pong
[16:03:20] <SinanTanilkan> 1. Update on the packages that are currently available
[16:03:32] <iikka> ok
[16:03:53] <iikka> linux/mac installers were available, no win installers due to compilation issue, qtactiveqt
[16:04:03] <lars> SinanTanilkan: pong
[16:04:09] <iikka> fix for qtactiveqt has been picked by sergio
[16:04:25] <iikka> current issues in packaging:
[16:04:30] <iikka> QTBUG-28336
[16:04:49] <iikka> this needs to be fixed in the build system
[16:05:14] <iikka> i.e. qtbase/qmake
[16:05:51] <treinio> yes, it's a P0, ossi is on it
[16:05:56] <iikka> then the documentation, works on linux, on mac docs are still missing, make docs does not work (johanna?)
[16:06:19] <iikka> docs on win, don't know because no win installer lsat night
[16:06:22] <iikka> is johanna online?
[16:06:35] <johanna> yes, problems in doc installing on mac and win
[16:06:48] <johanna> andre_ is working on the mac issue, right?
[16:07:00] <andre_> iikka: currently all make docs in debug_and_release builds are probably breaking
[16:07:39] <iikka> andre_: for mac should it be then "-debug -release" only?
[16:07:42] <andre_> johanna: the issue is identified, ossi is working on a solution, and at least for mac there is some kind of workaround
[16:08:00] <iikka> should we change configure options or wait for the actual fix?
[16:08:09] <johanna> andre_: ok, how about the win failure, is that same issue?
[16:08:18] <andre_> johanna: likely
[16:09:04] <thiago> iikka: that's debug-and-release
[16:09:07] <kkoehne> iikka: I don't think we can go away from -debug-and-release on windows at least.
[16:09:23] <iikka> ok
[16:09:27] <andre_> iikka: the workaround is to copy docs/* from a linux package and cp -a $SRC/qt*/example/*  $INSTALL/qtbase/example
[16:09:51] <andre_> iikka: "-debug -release" is different from -debug-and-release
[16:10:27] <andre_> the latter generates Makefile, Makefile.debug and Makefile.release instead of a single Makefile
[16:10:46] <iikka> ok
[16:10:53] <iikka> andre_: joaijala: so it's ok to copy those from linux package to mac package?
[16:11:08] <andre_> the docs target are currently in the Makefile.debug and Makefile.release, whereas they should be only in the "dispatcher Makefile"
[16:11:15] <hanne> if the workaround works on windows as well as on mac (given the docs are not in that installer either) , it sounds tempting to go for that solution for the RC
[16:11:20] <kkoehne> I think andres proposal in the agenda makes sense. Use qch / example files from Linux on Windows, Mac too.
[16:11:36] <andre_> iikka: it works for me locally. we should at least try that once on the build machines
[16:11:38] <thiago> the docs should be the same, shouldn't they?
[16:11:41] <iikka> ok
[16:11:43] <thiago> build once, deploy everywhere
[16:11:48] <thiago> where have I heard that before? :-)
[16:11:49] <iikka> qtactiveqt docs?
[16:12:02] <iikka> that's win specific, right?
[16:12:04] <johanna> that's something we need to do manually (copying the docs..)
[16:12:08] <thiago> it shouldn't require building the code to build the docs
[16:12:11] <andre_> thiago: in thoery yes, and if not it's kind of a different, but minor bug
[16:12:26] <andre_> johanna: just two lines in some script
[16:12:31] <thiago> we need to have a master doc set anyway to put online
[16:12:58] <andre_> kai just now reported back that the same workaround works on windows
[16:12:59] <hanne> can we try the copying for tonight's packages?
[16:13:19] <iikka> we are not at office anymore
[16:13:21] <hanne> yay for windows too :)
[16:13:24] <iikka> I don't have vpn
[16:13:31] <johanna> same thing here
[16:13:33] <lars> iikka: make docs should generate activeqt docs on linux as well.
[16:13:41] <iikka> lars: ok, good
[16:13:50] <iikka> treinio: was there any other blockers?
[16:14:19] <hanne> iikka, johanna: oh - so you can earliest do it tomorrow morning?
[16:14:24] <lars> iikka: if it doesn't, let's fix it.
[16:14:41] <iikka> hanne: correct
[16:14:46] <treinio> iikka, the webkit docs are missing as well
[16:14:49] <johanna> hanne: yes, but after that we only need to make installers again, that doesn't take so much time
[16:15:02] <treinio> jpasion sent a mail about a minute ago
[16:15:03] <iikka> true
[16:15:06] <SinanTanilkan> iikka: Can anyone else do this??
[16:15:09] <hanne> ok - then we could still have packages by 6 pm tomorrow evening
[16:15:26] <iikka> yes
[16:15:42] <hanne> iikka: was that a yes to me or SinanTanilkan ?
[16:16:08] <iikka> hanne: answer to question "then we could still have packages by 6 pm tomorrow evening"
[16:16:23] <iikka> I meant we should be able to get packages with copied docs
[16:16:35] <hanne> understand, so none else can help for tonights packages?
[16:17:08] <iikka> kai proposed to help
[16:17:13] <andre_> iikka: how long does a "manually guided" build cycle take?
[16:17:13] <hanne> it would feel better if we could test it tonight, and still have a package build round to go on
[16:17:16] <kkoehne> I could try at least.
[16:17:19] <iikka> I will sync with kai
[16:17:34] <iikka> andre_: on local machine?
[16:17:46] <hanne> kkoehne iikka : great - trying is more helpful than not at all
[16:18:02] <andre_> iikka: I meant bamboo
[16:18:20] <kkoehne> andre_: Just come over at the end of the meeting, and we can have at ry.
[16:18:21] <iikka> 1-2h
[16:18:24] <iikka> on mac
[16:18:27] <andre_> iikka: right now we can't build packages locally, or can we?
[16:18:35] <iikka> yes we can
[16:18:41] <mauricek> andre_: only for local testing then
[16:18:45] <iikka> check out qtsdk.git
[16:18:48] <andre_> mauricek: ?
[16:18:56] <mauricek> andre_: sergio just send a mail to you about his setup
[16:19:05] <iikka> and fetch the src package, that's all you need to run the build
[16:19:05] <andre_> iikka: I have qtsdk.git .... ah
[16:19:17] <lqi> shiroki: about this webkit issue, https://bugreports.qt-project.org/browse/QTWEBKIT-383 , only for windows?
[16:19:29] <iikka> it's one liner to make the build with the script (packaging is another step)
[16:19:43] <steveire> If we have packages by 6pm tomorrow evening, are they going to be released the following morning?
[16:19:54] <shiroki> lqi: didn't try other platforms. let me try
[16:19:57] <iikka> andre_: kkoehne: let's sync after mtg?
[16:20:00] <kkoehne> Anyway, any local package will likely pull in unwanted dependencies, etc, so it's not something to make shippable packages.
[16:20:01] <andre_> sahumada: wow. thanks!
[16:20:13] <sahumada> andre_: I read your mind
[16:20:20] <andre_> iikka: ok
[16:20:20] <lqi> shiroki: at least when beta2, i tried on mac, no that issue
[16:20:27] <lars> steveire: we need to test them IMO
[16:20:35] <steveire> lars: I agree.
[16:20:54] <steveire> lars: I'm just trying to make sure that's the plan.
[16:20:55] <hanne> steveire lars: we could then test them 24 hours,
[16:21:10] <shiroki> lqi: yeah. linux works fine too
[16:21:15] <steveire> Why not 48 as we've previously discussed?
[16:21:24] <iikka> treinio: sahumada: SinanTanilkan: will the fix from ossi arrive in time?
[16:21:36] <sahumada> iikka: no idea
[16:21:49] <ossi|tt> iikka: well, i now found the problem actually
[16:22:00] <shiroki> lqi: i added a comment on the bug report
[16:22:28] <lqi> shiroki: i just guess, maybe it's a font issue?
[16:22:43] <iikka> ossi|tt: is it fixable in this time frame?
[16:23:05] <SinanTanilkan> Ok. Are there any other issues we should raise regarding the packages and bugs we know about?
[16:23:06] <ossi|tt> iikka: seems pretty trivial. i only need to find out how the same thing is done somewhere else
[16:23:23] <hanne> steveire: it would just be great to get it out on thursday, and if it is only doc and packaging changes, maybe 24 hours would be ok?
[16:23:35] <hanne> it feels like we are very close
[16:24:04] <steveire> hanne: It would also be great to get it out on Friday.
[16:24:15] <iikka> sahumada: can you pick the fix when it's ready?
[16:24:24] <steveire> hanne: My point is that what was agreed before now seems to be ignored.
[16:24:24] <sahumada> iikka: can do
[16:24:41] <hanne> steveire: not ignored - just a proposal
[16:25:13] <johanna> and once we update submodules there is the risk that some other change causes build failures
[16:25:14] <shiroki> lqi: yes. it is a font issue. :)
[16:25:20] <shiroki> lqi: i think so too
[16:25:26] <steveire> The 48 hours was agreed. If you're saying that 24 hours is a proposal, then that's new :). People are talking as if it's already agreed.
[16:25:27] <johanna> or can we be sure that updating submodules now is safe?
[16:26:00] <iikka> sahumada: can we cherry-pick only needed changes?
[16:26:10] <sahumada> iikka: nop
[16:26:38] <lars> steveire: that's not agreed IMO
[16:27:11] <hanne> steveire: any way - 24 or 48 hours - the packages would need to be ready by 6 pm tomorrow
[16:27:31] <hanne> steveire: sorry - i did not mean that it was agreed - bad phrasing
[16:28:09] <lars> hanne: if we want to have a change to release tis week we need packages by tomorrow, yes
[16:29:10] <johanna> I'm not sure if you are aware that thursday is independence day in Finland, we wont be working (by default)
[16:29:25] <SinanTanilkan> johanna: Thanks for letting us know.
[16:29:31] <lars> johanna: nope, wasn't aware of that
[16:29:56] <hanne> johanna: should be fine if the packages are ready tomorrow.. :)
[16:30:00] <kkoehne> Okay, how likely is it that we really find no issues blocking an RC ourselves before Tuesday? Or are we already accepting that any RC released will not be the final package?
[16:30:38] <johanna> hanne: yep, no pressure for tomorrow.. :)
[16:30:47] <hanne> kkoehne: Tuesday?
[16:30:51] <SinanTanilkan> kkoehne: i think this increases risk, but it shouldn't stop us yet.
[16:30:59] <steveire> Still, it's not two working days on Finland. I'll leave it to lars about whether he thinks that matters. I hope no one thinks the 48 hours includes weekends though...
[16:31:15] <-- qt_imac_107 (~qa at 202.84-49-38.nextgentel.com) has quit (Quit: qt_imac_107)
[16:31:15] <andre_> kkoehne: there are quite a few little issues in docs left, so I would indeed expect the final not being exactly the rc
[16:31:39] <SinanTanilkan> Question:
[16:31:40] <lars> steveire: we IMO can't take all public holidays in the world into account.
[16:31:51] <steveire> I think that's reasonable
[16:31:55] <SinanTanilkan> My under standing is that the problems we are seeing are all related to documentation.
[16:32:03] <kkoehne> and examples
[16:32:06] <hanne> kkoehne, andre_ : docs should still be ok to fix up a bit before the final
[16:32:06] <thiago> guys, this is an RC
[16:32:20] <lars> SinanTanilkan: yes, but that's what makes a good or bad impression of the 5.0 release
[16:32:21] <andre_> hanne: that's what I think, ye
[16:32:27] <thiago> I'd say that if the package compiled and we have no brown paper bag problems, release it
[16:32:33] <thiago> THEN we do the actual release testing
[16:32:44] <thiago> if we find no issues, we do the *final*
[16:33:08] <SinanTanilkan> thiago: How does that relate to the 48 h rule?
[16:33:54] <thiago> I think we need to rethink it
[16:34:11] <SinanTanilkan> Ok. Let's do that then.
[16:34:21] <lars> thiago: what's your proposal then?
[16:34:27] <thiago> the proposal is:
[16:34:46] <thiago> this team produces packages and does a minimal testing to ensure that it built and it's not brown-paper-bag
[16:34:55] <thiago> if that is ok, we release it
[16:35:08] <thiago> it's an RC
[16:35:13] <lars> thiago: and you rather do a second rc in a week?
[16:35:14] <thiago> then the full testing cycle commences
[16:35:16] <thiago> yes
[16:35:23] <thiago> if there were issues, they get fixed and it's a new RC
[16:35:44] <thiago> if there were only minor things, we repackage with those minor things and call it final
[16:36:20] <hanne> but we wait until the packages tomorrow?
[16:36:22] <thiago> with those minor things fixed, I mean
[16:36:36] <thiago> do we have a built package set that, if released, wouldn't make us wear brown paper bags over our heads?
[16:36:37] <lars> ok, assume you still want a 48 hour rule for the final, right?
[16:36:39] <mauricek> hanne: we have to, as we have no windows packages at all right now
[16:36:54] <thiago> lars: they'll have had 1 week
[16:37:04] <SinanTanilkan> I think this is probably a better way to involve more people in trying out the packages.
[16:37:04] <lars> true :)
[16:37:06] <hanne> mauricek: they are ready any minute now though
[16:37:15] <mauricek> hanne: right...
[16:37:25] --> iikka_ (~iikka at 188.238.215.110) has joined #qt-releases
[16:37:29] <SinanTanilkan> Any objections to this proposal?
[16:38:03] <hanne> mauricek: or maybe not..
[16:38:33] <SinanTanilkan> Ok. there doesn't seem to be any objections.
[16:38:44] <SinanTanilkan> Let's agree on the next steps.
[16:38:46] <thiago> SinanTanilkan: shall I write the proposal up for the ML?
[16:38:56] <SinanTanilkan> thiago: Please do.
[16:38:56] <lars> thiago: please do
[16:38:56] <thiago> just for the benefit of others
[16:38:59] <thiago> ok
[16:39:19] <-- iikka (~iikka at gprs-internet-ffcdee00-119.dhcp.inet.fi) has quit (Ping timeout: 244 seconds)
[16:39:30] -*- andre_ opens a sidetrack: I propose to agree to not ship compiled examples in the final (but accept whatever will turn up in the RC packages)
[16:39:38] <SinanTanilkan> Assuming we have new packages in the morning, the QA team in Oslo can sanity test these packages.
[16:39:39] <andre_> this is related to https://bugreports.qt-project.org/browse/QTBUG-28349
[16:39:53] <thiago> SinanTanilkan: are they being built right now?
[16:40:09] <thiago> if so, the people in CA can do some initial testing later today
[16:40:20] <SinanTanilkan> thiago: True.
[16:40:30] <sahumada> thiago: no .. they start at 21:00 Oslo time (or so)
[16:40:35] <johanna> thiago: new build round will start arount 21 EET
[16:40:40] <iikka_> correct
[16:40:58] <SinanTanilkan> Ok. So earliest next packages would be tomorrow morning in europe?
[16:41:08] <sahumada> yes
[16:41:13] <johanna> SinanTanilkan: yes
[16:41:23] <SinanTanilkan> Ok. So I propose we ask the qa team in oslo to do sanity testing of those.
[16:41:25] <sahumada> if everything succeeds
[16:41:43] <lars> hmm... can't we start a new set of packages now?
[16:41:56] <iikka_> machines will reboot 21:00
[16:41:59] <sahumada> lars: the machines are rebooted at 21:00
[16:42:05] <iikka_> win packages won't finish in time
[16:42:10] <lars> *sigh*
[16:42:39] <sahumada> and the last packages that were supposed to be done by now failed on Windows
[16:42:47] <lars> iikka_: we're still not using jom then?
[16:42:55] <iikka_> tried
[16:43:07] -*- treinio gives +1 to andre_ 's proposal about example binaries
[16:43:09] <iikka_> it was failing in some install steps (johanna)
[16:43:30] <johanna> yes there was some problems with installing
[16:43:34] <kkoehne> iikka_: Well, that would still mean we could use it for compilation
[16:43:41] <mauricek> johanna: even with -j1 ?
[16:44:04] <johanna> mauricek: I ran out of time when trying that (I mean office hours..)
[16:44:23] <johanna> but manually testing building with jom and installing with nmake worked
[16:44:34] <johanna> so there is just something wrong in our script
[16:44:46] <lars> ok, after 5.0 we really need to do something about the packaging infra. the automatic reboot at 9pm and not using jom are large problems IMO.
[16:44:48] <SinanTanilkan> What do we want to do with the results from the sanity testing from the qa team? Do we want to bring their results into another release team meeting, and then release? Or do we want to ask them to say go ahead if they see no big problems?
[16:45:01] <mauricek> johanna: yeah, jom install on multiple instances breaks, that is why you have to do jom -j1 install
[16:45:24] <hanne> SinanTanilkan: can we have a new meeting same time tomorrow?
[16:45:26] <mauricek> lars: agreed.
[16:45:38] <SinanTanilkan> hanne: I think that is a good idea.
[16:45:47] -*- hanne gives +1 too to andre_ 's proposal about example binaries
[16:46:18] -*- mauricek in any case
[16:46:23] <lars> ok, let's get rid of them then. Just make sure people can easily figure out how to compile/run them themselves.
[16:46:32] <kkoehne> lars: Btw, linux make docs does not include qtactiveqt
[16:46:50] <mauricek> kkoehne: most likely because it is not cloned, right?
[16:46:54] <johanna> right now we are building the examples, removing that means update in build script, which means new build round tomorrow, right iikka?
[16:46:59] <lars> kkoehne: ok. should be possible to fix the pro files
[16:47:03] <lars> mauricek: it's cloned.
[16:47:29] <mauricek> lars: kkoehne: if we get doc in from linux, let's make a p2 for qtactiveqt doc and still go with that solution
[16:47:46] <lars> mauricek: agreed.
[16:48:00] <kkoehne> mauricek: Sure
[16:48:12] <SinanTanilkan> Ok. Next steps. QA team does sanity testing (assuming we have new packages), and we meet again tomorrow at the same time.
[16:48:14] -*- kkoehne guesses it's because of a big win32 {} scope in qtactiveqt.pro
[16:48:33] <lars> kkoehne: yes. need to pull the doc target out of it I guess
[16:48:40] <iikka_> what was the decision about example binaries, my internet connection was lost for a second
[16:48:41] <hanne> SinanTanilkan: and we do some fixing too before that..
[16:49:01] <SinanTanilkan> hanne: Agree :)
[16:49:03] <mauricek> SinanTanilkan: hanne: one thing i did not fully get now. Will we do the copying from docs and examples tomorrow morning and create new packages?
[16:49:04] <sahumada> iikka_: dont provide binaries
[16:49:12] <mauricek> or take the ones where doc is definitely broken for testing?
[16:49:47] <johanna> then we need to do another build round tomorrow (to remove examples), will take several hours to get new packages
[16:49:48] <lars> mauricek: we need to test/check docs before we release.
[16:49:51] <mauricek> or test what is there tomorrow morning and recreate packages with copied doc as a one shot for tomorrow afternoons packages?
[16:50:00] <hanne> mauricek: my understanding is that we do the copying  asap, and try to build new packages - either that happens before tonight or tomorrow morning
[16:50:13] <mauricek> hanne: ok, great
[16:50:18] <iikka_> mauricek: +1
[16:50:35] <SinanTanilkan> Ok.
[16:50:50] -*- andre_ took the liberty to put sergio's recipe on the wiki: http://qt-project.org/wiki/Building-Qt-Package
[16:51:00] <SinanTanilkan> I think we have been trough everything on the agenda. Anything else we need to discuss now?
[16:51:16] <iikka_> to summarise:
[16:51:29] <iikka_> - don't provide example binaries?
[16:51:36] <mauricek> iikka_: for the final
[16:51:42] <iikka_> ok
[16:51:57] <iikka_> - try the doc hack this evening?
[16:52:14] -*- thiago is ok for not providing example binaries
[16:52:18] <mauricek> yup
[16:52:26] <thiago> we have lost the qtdemo, haven't we?
[16:52:35] <mauricek> thiago: yup... gone since months
[16:52:40] <lars> iikka_: example binaries is not a showstopper for RC. so there it's ok to have or not have them... :)
[16:52:46] <iikka_> ok
[16:52:57] <treinio> thiago: yes, that would be the only point in having the binaries in place imo, but we don't have it
[16:53:01] <lars> thiago: yes, we'll need to create a new one
[16:53:23] <hanne> yes for 5.x
[16:54:25] <lars> hanne: of course :)
[16:54:31] <SinanTanilkan> Ok. Anything else?
[16:54:35] <thiago> so for now, there is no example launcher
[16:54:39] <thiago> no need to have them built
[16:54:45] <mauricek> thiago: exactly
[16:54:51] <thiago> when we have a new qtdemo, we'll think about it again
[16:54:54] <andre_> thiago: creator is de-facto the example launcher replacement
[16:55:06] <andre_> thiago: you open an example, click on run.
[16:55:08] <thiago> andre_: and it builds them :-)
[16:55:11] <andre_> (in theory)
[16:55:16] <treinio> andre_: ..and it rebuilds
[16:55:23] <treinio> yeah :)
[16:55:36] <andre_> thiago: right, but that's rather quick. a few seconds for the ones I checked
[16:56:13] <thiago> ok, anyone who opposes this?
[16:56:20] <andre_> I think it's definitely a good tradeoff, even given the fact that this started off as a workaround
[16:56:51] <thiago> btw, our installation of example.pro is an issue
[16:57:11] <mauricek> thiago: ssssshhhhhh ;)
[16:57:13] <thiago> but I'll report that after I build RPMs out of the RCs
[16:57:14] <thiago> later
[16:57:46] <andre_> ok
[16:58:16] -*- hanne thinks this meeting is over, and we can go over to regular irc'ing
[16:58:21] <kkoehne> thiago: which of the two example.pro files that we package do you mean ? ;)
[16:58:34] -*- mauricek wonders what hanne does differently now ;)
[16:58:36] <steveire> Is there a bug for that issue?
[16:58:44] <thiago> steveire: not yet
[16:58:45] <andre_> a solution would be to rename the module-level example.pro's into module-example.pro and put a example.pro subdirs .pro into qtbase/example
[16:58:46] <lars> kkoehne: nice ;-)
[16:58:56] <thiago> guys, we can discuss that *after* I report it
[16:59:03] <thiago> it's not an RC-release issue
[16:59:05] <SinanTanilkan> I propose we end the release team meeting.
[16:59:15] <mauricek> SinanTanilkan: +1
[16:59:15] <thiago> it's a final issue, which is what we have RCs for :-)
[16:59:19] <SinanTanilkan> Unless you have anything else that needs the teams attention?
[16:59:23] -*- hanne just wanted to help SinanTanilkan find the end of the meeting for his minutes :)
[16:59:35] <SinanTanilkan> Thanks for your time. See you tomorrow.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/releasing/attachments/20121204/f6e42d0a/attachment.html>