**** BEGIN LOGGING AT Fri Nov 01 02:59:59 2013
Nov 01 07:58:35 <Herrie> Morning
Nov 01 08:21:29 <morphis> Herrie: morning
Nov 01 08:21:58 <morphis> Herrie: thanks for fixing the bug with the memos apps
Nov 01 08:22:02 <morphis> s/apps/app/
Nov 01 08:39:17 <morphis> Herrie: merged
Nov 01 08:58:26 <morphis> Tofe: morning
Nov 01 09:03:39 <Herrie> Well it's not a bug, more a feature ;) Trying to get some low hanging fruits sorted ;)
Nov 01 09:04:41 <Herrie> The things that are within my knowledge and that's limited LOL
Nov 01 09:14:59 <Herrie> Sent out a tweet y'day to see if we can get some Enyo app developers to help out ;)
Nov 01 10:08:16 <Garfonso> morning
Nov 01 10:11:25 <morphis> Garfonso: morning
Nov 01 10:11:31 <morphis> Garfonso: how was/is your holiday?
Nov 01 10:12:51 <Garfonso> we have been in istanbul for five days. Was great. :)
Nov 01 10:13:54 <morphis> nice
Nov 01 10:14:23 <morphis> need some holiday too soon ..
Nov 01 10:35:57 <Herrie> Interesting read: http://arstechnica.com/gadgets/2013/10/googles-iron-grip-on-android-controlling-open-source-by-any-means-necessary/
Nov 01 10:36:33 <Herrie> Seems like Google is closing more and more things off from public.... Could mean some possibilities for other Open Source OS-es to pick up developers who aren't happy with this ;)
Nov 01 11:01:26 <morphis> Herrie: it seems like they are moving AOSP to be the core os for Android
Nov 01 11:02:06 <morphis> Garfonso: I've the console output for webapps already implemented
Nov 01 11:02:14 <morphis> it just needs to be included in a public build
Nov 01 11:05:39 <Tofe> morning!
Nov 01 11:06:33 <morphis> Tofe: morning
Nov 01 11:07:06 <Tofe> morphis: http://pastebin.com/J2CkAFbX  <-- this is the hybris debug, we can see it creates wrongly a nativewindow just before the egl error
Nov 01 11:07:18 <morphis> which line?
Nov 01 11:07:41 <Tofe> 1204
Nov 01 11:08:21 <morphis> Garfonso: when you've some time I think we need to look at preware to get it working correctly
Nov 01 11:08:27 <Tofe> well 1204 is the setFormat problem; the window is created some lines above
Nov 01 11:08:46 <Tofe> I still can't explain why it does that; we're right in the middle of hiding the input method
Nov 01 11:09:17 <JaMa> morphis: any idea about maliit-framework-qt5 failing like this:
Nov 01 11:09:18 <JaMa> | Could not find qmake configuration file linux-oe-g++.
Nov 01 11:09:32 <morphis> JaMa: I saw this once
Nov 01 11:09:45 <morphis> I cleaned things up and restarted the build and it worked then
Nov 01 11:09:59 <morphis> Tofe: ah I see:
Nov 01 11:09:59 <morphis> Oct 31 12:00:18 maguro maliit-server[1434]: PID: 1434 Tracepoint-E/create_window::wayland-platform
Nov 01 11:10:36 <JaMa> morphis: ok, I've seen it in jenkins qemux86-64 build and now got the same locally
Nov 01 11:10:47 <Tofe> morphis: as the GL code in Qt is in its own thread, I can't see the exact line that triggers this window creation
Nov 01 11:10:59 <morphis> JaMa: suspect there is something wrong with the staging
Nov 01 11:11:54 <JaMa> morphis: btw I've patch for meta-qt5/master to update cmake .bbappend which will make master branch incompatible with dora release
Nov 01 11:12:15 <morphis> JaMa: what does the patch change?
Nov 01 11:12:39 <JaMa> fixes building cmake after building qtbase, mmt
Nov 01 11:12:57 <JaMa> https://github.com/meta-qt5/meta-qt5/commit/b6df9940324fe8007fbbe7de076075485d50a925
Nov 01 11:13:31 <morphis> JaMa: ok, so we need our own branch for meta-qt5
Nov 01 11:14:01 <JaMa> I feared that you'll say that :)
Nov 01 11:14:25 <morphis> Tofe: you looked at the qtwayland repo? maybe they fixed something recently we still don't have in our version?
Nov 01 11:14:26 <JaMa> OR backport newer cmake to meta-webos-ports
Nov 01 11:14:44 <morphis> JaMa: if that works flawlessly it's fine for me
Nov 01 11:14:45 <JaMa> current cmake version in oe-core seems to handle more of qt5 related stuff
Nov 01 11:14:53 <Tofe> no, I didn't think of that; I'll have a look
Nov 01 11:15:17 <JaMa> well - it's my assumtion from the number of find_package(qt5* they have added since last release
Nov 01 11:15:42 <morphis> Tofe: looking at https://github.com/mer-packages/qtwayland currently
Nov 01 11:15:50 <morphis> JaMa: lets try that
Nov 01 11:16:02 <morphis> would be better instead of having a branch for meta-qt5
Nov 01 11:16:52 <JaMa> http://www.cmake.org/pipermail/cmake/2013-October/056020.html
Nov 01 11:17:15 <JaMa> announcement doesn't mention qt5 but I'll try to backport and see
Nov 01 11:17:48 <morphis> ok
Nov 01 11:18:53 <Tofe> morphis: they sure have interesting patches
Nov 01 11:19:15 <morphis> Tofe: some of them I already backported to our branch
Nov 01 11:20:40 <morphis> Tofe: take a look at ./src/plugins/platforms/qwayland-egl/qwaylandeglwindow.cpp of qtwayland
Nov 01 11:21:46 <morphis> Tofe: so somehow maliit is really recreating a window here
Nov 01 11:22:22 <Tofe> yes
Nov 01 11:22:57 <morphis> so the only path I can imagine where this happens is when it hides the window
Nov 01 11:23:28 <morphis> or at least: it doesn't looses the window but the egl surface
Nov 01 11:25:19 <morphis> JaMa: btw. valgrind builds fine here with the patch
Nov 01 11:25:24 <morphis> just need to check it in runtime
Nov 01 11:25:38 <Tofe> but I don't see how calling things like m_surface->hide() would trigger a window creation; do you ?
Nov 01 11:26:51 <morphis> Tofe: hm, it was right that maliit resizes the window to 0x0 when it hides?
Nov 01 11:27:04 <Tofe> I think so
Nov 01 11:31:19 <Tofe> I will try to follow the hide() call a bit
Nov 01 11:33:53 <morphis> ok
Nov 01 11:35:10 <JaMa> morphis: I've already pushed it to dora-next, so it will be in next rebase
Nov 01 11:35:15 <morphis> ok
Nov 01 11:38:28 <morphis> Tofe: maybe you can add debug calls to QWaylandEglWindow::eglSurface() too
Nov 01 11:38:44 <morphis> maybe we see that way where we loose the window
Nov 01 11:47:37 <Tofe> that's what I'm doing :)
Nov 01 11:48:01 <Tofe> I'm trying to see where setFormat is called with format=0
Nov 01 11:55:42 <morphis> Tofe: I am fixing the statusbar currently to work correctly
Nov 01 11:56:31 <morphis> there are some problems with the status bar services connector
Nov 01 11:57:03 <Tofe> ok, yes I saw a qml error somewhere about the wifi indicator, but I didn't have time to investigate
Nov 01 12:00:57 <morphis> I found the problem already
Nov 01 12:14:47 <morphis> Garfonso: new emulator build with the app console output is available now
Nov 01 12:21:49 <Tofe> morphis: what means "E" in Tracepoint-E ?
Nov 01 12:21:56 <Tofe> (in hybris)
Nov 01 12:22:12 <Tofe> ah, "End", maybe
Nov 01 12:22:17 <morphis> End or Error
Nov 01 12:22:42 <Tofe> Error ?
Nov 01 12:23:07 <Tofe> when eglError is set/reset ?
Nov 01 12:24:11 <Tofe> ah, no, ok, I read "of" when you wrote "or", sorry :)
Nov 01 12:25:11 <morphis> :)
Nov 01 12:27:49 <HaDAk> did we ever get that list over to enda?
Nov 01 12:29:24 <morphis> HaDAk: you saw my last comments regarding the list?
Nov 01 12:29:29 <HaDAk> no?
Nov 01 12:29:47 <HaDAk> did you see that i went through and broke it up a bit?
Nov 01 12:29:53 <morphis> I think most of the icons are already available
Nov 01 12:30:01 <HaDAk> nice.
Nov 01 12:30:09 <HaDAk> we should figure out which are and which aren't.
Nov 01 12:30:26 <HaDAk> i just looked for all the .png files in the doctor and pulled out a list of ones that i thought would be extremely helpful.
Nov 01 12:30:45 <HaDAk> to maintain continuity with previous versions of webos, in the UI
Nov 01 12:31:07 <morphis> HaDAk: look at github.com/openwebos/luna-sysmgr for example
Nov 01 12:33:32 <Tofe> morphis: when calling eglCreateWindowSurface, it changes the format from 4 to 0
Nov 01 12:38:58 <HaDAk> morphis: you're right. i'll download the repo and try to dig through.
Nov 01 12:52:15 <Herrie> HaDAk: When you have the list filtered, let me know. I wanted to look into it this week but didn't get to it yet due to construction stuff here at home :P
Nov 01 12:52:41 <HaDAk> Herrie: i understand. i'm working on another side project too. trying to split my time between them.
Nov 01 13:05:54 <Tofe> morphis: maybe I have an hypothesis: hide() is called, it sends wayland events to unmap the window. but no wl_roundtrip is asked. Then, Qt does a makeCurrent to handle "WM_TryRelease" event on a QuickWindow. Then makeCurrent calls window->eglSurface(), which will create the surface that, it seems, was not yet created. So now I'm trying to see what is that annoying QWaylandEglWindow that does things the wrong way.
Nov 01 13:15:22 <Tofe> morphis: my breakpoint in QWaylandEglWindow::QWaylandEglWindow is never reached, did I make a mistake ?
Nov 01 13:18:53 <Tofe> looks like we are happily playing with non allocated pointers...
Nov 01 13:35:38 <morphis> Tofe: can you try starting maliit-server with QT_QPA_PLATFORM=wayland-egl
Nov 01 13:35:47 <morphis> afaik there are two platforms: wayland and wayland-egl
Nov 01 13:36:11 <Tofe> I can try that yes
Nov 01 13:36:17 <morphis> but afaik we've configured wayland-egl as the default
Nov 01 13:37:06 <morphis> Tofe: https://github.com/webOS-ports/luna-next/pull/51
Nov 01 13:37:27 <morphis> Tofe: https://github.com/webOS-ports/luna-next-cardshell/pull/37
Nov 01 13:39:37 <Tofe> morphis: actually, the way I did the connect was that it was the only way I found to have a working Test "singleton"
Nov 01 13:39:54 <morphis> ah
Nov 01 13:41:08 <morphis> Tofe: let me rework that
Nov 01 13:41:22 <Tofe> otherwise it's quite ok, the syntax was horrible anyway
Nov 01 13:42:35 <morphis> :)
Nov 01 13:42:59 <Tofe> no, let's keep it that way. I'll merge.
Nov 01 13:43:33 <morphis> ok
Nov 01 13:51:27 <morphis> Tofe: you tried to disable the windows which the keyboard creates?
Nov 01 13:51:33 <morphis> afaik it uses three
Nov 01 13:51:45 <morphis> maybe one of them is causing the problem we're seeing here
Nov 01 13:52:09 <Tofe> no I didn't try yet
Nov 01 13:53:15 <morphis> Tofe: you're in #maliit?
Nov 01 13:53:28 <morphis> mikhas is one of the maliit developers
Nov 01 13:53:35 <morphis> seems like he's currently available
Nov 01 13:53:57 <Tofe> I don't know yet what precise question to ask
Nov 01 13:56:05 <morphis> yeah thats the hard thing :)
Nov 01 14:04:09 <Tofe> oh great. I changed a bit libhybris to get the traces, and now my build tree is all to be rebuilt
Nov 01 14:05:01 <morphis> :)
Nov 01 14:05:21 <morphis> what did you change?
Nov 01 14:05:35 <morphis> --enable-debug --enable-trace?
Nov 01 14:07:12 <Tofe> yes
Nov 01 14:08:38 <morphis> we can include them by default
Nov 01 14:09:30 <Tofe> we could, in the -dev image it would be logical
Nov 01 14:10:24 <morphis> both are adding just the possibility, right?
Nov 01 14:11:07 <Tofe> right
Nov 01 14:11:45 <Tofe> the "if" might cost a little bit cpu, but largely acceptable
Nov 01 14:18:38 <morphis> right
Nov 01 14:32:00 <Tofe> disable the hide() of the two subsidiary surfaces didn't help avoid the glError, so it seems it happens on the main window
Nov 01 14:36:15 <morphis> hm
Nov 01 14:38:43 <Tofe> and my gnexus is less and less stable... but given the hacks I do to the rootfs, I'm not that surprised.
Nov 01 14:39:53 <Tofe> disabling show() and hide() on those two surfaces didn't help either. I fear it's really about the main window.
Nov 01 14:43:37 <Tofe> maybe it's because they use a QQuickView ? I don't know.
Nov 01 14:45:01 <morphis> Tofe: QQuickView should not matter here
Nov 01 16:00:48 <Tofe> morphis: breakthrough (hum, partial): it's about transientParent
Nov 01 16:39:56 <Tofe> hum, very partially. I have also to disable show/hide for the two extra surfaces
Nov 01 16:50:41 <Tofe> ok, great, now we get a crash in qwlsurface.cpp, line 510: the surfaceBuffer is NULL
Nov 01 16:58:59 <Tofe> morphis: are we going to crash if a Q_ASSERT is not true in qtwayland ?
Nov 01 17:10:53 <morphis> Tofe: yeah!
Nov 01 17:12:01 <morphis> Tofe: afaik the Q_ASSERT should be disabled in release mode
Nov 01 17:12:44 <Tofe> ok, so it could be that we end up adding a buffer that is NULL in qtwayland.
Nov 01 17:15:02 <morphis> possible
Nov 01 17:15:35 <Tofe> I'll add a qWarning, to be sure.
Nov 01 17:16:39 <morphis> ok
Nov 01 17:17:29 <Tofe> (and I'll test the pointer...)
Nov 01 17:27:52 <Tofe> morphis: with the sideload, where is the partition located ?
Nov 01 17:28:14 <morphis> ?
Nov 01 17:28:21 <morphis> it's using the userdata partition
Nov 01 17:28:34 <morphis> creating a webos directory in it and extracting things there
Nov 01 17:29:55 <morphis> Tofe: https://github.com/webOS-ports/android-update-package/blob/master/webos_deploy.sh
Nov 01 17:30:33 <Tofe> because I think I've reached the mount limit or something like that, and it won't let me remount rw the partition
Nov 01 17:31:22 <morphis> our initrd does some bind mounts
Nov 01 17:31:27 <morphis> but / shouldn't be ro yet
Nov 01 17:32:15 <morphis> have to leave for some time
Nov 01 17:32:30 <Tofe> ok
Nov 01 19:18:35 <Tofe> morphis: I've solved the keyboard hide/show issue
Nov 01 19:30:25 <Herrie> morphis/Tofe: Do we have a solution for messaging (SMS and the likes) already? If not, what's the idea for that?
Nov 01 19:31:34 <Herrie> Tofe: Great news, this means the kb will work properly now?
Nov 01 19:31:44 <Herrie> Or still more hurdles to take?
Nov 01 19:43:57 <Tofe> Herrie: well, it's still not perfect, plenty of bugs here and there, but at least the kb is there
Nov 01 19:44:45 <Tofe> Herrie: for telephony, afaik, we do have the infrastructure, but not the UI yet
Nov 01 20:17:58 <Tofe> morphis: https://github.com/webOS-ports/meta-webos-ports/pull/15
Nov 01 20:41:52 <Tofe> morphis: https://github.com/webOS-ports/luna-next-cardshell/pull/38 & https://github.com/webOS-ports/luna-next/pull/52
Nov 01 20:55:38 <Herrie> morphis: Is it me or email app icon not fixed yet in today's VBox?
Nov 02 00:32:44 <LarrySteeze> herrie, Tofe, Morphis: morning
Nov 02 00:35:29 <HaDAk> hey LarrySteeze
Nov 02 00:35:48 <LarrySteeze> HaDAk: yo
Nov 02 00:36:02 <HaDAk> we just haven't been able to catch up for a while, eh
Nov 02 00:46:14 <JaMa> morphis: images built fine with newer cmake, I'll try in runtime too, but I don't expect issues there
**** ENDING LOGGING AT Sat Nov 02 02:59:57 2013