**** BEGIN LOGGING AT Tue Jan 27 02:59:58 2015
Jan 27 06:32:05 <DougReeder> Garfonso, how do I trigger the contactlinker after I’ve added a contact:1 record to DB8?
Jan 27 06:55:10 <morphis> Herrie: ok
Jan 27 07:11:23 <morphis> Herrie: I've picked https://github.com/webOS-ports/meta-webos-ports/pull/102 now but the commit message doesn't help much
Jan 27 07:11:27 <morphis> leave the commit line as
Jan 27 07:11:34 <morphis> <component name>: bump SRCREV
Jan 27 07:11:42 <morphis> then in the body include what has changed
Jan 27 07:11:57 <morphis> no guesses or what might have changed
Jan 27 07:17:48 <morphis> new builds with a lot of changes (including dockmode) are triggered now
Jan 27 07:43:32 <Herrie> OK
Jan 27 07:51:37 <Tofe> Morning!
Jan 27 08:01:55 <Herrie|Veer> morning from office
Jan 27 08:08:14 <morphis> Tofe: morning!
Jan 27 08:10:02 <Herrie|Veer> morphis: I saw a commit for the webappmgr as well :) That will sort the area above the vkb + some focus issues when launching apps?
Jan 27 08:10:46 <morphis> Herrie|Veer: no, nothing to do with the area about the vkb
Jan 27 08:10:51 <morphis> just some app internal things
Jan 27 08:12:01 <Herrie|Veer> Ah ok... Looked like it might fix the broken vkb area just above the vkb, too bad it doesn't ;) Will become a pain to send sms currently :P
Jan 27 08:15:23 <Herrie|Veer> morphis: Working on the mcc bit now, just need to adjust some internal logics still
Jan 27 08:16:06 <Herrie|Veer> It returns the mcc after it already populated the countrylist so need to work around that.
Jan 27 08:52:16 <morphis> Herrie|Veer: maybe doing the call for the mcc first and then get the countrylist
Jan 27 08:53:47 <Herrie|Veer> Yeah that's what I do but the call finishes after the countrylist finishes :P
Jan 27 08:53:55 <Herrie|Veer> So need to work around that a bit
Jan 27 08:54:55 <morphis> ?
Jan 27 08:55:15 <morphis> so you're doing the call for the countrylist not in onResponse of the mcc one?
Jan 27 08:56:28 <Herrie|Veer> No not yet ;) Need to fix that :=
Jan 27 08:56:45 <Herrie|Veer> Been a bit busy these days, day job taking lots of time :P
Jan 27 09:54:52 <morphis> Garfonso: you tried the modifed validator with one of the latest builds?
Jan 27 10:11:00 <morphis> Garfonso: works fine, just the close call is missing
Jan 27 10:33:43 <Herrie|Veer> morphis: Any luck on finding the memory issues (particularly TP)?
Jan 27 10:34:17 <morphis> no
Jan 27 10:34:28 <morphis> but it must come from the underlaying graphics system
Jan 27 10:34:35 <morphis> we're cleaning all memory
Jan 27 10:34:51 <morphis> but somehow the memory used by a window is never really freed
Jan 27 10:35:09 <morphis> so we end up with 80M for each closed window
Jan 27 10:35:47 <morphis> that doesn't matter for some applications like the QML ones which don't have multiple windows and are going away once their window got closed
Jan 27 10:35:52 <morphis> but for the webappmanager it does
Jan 27 10:36:05 <morphis> also it seems like the GPU memory itself is freed correctly
Jan 27 10:36:21 <morphis> so not sure yet who remains using such an amount
Jan 27 10:42:20 <Tofe> morphis: so it's a leak when we destroy a childWindow, right
Jan 27 10:42:40 <morphis> yes
Jan 27 10:42:57 <morphis> Tofe: but it is the WebAppManager process which consumes all the memory
Jan 27 10:43:03 <morphis> so not a problem on the compositor side
Jan 27 10:43:13 <morphis> Tofe: just try it on the tp
Jan 27 10:43:26 <morphis> and monitor memory with top -p `pidof LunaWebAppManager`
Jan 27 10:44:43 <Tofe> Yes, I understand, I'm trying to see if a QML object might still be alive in the process instance
Jan 27 10:46:11 <Tofe> like the mRootItem ?
Jan 27 10:46:35 <morphis> I tried deleting it explicitly
Jan 27 10:46:40 <morphis> but that doesn't make any difference
Jan 27 10:46:45 <Tofe> ah. ok.
Jan 27 10:46:56 <morphis> also registered to the destroyed signal of certain objects
Jan 27 10:46:58 <morphis> all looked fine
Jan 27 10:47:29 <Tofe> so the QML hierarchy seems to be freed... that pretty much kills my main idea :)
Jan 27 10:48:51 <Tofe> what is surprising is that it doesn't crash, actually -- if the GPU freed the memory and Qt always uses direct pointers to it, it must be pretty unstable
Jan 27 10:49:31 <Tofe> how much does a child window cost ? ~80Mb ? 160Mb ?
Jan 27 10:52:49 <morphis> about 80Mb
Jan 27 10:52:59 <morphis> Tofe: you can do a
Jan 27 10:53:01 <morphis> mkdir /d
Jan 27 10:53:09 <morphis> mount -t debugfs none /d
Jan 27 10:53:49 <morphis> then check within /d/kgsl
Jan 27 10:53:54 <morphis> you will find a folder per pid
Jan 27 10:54:05 <morphis> if you look into the one from our WebAppManager
Jan 27 10:54:19 <morphis> and take a shot without any windows and with a window
Jan 27 10:54:23 <morphis> memory seems to be released
Jan 27 10:54:47 <morphis> also that we get messages like
Jan 27 10:54:47 <morphis> [ 151.340423] ion_carveout_allocate: heap sf doesn't have enough memory (ac000) the allocation of size 180000 failed.
Jan 27 10:55:02 <morphis> it lets me think that something with the kernel ION allocator wents wrong
Jan 27 11:05:56 <Garfonso> DougReeder: the db8 watch *should* fire automatically. But you can also call palm://com.palm.service.contacts.linker/forceAutolink manually.
Jan 27 11:11:25 <Garfonso> morphis: did not push my local commits, this should do it: https://github.com/webOS-ports/org.webosports.service.contacts.carddav/pull/65
Jan 27 11:12:47 <Garfonso> morphis: about 837 ... uhm.. what? ;) Actually the contacts object is created by loadable-frameworks... would expect more issues with events. ;) How do you get this log output? And what are the consequences of it?
Jan 27 11:15:28 <morphis> Garfonso: no sure, just saw it in the logs
Jan 27 11:15:52 <Garfonso> ok. Will investigate a bit.
Jan 27 11:16:25 <morphis> Garfonso: merged
Jan 27 12:07:54 <Tofe> morphis: I'm at work so it will have to wait a little bit
Jan 27 12:08:32 <Tofe> I just wanted to know if we had other objects of that size that were allocated when creating a child window, in addition to the GPU content
Jan 27 12:28:27 <Garfonso> contactlinker does not seem to work after first launch. After reboot it did its job.
Jan 27 12:29:33 <Garfonso> maybe an issue with db8 watches or activities in general?
Jan 27 12:49:48 <morphis> Tofe: not that I know
Jan 27 12:49:57 <morphis> Tofe: but thanks that you're going to take a look too
Jan 27 12:50:01 <morphis> it's rather frustrating
**** BEGIN LOGGING AT Tue Jan 27 16:12:14 2015
Jan 27 19:09:51 <Herrie> morphis: DockMode works nicely!
Jan 27 19:09:56 <Herrie> I think we need the padlock on it though
Jan 27 19:10:02 <Herrie> Let me take some screenshots from legacy
Jan 27 19:14:20 <morphis> Herrie: that was the point I was unsure
Jan 27 19:14:38 <morphis> but the display manager switches from on -> dockmode -> locked -> on
Jan 27 19:14:50 <morphis> so we should only display dockmode when we're in the right state
Jan 27 19:21:11 <Herrie> Basically it works OK now, it would just need to show the lock as well I think, let me find some screenshots
Jan 27 19:21:23 <Herrie> To what extend is com.palm.wan supposed to work?
Jan 27 19:21:31 <Herrie> Jan 27 19:10:46 mako LunaWebAppManager[1481]: file:///usr/palm/applications/com.palm.systemui/data/TelephonyService.js:238: JS ERROR: ReferenceError: Can't find variable: causeCode
Jan 27 19:21:57 <Herrie> causeCode is not defined as a var, can add that and that will most likely "solve" the error..
Jan 27 19:24:25 <Herrie> At least not for that function, it's in another. Should we define as global var or define twice?
Jan 27 19:24:49 <Herrie> https://github.com/webOS-ports/luna-systemui/blob/master/data/TelephonyService.js#L177
Jan 27 20:21:44 <Herrie|2> morphis: SMS doesn't seem to go out here yet... This is in the log that might give some clues? http://pastebin.com/aex2Vje7
Jan 27 21:36:13 <morphis> Herrie|2: if the SMS uses an array for the to field then it will not work yet
Jan 27 21:36:20 <morphis> fix for telephonyd is still pending
Jan 27 21:44:50 <Herrie|2> morphis: Ah....
Jan 27 21:44:52 <Herrie|2> OK ;)
Jan 27 21:44:59 <Herrie|2> I thought you fixed that one already?
Jan 27 21:45:03 <Herrie|2> Or still testing?
Jan 27 21:49:31 <Herrie|2> Ah that one wasn't merged yet ;)
Jan 27 21:49:42 <Herrie|2> Still I only have 1 phone number in my msg
Jan 27 21:50:01 <Herrie|2> But could be the way it's stored in DB that causese telephonyd not to send it ;)
Jan 27 21:56:32 <Herrie|2> morphis: https://github.com/webOS-ports/luna-systemui/pull/4
Jan 27 21:56:47 <Herrie|2> Looked at legacy and added missing bits + modified it
Jan 27 22:46:44 <djichthys> morphis: ping  - ichthys here with a new name
Jan 27 23:17:52 <Garfonso> Herrie|2: latest messaging service should support objects for the to field... so if you alter messaging app to use the object in the to-field directly, then sending might work.
Jan 27 23:19:40 <Garfonso> Herrie|2: this probably is a deeper bug. I'd say either there is no causeCode at all in this function and should not be checked or causeCode is a member inRespone or so.
Jan 27 23:32:13 <Garfonso> morphis: with latest build MoboReader can't authorize anymore. When title changes I get the message "Did not find the right frame" in web inspector log. Journalctl only shows the "open" call as last output from LunaWebAppManager. Help. :-/
Jan 27 23:37:53 <Garfonso> morphis: hm.. window.name is empty for my root window...? Can fix that. But it's a plain enyo 2 app... so maybe we should have a fallback here.
Jan 27 23:41:34 <Garfonso> morphis: hm.. in window.top.frames is just one frame and it has some enyo 2 name. Ok.. I don't really understand this now. Better go to bed and think about it tomorrow. ;)
**** ENDING LOGGING AT Wed Jan 28 02:59:59 2015