**** BEGIN LOGGING AT Sun Aug 23 02:59:58 2015
Aug 23 08:47:55 <Tofe> Morning
Aug 23 08:48:27 <Tofe> Herrie: I saw you opened a bug on dashboard... 3 hours ago ! Looks like little one was awake, eh ;)
Aug 23 10:11:57 <Herrie> Tofe: Yeah he woke up at 7
Aug 23 10:12:07 <Herrie> The bug was there already just not assigned to anyone ;)
Aug 23 10:12:09 <Herrie> Assigned it to you
Aug 23 10:50:52 <Herrie> Tofe: I'm not 100% sure the problem lays with LNC or with the code for dashboards itself tbh
Aug 23 10:59:21 <Herrie> Or a combination of both
Aug 23 11:03:03 <Herrie|2> morphis tried to implement some resolution independent solution to Enyo as well similar to the Units.gu we have in QML, but not sure that's 100% working correctly
Aug 23 11:05:14 <nizovn> DougReeder: i found scenario how to get webappmanager crash with filepicker:
Aug 23 11:05:14 <nizovn> open Serene Notes, import some text from file
Aug 23 11:05:14 <nizovn> close the app, open again, note imported text disappeared
Aug 23 11:05:14 <nizovn> try to import text file -> crash
Aug 23 11:07:06 <Herrie> nizovn: I'm not sure what would cause this, anything useful in journalctl?
Aug 23 11:07:22 <Herrie> I suspect there's still some bugs with the way windows are created and destroyed
Aug 23 11:07:30 <nizovn> gdb shows it's related to invalid cache
Aug 23 11:09:11 <Herrie> I have sometimes apps not appearing, even though logs show they're opening. The power menu sometimes doesn't appear
Aug 23 11:09:25 <Herrie> The Settings bug getting white after Maps yesterday
Aug 23 11:09:37 <Herrie> Then there's the browser 2nd card as well
Aug 23 11:09:44 <Herrie> They might all be related somehow
Aug 23 11:10:32 <nizovn> yeah, closing of card is broken definitely
Aug 23 11:15:58 <Herrie> It's probably a few lines of code to fix, but to debug is not easy
Aug 23 14:48:44 <Tofe> Herrie: ping
Aug 23 14:56:54 <Herrie|Veer> Tofe: Pong
Aug 23 14:57:13 <Herrie|Veer> In metro so connection might drop :P But will pick up from log then.
Aug 23 14:58:48 <Tofe> Herrie|Veer: for you, tapping on the dashboard doesn't hide it?
Aug 23 15:00:25 <Herrie|Veer> Tofe I think it does but it's display doesn't work well. Let me try all from Testr on N4 and TP with latest nightly in about 30 mins and tell you results.
Aug 23 15:01:34 <Tofe> I agree the display is cut
Aug 23 15:02:08 <Tofe> I thought the main issue was that we can't dismiss it; but it the issue is the display itself, then I can reproduce it
Aug 23 15:03:41 <Herrie|Veer> I think there are different kinds. I.e. For example the lower power (<5 %) on N4 misses the OK button, while on 3.x TP it shows (also on LuneOS I think but haven't tested for a while should drain TP :P). Also I noticed that the dashboards with longer text
Aug 23 15:03:41 <Herrie|Veer>  don't display properly. I.e. it doesn't wrap onto 2nd line. Serene notes does one of those.
Aug 23 15:04:17 <Herrie|Veer> So I think that the missing OK button might be the case in the issue you're looking at but not sure.
Aug 23 15:05:21 <Herrie|Veer> When the OK button is not visible, you cannot press it and therefore not dismiss. This is for sure the case with the <5% on N4
Aug 23 15:05:31 <Tofe> ok, obviously we don't communicate well about the required/available size of the dasboard
Aug 23 15:06:31 <Herrie|Veer> morphis tried some flexible approach but it doesn't work properly yet imho. Also the power off dialog on TP is an example of that. It's fine on N4 but not on TP
Aug 23 15:06:32 <Tofe> If I understand correctly, for that kind of window, the size is explicitely requested by the client, and no compromise can be made by the compositor
Aug 23 15:07:19 <Tofe> the issue being (maybe among other issues) that the size might be DPI dependent, hence the morphis approach to solving this
Aug 23 15:08:20 <Tofe> Do you know where we do the call on Testr for showing the dashboards with different heights ?
Aug 23 15:08:27 <Herrie|Veer> Tofe: Enyo 1 in legacy only allowed for px which is not workable with multiple targets with different resolutions and DPI. An approach similar to Units.gu should be used but the current one is buggy.
Aug 23 15:09:18 <Tofe> exactly, we should fix current one, I think we're on right track and it's just a matter of finding where we introduced a mistake
Aug 23 15:09:36 <Herrie|Veer> Tofe: It's in the source code. DougReeders code is OK to read. Not 100% sure where we deal with the height when processing. Either LNC or maybe luna-systemui?
Aug 23 15:10:18 <Herrie|Veer> Tofe: Agree. Once I'm home I run all tests on N4 and TP and see which are ok.
Aug 23 15:10:29 <Herrie|Veer> And dig for morphis code on the dashboards etc
Aug 23 15:10:39 <Herrie|Veer> Should be able to locate in the logs
Aug 23 15:11:04 <Tofe> Could be in luna-webappmanager, in the WebView scripts
Aug 23 15:11:45 <Herrie|Veer> Not sure. I think it was elsewhere but you might be right.
Aug 23 15:13:34 <Tofe> no, now that I see Testr code I think you were right; it should be in LNC, when reading the window propertiezs
Aug 23 15:13:54 <Herrie|Veer> I think I have found the problem for the charging alerts not working properly when on charger. Fix will be easy :)
Aug 23 15:14:02 <Tofe> I thought that we used some scripting to compute the height, but actually we use a window property
Aug 23 15:14:07 <Herrie|Veer> Will make a PR for that tonight :)
Aug 23 15:14:11 <Tofe> ok great!
Aug 23 15:14:42 <Herrie|Veer> This is in luna-systemui ;) Nothing too shocking but was annoying me :P
Aug 23 15:23:24 <DougReeder>   nizovn, thanks for finding that scenario - I haven’t been seeing many Luna crashes with the file input lately.
Aug 23 15:24:14 <DougReeder> There may be a correlation with reboot or app installation. :-S
Aug 23 15:26:05 <DougReeder> I havent turned up anything useful using journalctl yet.
Aug 23 15:26:34 <DougReeder> I need to tear up my deck more today; may not be back till late tonight.
Aug 23 15:29:15 <Herrie> DougReeder: Good luck with that :D
Aug 23 15:29:37 <Herrie> Happy I got all the heavy lifting done for the next couple of years at least ;) Just some small things to do still ;)
Aug 23 15:29:43 <Herrie> Tofe: Home now ;)
Aug 23 17:52:00 <Herrie> Tofe: Ping
Aug 23 17:52:07 <Herrie> or morphis: ping
Aug 23 18:12:14 <Tofe> Herrie: pong
Aug 23 18:13:56 <Herrie|Laptop> Tofe: I was debugging luna-systemui
Aug 23 18:14:55 <Tofe> yup
Aug 23 18:15:12 <Herrie|Laptop> Found that the charging status doesn't get reported correctly. Touchstone for example doesn't report DockConnected and DockPower as true while it should. Also when you charge with proper charger, USBConnected seems to be set to false and USBName set to wall, while charging is set to true
Aug 23 18:15:41 <Herrie|Laptop> The only references to DockConnected I could find were in LunaSysMgr and Powerd
Aug 23 18:15:51 <Herrie|Laptop> So I guess powerd is the culprit: https://github.com/openwebos/powerd/blob/f211f541617a6c86f080c9f717df48f27e0a19ab/powerd/charging/charger.c
Aug 23 18:18:23 <Herrie|Laptop> Not sure what would be the best way to fix this. I compared the luna-systemui/data/PowerDservice.js between OWO and 3.0.5. Seems that HP/LG introduced some additional code and that's a bit buggy anyway.
Aug 23 18:18:41 <Herrie|Laptop> I wonder if we should fix that or simply revert to the 3.0.5 code which we knew worked?
Aug 23 18:19:41 <Tofe> Well, at least it looks like the status isn't correct
Aug 23 18:19:47 <Herrie|Laptop> Tofe: Yeah
Aug 23 18:19:51 <Tofe> I don't if the implementation of nyx_charger_query_charger_status has changed
Aug 23 18:20:50 <Herrie|Laptop> Tofe: Nyx was implemented in OWO as HP's version of libhybris more or less
Aug 23 18:20:57 <Herrie|Laptop> Or at least that's how I understood it
Aug 23 18:21:08 <Herrie|Laptop> It didn't exist in legacy 3.x for sure
Aug 23 18:21:36 <Tofe> ah ok, that's quite a big change
Aug 23 18:21:46 <Herrie|Laptop> http://www.webosnation.com/open-webos-gets-website-all-its-own-nyx-hardware-abstraction-layer-gets-release
Aug 23 18:21:53 <Herrie|Laptop> Gives some background on Nyx
Aug 23 18:22:05 <Herrie|Laptop> "Nyx is the webOS portability layer used to isolate the remainder of webOS from dependencies on the hardware and core OS upon which it is running." Essentially what that means is that Nyx picks up where the Linux Standard Kernel drops off.
Aug 23 18:24:25 <Tofe> yes, I understand what it is, but I thought it was there for quite some time in WebOS
Aug 23 18:24:43 <Herrie|Laptop> Nope it was introduced in Open webOS
Aug 23 18:24:58 <Tofe> ok
Aug 23 18:25:13 <Tofe> but we still use it partially, isn't it ?
Aug 23 18:25:17 <Herrie|Laptop> These are the diffs between our version of the PowerdService.js and the 3.0.5 version ;) https://github.com/webOS-ports/luna-systemui/compare/webOS-ports/master...Herrie82-patch-1?quick_pull=1
Aug 23 18:25:21 <Herrie|Laptop> Tofe: Yeah we use it
Aug 23 18:26:49 <Herrie|Laptop> So ideally we make sure NYX reports the correct status for the Touchstone
Aug 23 18:27:26 <Tofe> Looking at https://github.com/openwebos/nyx-modules/blob/master/src/rockhopper/charger/device/charger.c it looks like they dropped the DOCK things
Aug 23 18:28:27 <Herrie|Laptop> Dock is touchstone and therefore TP specific for now
Aug 23 18:28:50 <Herrie|Laptop> But could in future be recycled for inductive charging devices as well
Aug 23 18:29:00 <Herrie|Laptop> It seems it's getting more mainstream little by little
Aug 23 18:29:21 <Herrie|Laptop> Some of the nexus & Lumia devices offer wireless charging with a Touchstone alike device
Aug 23 18:30:00 <Herrie|Laptop> Likely rockhopper didn't have inductive charging
Aug 23 18:30:16 <Herrie|Laptop> Rockhopper was some x86 device if I'm correct?
Aug 23 18:30:39 <Herrie|Laptop> nizovn: ping
Aug 23 18:31:52 <Herrie|Laptop> I see references to inductivepowered in https://github.com/openwebos/nyx-lib/blob/master/include/public/nyx/common/nyx_charger_common.h
Aug 23 18:32:55 <Herrie|Laptop> But it doesn't appear in https://github.com/openwebos/nyx-lib/blob/master/src/device/nyx_charger_impl.c
Aug 23 18:35:50 <Herrie|Laptop> I believe nizovn has played with nyx before
Aug 23 18:36:29 <nizovn> Herrie|Laptop: pong
Aug 23 18:36:37 <Herrie|Laptop> nizovn: You played with nyx before?
Aug 23 18:36:45 <nizovn> a little
Aug 23 18:37:02 <nizovn> vibration module
Aug 23 18:42:26 <Herrie|Laptop> You happen to have a touchstone for TP?
Aug 23 18:44:04 * Tofe gives up the debug of multiple browser window crash after a fierce battle of 4 hours. Wayland 1 - 0 Tofe.
Aug 23 18:46:40 <nizovn> Herrie|Laptop: no
Aug 23 18:57:08 <Herrie|Laptop> Tofe: :'(
Aug 23 18:57:38 <Herrie|Laptop> nizovn: I guess troubleshooting the Touchstone charger would prove challenging then ?
Aug 23 18:58:19 <Herrie|Laptop> Tofe: Maybe it's related to the open Settings, open Maps, close maps one. That one is easy to replicate?
Aug 23 18:59:07 <nizovn> what's wrong with touchstone charging?
Aug 23 18:59:58 <Herrie|Laptop> nizovn: It charges, but it doesn't report DockConnected and DockCharging correctly
Aug 23 19:00:30 <Herrie|Laptop> Sorry it's DockPower not DockCharging
Aug 23 19:00:44 <Herrie|Laptop> It says both as false while they should be true when it's charging ;)
Aug 23 19:01:19 <nizovn> i guess it was reported by special chip on legacy
Aug 23 19:01:39 <nizovn> ? not sure cm have support for it
Aug 23 19:02:48 <nizovn> ah no maybe i'm wrong
Aug 23 19:04:51 <Tofe> Herrie|Laptop: as far I as I can see, it's some resource that is not released when the first window is closed
Aug 23 19:05:23 <Tofe> It only happens in the context of multiple windows, because otherwise the whole client is destroyed and we don't see the issue
Aug 23 19:06:27 <Tofe> When the last window is closed, wayland cleans up a bit more and stumbles on the resources that are still there but attached to a window that has been destroyed
Aug 23 19:13:45 <Herrie|Laptop> nizovn: It was something called A6
Aug 23 19:13:57 <Herrie|Laptop> It works for charging on Android too
Aug 23 19:14:52 <Tofe> ( as a reminder for myself, here is the interesting log http://pastebin.com/0Uu7kBtQ )
**** ENDING LOGGING AT Mon Aug 24 02:59:59 2015