**** BEGIN LOGGING AT Sun Nov 01 02:59:58 2015 Nov 01 05:06:49 morning Nov 01 05:12:16 Somebody here decided 5:45 was a nice time to get up on a Sunday :P Nov 01 05:12:36 Tofe: BTW great work on the rest, all seems to be fixed (i.e. power menu, height, email app, banners etc) :D Nov 01 05:25:29 morning Nov 01 05:25:52 somebody decided a train at 7:50 was a great idea... Nov 01 05:36:45 Herrie: when the change of url is done without any user interaction, I don't thnk we get much feedback apart from urlChanged Nov 01 06:02:49 TOfe: LOL ;) Nov 01 06:03:12 Little ones are harder to "train" at this age :P Nov 01 06:19:30 Tofe: I'll put latest nightly on my TP today and stress test with 100+ apps to see how things are working Nov 01 06:19:35 And if I see odd things in logs Nov 01 07:43:01 Herrie: if you have reproducible scenario of crashes, I'll take it Nov 01 07:44:22 I didn't investigate cross-domain XHR request yet Nov 01 07:52:31 Tofe: In Testr the HTML5 notifications will cause a hard crash Nov 01 07:55:40 Another scenario: Try to add an email account from Email app. Hard crash of the email app too Nov 01 07:55:59 But at least gets back to launcher. When you have Preware open, open Email, try to add account, launcher is gone too :P Nov 01 07:57:23 And after a crash the power menu is broken too (IE doesn't show anymore and N4 will just turn off) Nov 01 07:57:43 But this is what we had before with WebKit already too when luna-webappmgr would crash Nov 01 07:58:34 It could be that the (noWindow) apps that get loaded on initial launch of luna-webappmanager don't get loaded anymore after a crash when luna-webappmanager is restarted Nov 01 07:59:07 Best would be to make sure it doesn't crash of course, but when it does it would be good to check if these apps are still running when it's launched again and if not, run them Nov 01 07:59:14 I.e. Email and systemui Nov 01 08:02:59 I'm not sure how you could achieve a check in linux to see if a process is still running and if not start it Nov 01 08:11:02 Seems we might be able to do this with inittab. But I'm sure ka6sox or morphis might have a better solution maybe? Nov 01 08:12:14 Seems it can also be done in upstart & systemd Nov 01 08:12:42 This one gives examples for mysql but I see no reason why we couldn't do that for luna-webappmanager? https://www.digitalocean.com/community/tutorials/how-to-configure-a-linux-service-to-start-automatically-after-a-crash-or-reboot-part-1-practical-examples Nov 01 08:16:12 Seems we already have some casese where we do it with systemd Nov 01 08:17:23 https://github.com/webOS-ports/meta-webos-ports/blob/fido/meta-luneos/recipes-multimedia/pulseaudio/pulseaudio_6.0.bbappend#L18 and https://github.com/webOS-ports/meta-webos-ports/blob/fido/meta-luneos/recipes-qt/maliit/maliit-framework-qt5_git.bbappend#L14 Nov 01 09:10:37 We also have these of course: https://github.com/webOS-ports/webos-systemd-services Nov 01 09:13:57 Hmmz and appearantly we already have .service file in /lib/systemd/system/LunaWebAppManager.service that should force it to restart Nov 01 09:14:14 So it could be simply a case of making sure the apps that launch at boot are still running if not run them Nov 01 10:49:56 morning Nov 01 10:50:35 if there are multiple messages, do we want to have multiple notifications? (Still not sure about the wording) Nov 01 11:05:51 Garfonso: morning Nov 01 11:06:05 Garfonso: In legacy it has 1 for each arriving message Nov 01 11:06:39 P.s. Since we moved to QTWebEngine we get some funny number with Google C+DAV account name instead of xxxx@gmail.com Nov 01 11:07:12 it has a banner for each message, but there is only one notification that has a count in its icon or something like that. Nov 01 11:07:23 hm.. then probably the name retrival does not work anymore. Nov 01 11:07:40 Garfonso: Yeah see 2nd part in this log: https://bpaste.net/show/da700a0ce5f8 Nov 01 11:07:40 the number is Date.now(), i.e. system time as linux timestamp. ;) Nov 01 11:08:02 You think it's the Oct 31 16:22:42 mako LunaWebAppManager[2242]: WARNING: 16:22:42.474: CONSOLE JS: XMLHttpRequest cannot load https://accounts.google.com/o/oauth2/token. No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin 'file://' is therefore not allowed access. or the one after it? Nov 01 11:08:15 That's probably osmething Tofe still would need to fix ;) Nov 01 11:27:27 that is probably a configuration for QTWebEngine somewhere... Nov 01 11:27:57 you get the same issues, if you run an local app in chrome for example... there are some command line parameters to fix that in the case of chrome Nov 01 11:29:19 IIRC --disable-web-security was necessary for file:// to access web-apis. Nov 01 11:30:50 Garfonso: We thought we had that covered with: https://github.com/webOS-ports/luneos-components/blob/master/modules/LuneOS/Components/LunaWebEngineView.qml#l53 Nov 01 11:30:56 But seems we might need something more ;) Nov 01 11:32:51 yeah, that's another switch.. it reads like it allows app filesystem access (not sure if that is restricted further... so might be a security issue?). There is a parameter for that in chrome browser, too. Nov 01 11:35:23 Well we also have https://github.com/webOS-ports/luneos-components/blob/master/modules/LuneOS/Components/LunaWebEngineView.qml#l55 Nov 01 11:35:38 That should work I'd say but could be there's something missing in QtWebEngine still Nov 01 11:36:01 There were a few things Tofe had to backport from 5.6 alpha and also needed to simply implement because it wasn't there yet Nov 01 11:49:03 is qtwebengine already in the nightlies? Nov 01 11:51:13 Garfonso: Yes Nov 01 11:51:20 It's working pretty well Nov 01 11:51:28 Just a few minor bugs to fix Nov 01 11:52:16 Browser app was migrated last month, redering of Enyo 1/2 apps this week Nov 01 11:55:59 Not sure if you knew but Galaxy Nexus is back working too ;) Nov 01 11:59:15 yeah, I heard that.. not really dug into it, yet. Currently the Galaxy Nexus is running android here and acting as Wifi Phone. ;) Nov 01 12:01:36 Garfonso: Ah ok :) Nov 01 12:01:40 When is the baby due? Nov 01 12:16:25 14th of November Nov 01 12:18:07 :) Nov 01 12:18:41 All ready now? Well at least as much as you can be :P I know from experience it's difficult to imagine what will be ;) Nov 01 12:29:20 yeah... all ready, at least as far as we have planned... not sure what will be. Nov 01 13:14:15 It'll be OK ;) FIrst couple of days are strange, then you'll get used to it ;) You'll just miss your time alone together ;) Nov 01 13:14:38 Garfonso: BTW you remember why the option to select multiple apps at once in Preware 2 was disabled? Nov 01 15:08:02 Herrie: no clue.. maybe it was never implemented? Nov 01 15:09:28 I see functions for multiple app install, but seems it's not hooked up in UI Nov 01 15:09:47 Which is quite a pain when you want to stress test 100+ apps and need to install them 1 by 1 Nov 01 15:12:50 the functions for multiple app install are there, if an app requires other packages. Then thouse *should* be installed automatically. It is right, that this never was wired to the UI in preware 2. Nov 01 15:13:22 but a multi install for independent app was never there, I guess. Nov 01 15:17:41 Ah ok :) Nov 01 15:21:44 hm... I get a strange error message, when trying to add the notification: "Error: an application id or a service id must be set on the notification message." Nov 01 15:26:49 What is your syntax? Nov 01 15:26:51 luna-send -m "org.webosports.service.messaging" (for setting the service id) doesn't work, either. Only -a (which is appId). Nov 01 15:27:17 You need to put it as one of the parameters in the luna-send Nov 01 15:28:46 see here: https://bpaste.net/show/23c4a3f24aa2 Nov 01 15:31:40 You might want to use a luna instead of palm Nov 01 15:32:33 that does not matter. Nov 01 15:32:50 the first command does not work, the second does work. The only difference is -a instead of -m. Nov 01 15:33:36 That means the notification service can read the app id (-a) but somehow has issues with the serviceId (-m). Maybe it is not "serviceId" as expected here https://github.com/webOS-ports/luna-next-cardshell/blob/master/qml/LunaSysAPI/NotificationService.qml#L87 Nov 01 15:38:17 Interesting... something else seems to filter out service id. Nov 01 15:53:02 experimented a bit more: https://bpaste.net/show/c1ff72fc4f6b Nov 01 16:13:59 Herrie: notifications in messaging service: https://github.com/webOS-ports/org.webosports.messaging/pull/31 - but currently not working, because of the bug... Nov 01 16:36:47 Garfonso: Cannot I didn't even know -m existed Nov 01 16:36:50 Only used -a ever Nov 01 16:38:02 -m sets serviceid and not appid, so it is what a node.js service would do. Nov 01 16:38:25 I had one case where I needed to set -m for a specific reason (can not remember, though). So I learned about that. Nov 01 16:41:27 not sure who wrote the code to process dbus in qml. Maybe Tofe or morphis can have a look at http://issues.webos-ports.org/issues/1007 Nov 01 16:44:07 Can't you use -a or that won't work? Nov 01 16:44:21 I know it's not proper, but as a work around or it wouldn't allow that? Nov 01 16:47:43 the service always uses "-m". That is what the framework does. Nov 01 16:48:57 Not sure if that is easy to hack or not... but it possibly has quite some side effects, because it would affect all node.js services. Nov 01 16:51:09 I think they both worked on it Nov 01 16:51:14 It's probably a small fix Nov 01 16:51:21 I think they also might not know about -m ;) Nov 01 16:56:47 at least the one who build the notification service was aware that there might be a serviceId instead of applicationId. Not sure why it is not received, though... maybe luna-send and node.js does something different than what luna-next expects. Nov 01 17:05:12 Garfonso: I'm completely lost on the notification implementation ;) Nov 01 17:05:28 How it all flows from Enyo to luna-next and luna-next-cardshell :P Nov 01 17:06:23 This seems to suggest that there's only an ownerId which I guess is -a? https://github.com/webOS-ports/luna-next/blob/master/plugins/shell/notifications/notification.cpp#L23 Nov 01 17:07:02 we need simply to add serviceId/serviceName here: https://github.com/webOS-ports/luneos-components/blob/master/modules/LuneOS/Service/plugin/lunaserviceadapter.h#L34-L41 Nov 01 17:09:19 and use LSMessageGetSenderServiceName like https://github.com/webOS-ports/luneos-components/blob/master/modules/LuneOS/Service/plugin/lunaserviceadapter.cpp#L87-L90 Nov 01 17:13:29 nizovn: You want me to quickly mock up something and PR so you can review? Nov 01 17:14:10 i can review if needed :) Nov 01 17:20:54 https://github.com/webOS-ports/luneos-components/commit/0c52dc4cec8a70f52237bbbc2ef73b876e9cffcb Nov 01 17:21:09 Seems I did to the ports branch instead of my own fork :S Nov 01 17:22:49 you should use LSMessageGetSenderServiceName instead LSMessageGetServiceID Nov 01 17:26:45 https://bpaste.net/show/ec09fd331e96 ? Nov 01 17:28:15 yes Nov 01 17:32:26 OK kicking off a new build with this change & Garfonso's. Let's see if it works :P Nov 01 17:36:08 Herrie: Where do you guys usually store device-specific modifications? meta-smartphone? Nov 01 18:29:29 Andolamin: Let me check Nov 01 18:33:22 Seems meta-smartfone as per http://webos-ports.org/wiki/Porting_Guide#Creating_an_OpenEmbedded_machine_layer ? Nov 01 18:34:10 Though it seems there's no Fido branch :S Nov 01 18:34:16 JaMa: ^ Any ideas? Nov 01 18:34:53 Ah seems JaMa is not around... Nov 01 18:34:59 morphis: ^ Nov 01 18:36:27 Ah I guess we use the one frm shr and not our own fork Nov 01 18:37:11 Seems we do: https://github.com/webOS-ports/webos-ports-setup/blob/testing/conf/layers.txt#L5 Nov 01 18:38:48 JaMa manages that one I think Nov 01 19:01:53 Herrie: That's what I was thinking, but I'm not sure that meta-smartphone would be the best place to store Raspberry Pi stuff. Plus, there's the whole question of what to do with the meta-raspberrypi layer. Nov 01 19:11:44 I thought you said that the meta-raspberrypi could be used from upstream now? Nov 01 19:11:56 We can link to that directly if needed I guess in our build Nov 01 19:17:35 Right Nov 01 19:17:59 We need to override the machine config to bring in the touchscreen overlay but it seems to work correctly with upstream now. Nov 01 19:18:48 I guess we'll just have to add it to layers.txt. Seems silly to clone it for every machine, but I guess that's what happens for meta-smartphone right now anyway Nov 01 19:25:11 I'll ping JaMa about it when he's around. He's in Czech so same timezone with me in Amsterdam ;) Nov 01 19:25:22 At least when he's not over in the US @ LG SVL :P Nov 01 19:26:23 Heh, I'm sure there are people who wish he was in the office with us, though. Nov 01 19:28:20 I'm planning on openning a PR today for some of the raspberrypi configuration files. I can't get /etc/palm/luna-platform.conf or /etc/luna-next/environment.conf to get picked up in my build, but those are both things that can be pulled into the official layers without impacting anything else. Nov 01 19:30:18 Nevermind, apparently /etc/luna-next/environment.conf WAS picked up in this build. Nov 01 19:30:32 still missing luna-platform.conf, though. Not sure what I'm doing wrong there. Nov 01 20:18:05 Herrie: What does QT_COMPOSITOR_NEGATE_INVERTED_Y do? Nov 01 20:34:23 Andolamin: Tofe would know that one Nov 01 20:37:28 I'd say that's when coordinates aren't working properly. Some Google search gave " We need inverted-Y for Qt5 apps runing on Wayland platform". So as long as you don't have issues in QML apps you don't need it I guess Nov 01 20:40:10 Yeah - trying to work out a way to fix this upstream meta-raspberrypi issue. Not sure why, but display is rotated 180 when I use upstream. I can rotate it back in the raspberrypi config, but that only rotates the display, not the touchscreen coordinates, so they're all off. Nov 01 21:17:08 Fixed it. Upstream looks good now. Nov 01 21:41:07 Andolamin: Nice :D Nov 01 21:41:44 Herrie: Indeed. Resuming trying to fix pulse audio now. Nov 01 21:43:52 How necessary is SDL support? Nov 01 21:54:02 SDL? Nov 01 21:57:24 Tofe: https://github.com/webOS-ports/luna-webappmanager/blob/e55c0d7e4d1abedf9b0837d1fda530ffefe8f060/src/qml/ApplicationContainer.qml#L195 That might be the culprit somehow for Macaw? Nov 01 21:58:01 Herrie: Native app development library: https://www.libsdl.org/ Nov 01 21:58:54 Seeing that https://github.com/webOS-ports/luna-webappmanager/blob/675f9d956968f20f585da2da034c8d1f719aa6cd/src/webapplication.cpp#l379 will only return true for https://github.com/webOS-ports/luna-webappmanager/blob/675f9d956968f20f585da2da034c8d1f719aa6cd/src/webapplication.cpp#l130 Nov 01 21:59:01 Which macaw is not of course Nov 01 22:00:05 Andolamin: Not sure tbh, Tofe or Morphis might know. It's not working with the R Pi or... ? Nov 01 22:00:45 Yeah - not working on the Pi Nov 01 22:01:27 As far as I can tell the LuneOS image only uses it for a couple of test apps. I suppose it's only need for backwards compatibility with some of the old PDK apps, maybe? Nov 01 22:03:02 Andolamin: We don't support PDK apps currently AFAIK Nov 01 22:03:37 Makes sense. Probably missing the PDK APIs there. Nov 01 22:03:38 ONly Enyo 1/2 + QT (QML / C++) and some OpenGLES stuff Nov 01 22:03:48 Andolamin: Unfortunately yes Nov 01 22:03:48 I don't think SDL is necessary, honestly. Nov 01 22:04:02 Andolamin: Well would be nice for OpenGLES stuff Nov 01 22:04:15 But if Pi doesn't have it you could leave it our Nov 01 22:04:23 LIke phone app on tablets Nov 01 22:04:36 Yeah, it's something that should work better as the VC4 drivers mature Nov 01 22:05:02 What's the best way to test audio? Does audio in the browser work? Nov 01 22:05:18 Andolamin: It doens't in YouTube Nov 01 22:05:29 Herrie: your analysis seems correct Nov 01 22:05:31 It did before, not sure what broke it (QT 5.5 or webengine) Nov 01 22:05:41 Volume up/down I use normally Nov 01 22:05:47 You can use the slider in Settings for that I guess Nov 01 22:07:39 Herrie: but then, it would mean we allow cross-domain for all system apps ? What was this test for ? Nov 01 22:08:06 (knowing that system means "file:///" in our case) Nov 01 22:10:24 gotta get some sleep, we'll discuss that tomorrow :) Nov 01 22:10:39 gn8 all! Nov 01 22:14:31 Tofe: mPriveliged should be for private bus access which is OK with com.palm, org.webosports and org.webosinternals Nov 01 22:14:50 But I don't think the cross domain should be restricted to those 3 Nov 01 22:15:32 Anyway off to bed here too was 5:45 this morning LOL Nov 01 22:15:36 Gn8 Nov 01 22:16:05 Herrie: Allowing cross domain behavior extended to all apps in legacy webOS Nov 01 22:24:28 I have audio! Nov 01 22:25:06 Bit of a delay, but it plays Nov 01 22:29:09 No idea how to integrate it into the meta layer, but at least it works **** ENDING LOGGING AT Mon Nov 02 02:59:59 2015