**** BEGIN LOGGING AT Mon Jun 08 02:59:58 2020 Jun 08 06:31:03 Morning ! Jun 08 06:38:50 https://github.com/search?q=org%3AwebOS-ports+vboxtouch&type=Commits the ones ine april are about this mosue issue Jun 08 06:38:53 mouse* Jun 08 07:56:26 Morning! Jun 08 08:02:43 Tofe: Merged the phone app PR. The audio-service needs adding a line. Can you take care of that? Afterwards OK to merge as well Jun 08 08:02:57 For the logging changes would like to get some feedback from JaMa Jun 08 08:06:44 Herrie: yes, I wasn't sure how to that the "right" way... removing an element from SRC_URI feels wrong, but I didn't see any other way so far Jun 08 08:09:18 Tofe: I'm not too familiar with all the logging mechanisms, JaMa might be more familiar with it from a Yocto point of view etc Jun 08 08:09:26 So let's see what he says. Jun 08 08:10:47 For audio-service I don't see how I missed those 2 API calls initially... Because I check the c-files normally... Maybe I missed it or had it in another iteration somewhere Jun 08 08:18:40 Well, it wasn't hard to fix anyway Jun 08 08:21:46 No ;) Jun 08 08:22:28 For ACG main thing remaining now would be the QML settings app recipe I guess Jun 08 08:22:41 I had an initial stab at it but not much luck I guess Jun 08 08:23:24 WIP is in: https://github.com/webOS-ports/org.webosports.app.settings/commit/be6af01496a1dd58ed78736491b3c46e5d23fd1d#diff-24724d5413fc03e0f537dac5d19dcf33 Jun 08 08:23:37 I guess the .in.in should still be called .in.in Jun 08 08:23:43 And not .in like I did Jun 08 08:24:32 I'll have another stab at this today I guess Jun 08 08:24:38 If stuck will ask for help ;) Jun 08 08:24:43 ok :) Jun 08 08:24:50 I'll fix the missing line right away Jun 08 08:28:14 Herrie: speaking of which, should I add it for all the service names ? Jun 08 08:28:54 it doesn't make sense to add it to old service names if it's not in the compat-api groups Jun 08 08:31:48 Ah OK Jun 08 08:31:56 Well we anyway only call it from our apps Jun 08 08:31:59 So we control it's calls Jun 08 08:32:08 There shouldn't be legacy names Jun 08 08:39:52 Phone app merged Jun 08 08:40:07 yep, saw that, thanks Jun 08 08:40:24 now it's just the logging stuff Jun 08 08:41:00 I'll bump SRCREV and add it to my meta-webos-ports PR Jun 08 08:41:51 Yes :) Jun 08 08:50:29 now, time to update my testing build Jun 08 09:26:02 Same here shortly finishing up a few other things for now ;) Jun 08 11:00:53 OK new build ready here Jun 08 11:01:00 Now will check QML settings Jun 08 12:23:48 Tofe: I'm seeing these also: 2020-06-08T12:19:19.453601Z [14.033883057] user.warning luna-next [] LS_INVALID_PAYLOAD {} It is ambiguous to have "subscribe" in one reply call.Service org.webosports.luna, uri luna://com.webos.service.systemservice/getPreferences Jun 08 12:23:48 2020-06-08T12:19:24.431051Z [19.011335106] user.warning luna-qml-launcher [] LS_INVALID_PAYLOAD {} It is ambiguous to have "subscribe" in one reply call.Service org.webosports.app.phone-857, uri luna://com.palm.systemservice/getPreferences Jun 08 12:23:58 Seems we're subscribing to services that only give 1 reply? Jun 08 12:24:40 yes, I guess we've put "subscribe: true" in the QML property were it wasn't needed Jun 08 12:24:45 where Jun 08 12:36:16 https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.8-Adreno-MSM-Driver damn, nothing for rosy, mido or tissot Jun 08 12:45:26 Tofe: Yeah Jun 08 12:45:41 I also noticed that there's an issue with installing apps via Preware. All seems to be OK, but then they don't appear Jun 08 12:45:54 Probably something small, but didn't find it yet Jun 08 12:48:09 Logs tell me: Jun 08 14:20:17 qemux86-64 ls-hubd[265]: [] [pmlog] ls-hubd LSHUB_NOT_LSTED {"SERVICE_NAME":"com.palm.deviceprofile","EXE":"/usr/sbin/org.webosports.service.ipkg","APP_ID":"com.palm.configurator","PID":641} Service not listed in service files (cmdline: /usr/sbin/org.webosports.service.ipkg) Jun 08 12:48:22 But I suspect that was there before, because we never had a com.palm.deviceprofile service on our images Jun 08 12:49:37 Herrie: isn't there a missing "luna-send" permission ? I think I saw soemthing like that... Jun 08 12:49:43 ^ ah no it was power menu Jun 08 12:49:46 Relevant bits from log: https://paste.ubuntu.com/p/fdbf693VVk/ Jun 08 12:50:19 Last few lines look weird... all seems to go OK and then fails Jun 08 12:51:32 I do see: Jun 08 14:20:34 qemux86-64 LunaWebAppManager[780]: CRITICAL: 14:20:34.409: Uncaught TypeError: Cannot read property 'pkg' of undefined Jun 08 12:51:50 Jun 08 14:20:34 qemux86-64 LunaWebAppManager[780]: [780:780:0608/142034.409930:INFO:CONSOLE(273)] "Uncaught TypeError: Cannot read property 'pkg' of undefined", source: file:///usr/palm/applications/org.webosports.app.preware/build/app.js (273) Jun 08 12:52:31 Ah it might help if we merge this one: https://github.com/webOS-ports/preware/pull/36/files Jun 08 13:24:59 Herrie: don't we already take the acg branch ? Jun 08 13:25:16 iirc you already PR'ed a lot of these branches Jun 08 13:50:36 Tofe: Ah seems you're right, wrong clue ;) Jun 08 13:50:42 Should we merge your meta-wop? Jun 08 13:50:48 Since JaMa seems happy with it Jun 08 14:13:14 I think that's fine yes Jun 08 14:13:37 Can you also merge pmlogdaemon ? Jun 08 14:14:04 then we can bump it and you'll get all the logs in the journal Jun 08 14:14:09 It does help :p Jun 08 14:37:40 Tofe: Done & done, also bumped SRCREV Jun 08 14:37:49 Checking the Settings-qml change now locally Jun 08 14:38:33 thanks Jun 08 14:55:01 Tofe: https://github.com/webOS-ports/org.webosports.app.settings/pull/106 Jun 08 14:57:26 Was easy enough in the end it seems ;) Jun 08 15:05:44 Tofe: Some more cosmetics: https://github.com/webOS-ports/luna-webappmanager/pull/88 Jun 08 15:17:21 Herrie: won't we miss the client-permission file ? I guess we don't rely on webos_app for this one ? Jun 08 15:18:00 Tofe: For Settings we should be OK, we define requiredPermissions Jun 08 15:18:11 client-permissions should be for services only if I'm correct Jun 08 15:18:21 Our QML apps don't define API's? Jun 08 15:18:25 no it's api for services Jun 08 15:18:38 client is for requesting permissions Jun 08 15:18:49 generated through webos_api in many app case Jun 08 15:19:27 It generates them ;) Jun 08 15:19:34 I have them at my end with my recipe Jun 08 15:19:58 ok good, but doesn't it also generate the role file ? I though it took care of all of them Jun 08 15:20:08 Ah wait maybe that's the JS version that doesn it Jun 08 15:20:39 I'm also a bit surprised it "detects" the "appinfo.something.json" Jun 08 15:20:52 I have org.webosports.app.settings.app.json for example Jun 08 15:21:27 Seems that comes from JS according to manifest Jun 08 15:22:05 For QML we don't have any client files Jun 08 15:22:26 ok so it doesn't see the appinfo Jun 08 15:22:36 which doesn't surprise me... Jun 08 15:22:48 we'll have to provide it Jun 08 15:23:11 This is my manifest for the QML version: https://paste.ubuntu.com/p/SX7pFDZtWy/ Jun 08 15:23:57 And I have role files for all the QML "Settings-apps" in the roles.d folder Jun 08 15:24:02 ok; so it just "sees" what is generated during the build Jun 08 15:24:08 Yes Jun 08 15:24:19 Then we need to generate the client files as well Jun 08 15:24:38 Hard to test on Qemu for now Jun 08 15:24:49 or find a way to get webos_app working with this settings apps, but not sure what would be the easiest approach Jun 08 15:25:34 I'd say, we know we need the client files, so when we do get them in the packaged files and the manifest also looks good, we're good to test on runtime Jun 08 15:25:54 I can have a stab at it tonight Jun 08 15:26:01 I'll start from your PR Jun 08 15:26:07 I don't think we need them really Jun 08 15:26:25 We don't have them for any other apps Jun 08 15:26:53 Ah wait we do Jun 08 15:26:57 Head still not great today Jun 08 15:27:10 I'll continue on top of your work then :) Jun 08 15:27:10 Better compared to yesterday, still not great Jun 08 15:27:51 I have some ideas on how to do it; also, the role files need to have the correct allowed names, we can't have "" anymore Jun 08 15:28:10 OK Jun 08 15:28:14 not sure what we use in the QML files though Jun 08 15:28:25 Well that should be easy enough in general Jun 08 15:28:35 yes, that's fine Jun 08 15:28:53 the client files will be a bit tedious to have in a generic way Jun 08 15:29:45 but maybe I can just put org.webosports.app.settings.* or so Jun 08 15:30:14 do we regroup all the permissions ?... Jun 08 15:31:54 Tofe: Not sure I understand your question Jun 08 15:31:58 that way we just would have one role file, one client file, maybe I'll just use one service name for all Jun 08 15:33:10 for instance if all the qml setting apps use "org.webosports.app.settings.qml-" for their LS2 name Jun 08 15:33:26 then it simplifies the permissions a lot Jun 08 15:33:49 but it also means we merge all the permissions in one group for "org.webosports.app.settings.qml-*" Jun 08 15:34:11 Tofe: Yeah that would work as well I gues Jun 08 15:34:13 +s Jun 08 15:34:55 From a security point of view I think it's ok Jun 08 15:44:25 Tofe: Well yes, it's anyway our own app and we know what it does Jun 08 15:50:01 I guess the goal of that "security" is to avoid injected code misbehaving or something like that; it doesn't necessarily improve the security for a qml app Jun 08 16:15:36 Tofe: Yeah well the whole ACG is good in general to lock things down (just like we should still move to read-only rootfs at some point) Jun 08 16:23:00 yes, it's far better than the public/private stuff Jun 08 16:28:17 Herrie: can I force push on your herrie/qml-based-acg branch ? It seems easier Jun 08 16:28:37 Tofe: sure Jun 08 16:28:40 Go ahead Jun 08 16:30:30 ok, let me just test this on device; the generated files seem fine Jun 08 16:30:48 ah, I forgot some cleanup Jun 08 16:31:59 done Jun 08 16:33:01 also I'll add acg permissions for wifi & bluetooth Jun 08 16:33:54 in the end I didn't do what I described here :) I just removed the role file, which went in the way of the webos_app script, and the rest was already pretty well matching Jun 08 16:35:04 Damn, it would be useful to get groups corresponding to an API Jun 08 16:35:33 I'll sketch up some script to do that on command line Jun 08 16:36:28 yes, everything LGTM today Jun 08 16:36:51 JaMa: that's a very good Monday then Jun 08 16:37:32 or very resigned :) Jun 08 16:38:09 but compared to everything else I really like your changes and SVG (even with only a little time I was able to look at them) Jun 08 16:39:10 Herrie: I realize we don't use LS2 for wifi or bluetooth, so you already did what was needed :) Jun 08 16:41:13 JaMa: I don't really understand why LG, when choosing between journald and PmLogDaemon, went for the latter. Everything is already in the syslog message, so PmLogDaemon only advantage is that is filters and splts the flow into different text messages. But that can be done with journalctl too, that was a bit the point of journald... Jun 08 16:41:40 s/messages/files/ Jun 08 16:43:17 Anyway. If at some point PmLogDaemon causes an issue or conflict, I think I know what i'll propose. Jun 08 16:44:43 Tofe: I agree, and IIRC I've asked the same internally when we were switching from upstart to systemd, but probably didn't get reasonable answer Jun 08 16:45:33 Tofe: but some products are really trying to disable journal, so maybe PmLogDaemon was seen as common component for all builds, but given how different everything else is between projects I don't see much benefit in it Jun 08 16:46:01 I was patching 5 different versions of luna-service2 right now just to make it buildable with newer Yocto .. Jun 08 16:46:33 Well, PmLogLib is a convenient piece to have between an app and the syslog call Jun 08 16:47:00 and I'm not even going to ask LS2 team to integrate this, because everybody is bussy just reshuffling commits in all this mess with no time for actual new work :) Jun 08 16:47:01 but afterwards, journalctl has all the filters it need to efficiently analyze an issue Jun 08 16:47:40 JaMa: ah ! you remind me I should discuss a luna-service2 PR with Herrie :) Jun 08 16:48:48 Herrie: https://github.com/webOS-ports/luna-service2/pull/9 <-- we don't need that one, in the end ? I thought we could just not merge it right now, test legacy apps, and see if it's needed in the end Jun 08 16:49:59 JaMa: 5 versions... damn. Jun 08 17:01:47 Tofe: Yeah I guess we're OK without that one Jun 08 17:03:51 I still have issues with C+DAV Google though Jun 08 17:04:28 All seems OK, nothing in logs but account doesn't show up in C+Dav app, however does show up in accounts app Jun 08 17:04:51 Could be ACG related or something I messed up in service renaming etc Jun 08 17:05:03 But no sync happens Jun 08 17:05:07 webos-connman-adapter[2279]: [] [pmlog] LS_REQUIRES_SECURITY {"SERVICE":"org.webosports.app.settings.deviceinfo-3728","CATEGORY":"/","METHOD":"getinfo"} Service security groups don't allow method call. Jun 08 17:05:14 ^I'll look into that one Jun 08 17:05:24 Don't really see any clues in logs though Jun 08 17:06:17 Tried ls-monitor no clues there either Jun 08 17:06:42 Ah that one should be easy to fix for connman Jun 08 17:08:30 yes, networking.query Jun 08 17:11:08 looks like we are missing some API for com.android.properties/getProperty Jun 08 17:12:30 Tofe: I thought I covered that one but could be I forgot it altogether.... Jun 08 17:12:37 should be exposed by our android-property-service Jun 08 17:13:10 let's say we'll put this API in the "system" group Jun 08 17:13:20 or "settings" Jun 08 17:15:07 good to go for the settings-qml PR, I'll merge it Jun 08 17:22:00 new meta-webos-ports changes, just in time, builder is just finishing testing builds :) Jun 08 17:33:36 Tofe: Seems I noticed android-property-service somewhere but forgot to migrate it Jun 08 17:33:41 I don't have it locally either Jun 08 17:34:01 It's a bit of a weird one anyway compared to our other services Jun 08 17:34:33 Tofe: You'd need some more files, only API wouldn't work I'd say Jun 08 17:36:58 role file is there, manifest too Jun 08 17:37:33 yup, it works now Jun 08 17:40:16 ok, settings app is now migrated Jun 08 17:42:50 How did the halium-7.1 rebuild go btw, I didn't have a look yet Jun 08 17:44:23 Tofe: I guess we might need to patch the halium-devices repos Jun 08 17:44:29 I sent a PR but don't think it got merged yet Jun 08 17:45:53 ah, mido failed Jun 08 17:47:59 TheMuppets it couldn't find? Jun 08 17:48:45 I didn't look closely at the error Jun 08 17:48:56 but was related to vendor yes Jun 08 17:49:08 OK Jun 08 17:49:38 ah, you beat me to the merge Jun 08 17:50:57 Herrie: I've just scheduled a new build Jun 08 17:51:17 Tofe: Fixed Jun 08 17:51:22 Switched device repo to mine as well Jun 08 17:51:33 I fixed some of the qseeproxydaemon etc there Jun 08 17:52:17 good :) Jun 08 17:53:12 I.e. some of these: https://github.com/Herrie82/android_device_xiaomi_mido/commits/pgz-14.1 Jun 08 17:55:02 yes, I remember these Jun 08 17:55:17 I was surprised to see qseeproxydaemon in my logfiles last week, since I was sure I tackled it LOL Jun 08 17:55:20 Now I saw why Jun 08 17:57:13 Seems I never updated halium-devices and probably used a locally created file ;) Jun 08 18:23:21 Tofe: Well hopefully this one is better Jun 08 19:02:16 Herrie: I have tested tissot with my own sim card Jun 08 19:02:21 - pincode works Jun 08 19:02:33 - calling works, but no sound in or out Jun 08 19:03:03 so I was on right track: it is "just" a sound issue Jun 08 19:03:10 Tofe: OK Jun 08 19:03:12 probably the routing is wrong Jun 08 19:03:21 Incoming call shows on display? Jun 08 19:03:21 and/or not switch to call mode Jun 08 19:03:34 ...oops I didn't try that :D Jun 08 19:04:04 but I'd say "probably" Jun 08 19:04:06 I remember it wouldn't show, reject call to voicemail and just show something in logs Jun 08 19:04:21 I'm only testing output calls for now Jun 08 19:05:14 now I wonder what kind of debug I could do to determine what's wrong Jun 08 19:05:44 It could be something in pulseaudio somehow? Jun 08 19:06:08 Might be good to check piggz work on mido Jun 08 19:06:15 For SFOS Jun 08 19:06:23 maybe yes Jun 08 19:06:27 I'll have a look Jun 08 19:08:44 https://github.com/piggz/droid-config-mido?files=1 Jun 08 19:09:02 I didn't see anything just now but on phone Jun 08 19:11:44 I quite don't know what their "x-policy" is for puleaudio Jun 08 19:11:50 pulseaudio* Jun 08 19:14:08 Tofe: This might be a hint: https://github.com/piggz/droid-config-mido/blob/master/sparse/etc/ohm/plugins.d/accessories.ini#L4 Jun 08 19:14:12 Seems to specify jack Jun 08 19:14:20 If we have it wrong could lead to muted audio Jun 08 19:22:22 That's the only clue I have so far Jun 08 19:23:44 Tofe: x-policy seems mainly for FM Jun 08 19:23:52 Which we don't support yet anyway Jun 08 19:25:49 Tofe: You've seen this? https://elinux.org/images/6/61/Elc_telephony_piirainen_0.pdf Jun 08 19:25:59 sounds finnish Jun 08 19:28:30 It's a little outdated but might give some clues as to architecture and how things work Jun 08 19:29:14 the ohm policy might be a clue Jun 08 19:29:54 Well but we don't have ohm ;) Jun 08 19:33:11 Don't think we'd want/need it either? Jun 08 19:33:21 I have no idea, so far Jun 08 19:33:29 Pretty sure there should be an existing webOS equivalent somehow Jun 08 19:33:53 I.e. https://github.com/webOSose/audiod-pro Jun 08 19:35:47 And https://github.com/webosose/com.webos.service.audiooutput Jun 08 19:36:28 Ohm is here: https://git.sailfishos.org/mer-core/ohm Seems to originate from Maemo Jun 08 19:38:46 seems to be related to power policies Jun 08 19:42:16 Tofe: And this? https://github.com/mer-hybris/audioflingerglue Jun 08 19:42:36 Or ? https://github.com/mer-hybris/audiosystem-passthrough Jun 08 19:43:33 I think this is more related to the gstreamer droidmedia thing Jun 08 19:43:53 and maybe that's related, but in this case it's a big rework for us Jun 08 19:50:48 Could simply be some missing pulseaudio config or the droid modules Jun 08 19:51:34 could be as well Jun 08 19:51:45 though they don't seem to misbehave Jun 08 20:22:14 Tofe: not misbehaving !== properly configured ;) Jun 08 20:22:30 agreed Jun 08 20:22:50 it could also be that the volumes are set to zero :p Jun 08 20:28:10 Pretty sure pulseaudio has some kind of monitor Jun 08 20:28:29 Or maybe good old ls-monitor could help even Jun 08 20:31:00 I'll try tomorrow, I'm a bit tired now :) Jun 08 20:35:38 Same here, bedtime Jun 08 20:35:52 Pactl might actually be useful here it seems Jun 08 20:38:24 yes, I already had a look earlier, but maybe I missed a clue Jun 08 20:38:28 gn8 ! **** ENDING LOGGING AT Tue Jun 09 02:59:57 2020