**** BEGIN LOGGING AT Wed Apr 03 02:59:57 2019 Apr 03 06:49:26 Morning! Apr 03 06:50:03 JaMa: reading https://manpages.debian.org/stretch/systemd/systemd-logind.8.en.html I mainly see conflicts between logind and our LunaSysMgr/LunaAppManager/sleepd/etc Apr 03 07:07:34 Morning! Apr 03 07:59:50 We might want to bump our pulseaudio-modules-droid : https://github.com/mer-hybris/pulseaudio-modules-droid/commit/8ff57cbc0dde7ef021d0e09a3d09b734c2e9f850 Apr 03 09:07:53 Tofe: That makes sense, seeing Warrior ships with PA 12.2 Apr 03 09:09:54 Not surprised audio doesn't work without it ;) Apr 03 09:12:22 If that works, and we disable logind, there's still the wifi middleware to debug a bit and then tissot is good to go Apr 03 09:13:10 btw: ideally, we should provide more sensible initial values for systemd-rfkill, so that at first boot it unblock wifi by default Apr 03 09:19:57 Yes, and I guess we should test some 3.4 kernel devices as well ;) Apr 03 09:20:01 Those are always tricky Apr 03 09:20:14 3.18 is more or less OK but 3.4 well ;) Apr 03 09:25:06 Herrie|Laptop: I was testing on hammerhead yesterday and even without deeper testing like Tofe it worked better for me than current testing build Apr 03 09:25:50 Tofe: should I disable systemd logind PACKAGECONFIG before it starts another qtwebengine build for me? now it's finishing tissot, hammerhead is next Apr 03 09:26:35 or do you think we can get away with logind.conf modifications? those wouldn't cause any big rebuilds Apr 03 09:32:03 JaMa: I won't be able to PR the logind.conf modification before tonight; but disabling it in PACKAGECONFIG is fine for me Apr 03 09:34:16 The idea being to set "HandlePowerKey=, HandleSuspendKey=, HandleHibernateKey=, HandleLidSwitch=, HandleLidSwitchExternalPower=, HandleLidSwitchDocked=" all to "ignore" Apr 03 09:42:16 Tofe: ok, will let the build finished with logind and then try to modify logind.conf only Apr 03 09:42:33 the build might not finish before you return tonight :/ Apr 03 09:43:34 BTW: 5.12.3 should be relasesed April 12, so I guess we can wait a bit more for that one Apr 03 09:44:03 I'll merge 5.12.2 to meta-qt5 as is even without successful test with LuneOS Apr 03 09:44:11 famous last words :P (re: qt relese) Apr 03 09:46:15 JaMa: https://codereview.qt-project.org/#/c/256191/ has just been merged, that'll be one less patch on our side for 5.12 Apr 03 09:47:40 merge to 5.12.3 branch also finished very recently, so it depends if they will update it again, otherwise it will wait in 5.12 for 5.12.4 Apr 03 09:48:09 [17:09:23] Branching from '5.12' -> '5.12.3' finalized yesterday and so on all changes for Qt 5.12.3 release needs to be done in '5.12.3' Apr 03 09:48:09 ah, too bad Apr 03 09:48:14 already missed Apr 03 09:48:31 but I have the updated version of your patch already so it should apply the same in 5.12.2 and 5.12.3 Apr 03 09:48:36 sure Apr 03 10:45:35 I have the pulseaudio-modules-droid upgrade ready but I'm wondering why we still need "droid-extcon.c: adapt to new pulseaudio 5 APIs" patch when upstream already supports new PA, let me try to build without this one Apr 03 11:00:41 https://github.com/webOS-ports/meta-webos-ports/pull/327 Apr 03 11:02:38 it could be that it's useless now Apr 03 11:03:21 interesting, you prefer using sed on the installed file instead of providing an override .conf ? maybe so that future new options will be present ? Apr 03 11:03:47 yes and also shows the "diff" of what we were interested in Apr 03 11:04:41 if we provide our own logind.conf, then few years from now the upstream logind.conf might look a lot different and you would need to read the commit message to find out why we changed it Apr 03 11:04:50 with sed it's more self-explanatory Apr 03 11:05:05 but on the other hand I should make git patch for the original logind.conf.in Apr 03 11:05:26 so that we'll notice when the .conf file is changed in upstream in a way that it won't no longer work with the sed Apr 03 11:08:43 updated the PR with patch Apr 03 11:10:33 ok thanks :) Apr 03 12:51:30 mtp-server will need similar fix as android-property-service: mtp-server/0.0.3+gitrAUTOINC+6159f1a143-r0/recipe-sysroot/usr/include/hybris/properties/properties.h:28:10: fatal error: android-config.h: No such file or directory Apr 03 12:51:55 Tofe: this is one of the dangers of meta-android/conf/layer.conf:SIGGEN_EXCLUDERECIPES_ABISAFE += "libhybris" Apr 03 12:52:30 nothing gets rebuilt because of libhybris SRCREV bump and now I'm getting new errors caused by new libhybris, just because something else caused the rebuild Apr 03 12:53:03 and it gets worse, when the issue is some subtle ABI change in libhybris libraries (hopefully not this case) Apr 03 12:53:33 but I was ready to blame the bluetooth link error on this (except that it happens on qemux86 as well) Apr 03 12:54:30 and the 2nd pulseaudio-modules-droid patch is still needed: git/src/droid/droid-extcon.c:125:14: error: 'pa_device_port {aka struct pa_device_port}' has no member named 'is_output' Apr 03 13:19:35 JaMa: we should have a libhybris-headers or something like that Apr 03 13:21:51 There's a difference between depending on virtual/egl and depending on libhybris, but bitbake doesn't let us word it Apr 03 13:22:00 or ?... Apr 03 13:41:42 yes and libhybris is MACHINE_ARCH Apr 03 13:42:18 so it's not only between mesa and libhybris, but also each MACHINE libhybris Apr 03 13:42:29 that's why we have SIGGEN_EXCLUDERECIPES_ABISAFE **** BEGIN LOGGING AT Wed Apr 03 14:10:19 2019 Apr 03 14:45:33 there are new unstable/warrior images on jenkins now, but wait with testing until I merge the fix for pa, mtp-server and logind.conf Apr 03 14:51:46 https://github.com/webOS-ports/mtp-server/pull/3 Apr 03 14:52:05 I guess I can go ahead and just merge it Apr 03 14:58:50 all merged Apr 03 15:03:50 thanks ! Apr 03 15:06:13 it's not tested yet, but at least there will be images to test from jenkins soon Apr 03 15:06:42 I've build tested it on tissot and hammerhead build is somewhere in the middle Apr 03 15:07:18 Tofe: what's next on the list? rfkill config? Apr 03 15:07:50 yes, rfkill Apr 03 15:07:51 I've quickly scanned through systemd-config if there is something for rfkill and haven't noticed anything Apr 03 15:08:05 nor in /etc/systemd from systemd recipe Apr 03 15:08:22 looks like it would be just putting an initial file on rootfs for first boot, in /var/lib/rfkill or something like that Apr 03 15:08:24 but I haven't spend much time on it Apr 03 15:08:52 good, that could be done from systemd-config (without big rebuild) Apr 03 15:08:58 like echo 1 > /var/lib/rfkill/wlan Apr 03 15:09:14 that could be done from initrd too Apr 03 15:09:28 we have a first boot scenario there too, for copying initial data Apr 03 15:09:30 ok Apr 03 15:11:17 but I'll need to look a bit more where it would be more appropriate Apr 03 15:11:43 though creating it before systemd is starting looks like a good idea :) Apr 03 15:13:08 https://github.com/systemd/systemd/blob/master/src/rfkill/rfkill.c#L127 says /var/lib/systemd/rfkill/wlan could work Apr 03 16:59:08 new tissot build flashed, let's see now... Apr 03 16:59:29 yep, sound works ! Apr 03 17:12:34 JaMa: I see we also have systemd-networkd.service... I don't think it's compatible with connmanctl. Apr 03 17:17:50 mmh doesn't help with networkd masked Apr 03 17:19:43 Tofe: local build or from jenkins? Apr 03 17:19:50 local Apr 03 17:20:05 I updated meta-wop and smartphone Apr 03 17:20:11 ok, because the one on jenkins just finished, now doing hammerhead, so they aren't rsynced yet Apr 03 17:20:41 my local still doing maliit NOTE: Running task 3399 of 6693 Apr 03 17:21:10 and [11920/17952] inside qtwebengine Apr 03 17:31:33 ahah {"returnValue":false,"errorCode":-1,"errorText":"Denied method call \"findnetworks\" for category \"/\""} Apr 03 17:34:22 ok then, this has nothing to do with warrior Apr 03 18:40:02 I wanted to drop meta-raspberrypi layer, because it's last which wasn't marked as compatible with warrior (needed for latest oe-core/warrior revision), but unfortunatelly removing them causes rebuild for pretty much everything http://jenkins.nas-admin.org/job/LuneOS/view/unstable/job/luneos-unstable_qemux86/180/console started rebuilding glibc Apr 03 18:40:48 so I'll revert the removal and we'll wait a bit more for rpi PR to be merged, don't update to latest webos-ports-setup until then (3rd to last commit is fine, even when it causes back rebuild) Apr 03 18:53:54 or rather if you do update, then apply https://github.com/agherzan/meta-raspberrypi/pull/406 manually to your meta-raspberrypi checkout Apr 03 20:29:23 Hello! Apr 03 20:43:54 Andolamin: hi !! Apr 03 20:44:09 It's been a while... Apr 03 20:44:18 How are things around here? Apr 03 20:58:18 Pretty busy lately Apr 03 20:58:32 We were upgrading from pyro to sumo Apr 03 20:58:47 but then, why not to thud Apr 03 20:59:02 and finally we're not testing builds on warrior Apr 03 20:59:15 all this in one month :) Apr 03 20:59:21 -not Apr 03 21:00:15 And at the same time, I got a mainline kernel running on N5 -- though it's still pretty alpha quality Apr 03 21:00:44 Nice! Apr 03 21:01:02 I'm planning on doing a PoC LuneOS port to Jetson TX1/TX2/Nano Apr 03 21:01:05 yup, quite satisfying :) Apr 03 21:01:22 * Tofe googles Jetson... Apr 03 21:02:00 wow ! Apr 03 21:02:06 I've TX1 as well :) Apr 03 21:02:20 I have a TX2 already, and a Nano on order. The Nano uses the same SoC as the TX1, just slimmed down a bit, so if it works on the Nano, should also work on the TX1 Apr 03 21:03:06 what are you using them for? Apr 03 21:03:26 What kind of SDK do they provide ? Android ? Mainline ? Homemade ? Apr 03 21:03:34 I do have working yocto builds on my TX2 currently, running an application stack of Qt + Nodejs 11.11.0 Apr 03 21:03:57 using git://github.com/madisongh/meta-tegra I assume Apr 03 21:04:18 Ok, so actually a LuneOS port wouldn't be so far away! Apr 03 21:04:42 Currently using sumo Apr 03 21:04:59 Yeah, based on that meta-tegra BSP Apr 03 21:05:35 I'm using Qt 5.12.2 Apr 03 21:05:38 Our pair of shaders for the card shadows would be quite at ease here :p Apr 03 21:05:40 From boot2qt Apr 03 21:06:03 Yeah, it has a huge GPU compared to most embedded platforms Apr 03 21:06:13 We are one patch away from using Qt 5.12 too Apr 03 21:06:36 Would love to port my work over to a LuneOS system Apr 03 21:06:39 one patch we know about and one fix we don't know (the bluetooth issue) Apr 03 21:07:56 Once I get my hands on one, I'm also thinking about building a LuneOS tablet using the Pi 3 A+ Apr 03 21:08:16 It's slim enough it should fit into a fairly reasonable 3D printed case, with battery, speakers, etc Apr 03 21:11:35 Also, any interest in a Plex app for LuneOS? Thinking about adapting https://github.com/plexinc/plex-media-player Apr 03 21:18:11 can be interesting, but can also be quite some work :) The stack is already available on our side, though, so that's fine Apr 03 21:23:41 heh just when I was thinking that we're already spread too thin and should support less devices but better :) Apr 03 21:24:20 but I can still drop those meta-rpi layers again even for the cost of another rebuild Apr 03 21:28:17 JaMa: I agree we're very thin right now Apr 03 21:33:13 my warrior builds finally finished, but triggering the master with qt 5.12.2 now, will call it a day and test warrior tomorrow **** ENDING LOGGING AT Thu Apr 04 02:59:57 2019