**** BEGIN LOGGING AT Tue Mar 15 02:59:58 2016
Mar 15 05:28:51 <Herrie> Andolamin: I see you got your new toy :D
Mar 15 05:29:13 <Andolamin> Herrie: Indeed I did. And, I *think* I might have figured out what the build issue was
Mar 15 05:29:39 <Andolamin> I hope, at least. I've *thought* I had it figured out about a dozen times now.
Mar 15 05:30:20 <Andolamin> Of course, the issue is definitely in the graphics stack somewhere, so every new build attempt requires a complete qt rebuild
Mar 15 05:32:16 <Andolamin> It builds and boots but can't start the cardshell because of some issue in the graphics stack
Mar 15 05:33:53 <Herrie> Andolamin: Hmmz that's a pain
Mar 15 05:34:02 <Herrie> Especially when you need to rebuild QT all the time :P
Mar 15 05:34:06 <Andolamin> Yeah
Mar 15 05:34:16 <Andolamin> Each test iteration takes ~2 hours
Mar 15 05:54:16 <Herrie> Luckily we normally don't rebuild it that often when it's working properly
Mar 15 05:55:30 <Andolamin> Yeah - normally tweaking something minor just takes a few minutes to do a biuld
Mar 15 05:55:37 <Andolamin> s/biuld/build
Mar 15 05:57:54 <Andolamin> On qtwebengine now, so nearly done
Mar 15 06:34:37 <Tofe> Morning!
Mar 15 06:36:53 <Andolamin> Morning
Mar 15 06:37:17 <Andolamin> Herrie: Progress! Need to rebuild everything again, but much more confident that the next build will be a good one
Mar 15 06:40:06 <Herrie> Andolamin: Nice :D
Mar 15 06:40:16 <Herrie> Tofe: morning
Mar 15 06:41:12 <Tofe> Andolamin: good luck :)
Mar 15 06:41:42 <Tofe> (and patience, in this case...)
Mar 15 06:44:37 <Andolamin> Looks like during the Pi 3 merge they bumped the gpu drivers, which started causing my driver patches to silently fail. The build would go through, but it would fail at run.
Mar 15 06:45:06 <Andolamin> Removing all of my patches seems to work. On a positive note, it also means we should get wayland-egl now.
Mar 15 06:49:37 <Tofe> ah, a silent fail is nasty
Mar 15 06:50:17 <Tofe> cool for wayland-egl :) I'm not sure I see the consequences, but it looks like a good thing
Mar 15 07:03:40 <Andolamin> Tofe: It's just one step closer to eliminating the MACHINE_ARCH issue with the Pi boards
Mar 15 07:03:47 <Andolamin> So, faster builds primarily.
Mar 15 07:04:19 <Andolamin> I think that wayland-egl draws directly to the framebuffer, as well, so it might be a tad faster than wayland, but not sure it would be noticeable.
Mar 15 07:15:50 <Herrie> Tofe: What you're working on currently? Or having a little break :D
Mar 15 07:57:16 <Herrie|Veer> morning :)
Mar 15 08:11:06 <Tofe> Herrie|Veer: yes, having a little break :p
Mar 15 08:12:09 <Tofe> Then I'll propably try to improve a bit more the phone app
Mar 15 08:14:24 <Herrie|Veer> Tofe: OK, well deserved break :)
Mar 15 08:16:47 <Tofe> And it's a bit calmer now, with the waiting for jethro and qt 5.6 releases
Mar 15 08:17:22 <Herrie|Veer> Tofe: jethro is out, just not working it seems for us :P
Mar 15 08:17:34 <Herrie|Veer> 5.6 you had issues with still right?
Mar 15 08:34:32 <Tofe> Herrie|Veer: well, 5.6 isn't out yet
Mar 15 08:34:55 <Tofe> though I doubt that the crash we had will be fixed for the release
Mar 15 08:34:59 <Herrie|Veer> Tofe: RC is ;)
Mar 15 08:37:23 <Tofe> Herrie|Veer: but the switch is easy to do, just the crash will be a long debug I feel...
Mar 15 08:38:09 <Herrie|Veer> Tofe: Yeah I think so too... It's probably in something that other projects don't use yet... Since they're way behind in QT terms still :P
Mar 15 08:39:25 <Tofe> Herrie|Veer: I don't know if the plasma project uses the cutting edge versions
Mar 15 08:42:00 <Herrie|Veer> Tofe: I think they're still on webkit
Mar 15 08:42:38 <Tofe> well, our crash is independant of webkit/webengine, in our case
Mar 15 08:43:33 <Herrie|Veer> Tofe: It's probably more something in hybris/qpa/wayland
Mar 15 08:43:49 <Herrie|Veer> Which others usually also don't experience because we're ahead :P
Mar 15 08:54:47 <Tofe> Also qtwayland isn't used by ubuntu, jolla is still on Qt 5.2, and desktop wayland compositors don't use hybris...
Mar 15 08:57:22 <Herrie|Veer> Tofe: Might be interesting to see how things behave on qemu or Pi 2/3 devices to see if we can eliminate some possible causes
Mar 15 09:09:16 <Tofe> The RPi doesn't use hybris, right ? It uses mesa ?
Mar 15 09:11:44 <Herrie|Veer> Tofe: I think so. Qemu uses mesa too I think. It might just exclude some causes so we can focus on others. If it's happening on qemu too it might mean something with 5.6 in general?
Mar 15 09:22:23 <Tofe> yes, that would exclude hybris from the group of suspects
Mar 15 09:28:45 <Herrie|Veer> Tofe: Should be fairly straight forward for you to do the qemu one right?
Mar 15 09:30:20 <Tofe> Well, it's been a while since I last built a qemu target, and I'll have to redo some little things for getting Qt 5.6 RC back on line
Mar 15 09:35:24 <Herrie|Veer> If you feel like giving it a go, go ahead :) If not I can ask Andolamin to look into 5.6 for Pi once his current build is working & integrated.
Mar 15 13:54:52 <Andolamin> JaMa: ping
Mar 15 13:56:00 <Andolamin> or maybe Tofe might know too... Tofe: ping?
Mar 15 13:57:56 <Herrie|Veer> Andolamin: I might know too but maybe not :P
Mar 15 13:58:09 <Herrie|Veer> Still no luck with latest build?
Mar 15 13:58:20 <Andolamin> Maybe. Do you know what MACHINEOVERRIDES = "raspberrypi2:${MACHINE}" in that raspberrypi 3 machine conf is actually doing?
Mar 15 13:59:56 <JaMa> it makes all raspberrypi2 overrides applicable for raspberrypi3 as well
Mar 15 14:00:18 <Andolamin> Ah, OK. Thanks!
Mar 15 14:03:20 <Andolamin> Herrie|Veer: I made another couple of attempts last night. I had to set up a meta-raspberrypi branch that doesn't include all the recent userland changes, but was able to get a working build for the raspberrypi2.
Mar 15 14:08:21 <Herrie|Veer> Andolamin: Is that including wayland-egl or excluding?
Mar 15 14:08:29 <Andolamin> Excluding
Mar 15 14:11:41 <Andolamin> I'll keep working on reintegrating those patches, though. When I get it working it will just be adjusting the install path for the pthreads library and we'll be able to eliminate all qt5 changes for the pis
Mar 15 14:19:13 <Herrie|Veer> Andolamin: OK. This is Qt 5.5 still right?
Mar 15 14:19:49 <Andolamin> Yes
Mar 15 14:21:18 <Herrie|Veer> After we got 5.5 OK we might want to experiment a bit with 5.6. We have some issues on our Android targets, not sure if they exist on qemu/pi 2/3 too.
Mar 15 14:24:19 <Andolamin> Yeah - completely different stack on the Pi, and there are official cross-compilation instructions for the Pi up the Qt website, so I think it would be relatively straight forward
Mar 15 14:25:19 <Herrie|Veer> Just want to exclude some causes
Mar 15 14:25:40 <Andolamin> Yeah - for sure
Mar 15 14:27:28 <Herrie|Veer> So we know if it's qt 5.6 or Android specific bits
Mar 15 14:30:41 <Herrie|Veer> I can imagine that the DBUS rework in 5.6 might cause the issues somehow, but could be something else altogether as well.
Mar 15 14:36:09 <Herrie|Veer> So if we could narrow it down somehow that would be good.
Mar 15 14:36:23 <Herrie|Veer> But that's after we get things up & running in our build env :)
Mar 15 14:54:43 <Tofe> Andolamin: pong, but no, I wouldn't have been able to answer that!
Mar 15 14:55:54 <Andolamin> No problem!
Mar 15 14:57:02 <Tofe> I'll setup a qemu target in the following days
Mar 15 14:57:12 <Tofe> (for Qt 5.6 build)
Mar 15 14:57:27 <Herrie|Veer> Tofe: Nice :)
Mar 15 14:58:16 <Herrie|Veer> Seeing the DBUS rework in 5.6 that could be a cause too, though not very likely. But we use it in some bits in LNC for BlueTooth, oFono etc
Mar 15 15:00:42 <Tofe> I don't think it's that; the crash is very local and on very specific scenarios: only when starting an app, and it crashes immediately in the compositor; I find it very unlikely that DBus is involved here
Mar 15 15:01:27 <Tofe> (well, not exactly when starting an app, more when a QML surface is hidden or something like that)
Mar 15 15:02:37 <Herrie|Veer> Tofe: OK :)
Mar 15 15:02:57 <Herrie|Veer> Well qemu and Pi are interesting targets to see if it's specific to our Android targets
Mar 15 15:03:09 <Herrie|Veer> So we can at least exclude other causes :)
Mar 15 15:04:52 <Tofe> It's just that I will have to make space for it, so I'll hve to remove all my N4 build :p
Mar 15 15:06:38 <Herrie|Veer> I think we should sponsor you a larger SSD :P
Mar 15 15:07:48 <Tofe> oh well, it's not that big a problem :)
Mar 15 15:09:10 <Tofe> I think it's my rational side: I just built the mako build, so I don't want to throw it away; but actually it's just starting the build when leaving for work, and getting the build done when back home
Mar 15 15:09:37 <Herrie|Veer> :)
Mar 15 16:34:13 <Andolamin> raspberrypi3 builds are still failing
Mar 15 16:34:44 <Andolamin> Tofe, JaMa: https://gist.github.com/Andolamin/2405776e5dd10d4efce9
Mar 15 16:35:13 <Andolamin> Seems the issue is in building webos-keyboard
Mar 15 16:37:22 <Andolamin> I checked the listed sysroot folder (/media/alan/Workstation/builds/luneos/webos-ports/tmp-glibc/sysroots/raspberrypi3) and all of the "missing" libraries are there
Mar 15 16:53:14 <Herrie|Veer> Andolamin: Not sure why that would fail for pi3 and not for pi2 :s
Mar 15 16:53:25 <Andolamin> Right? Right?!
Mar 15 16:56:36 <Herrie|Veer> Andolamin: We have some outstanding issue in the bugtracker for webos-keyboard/maliit which rings a bell but it doesn't prevent us from building currently
Mar 15 16:57:18 <Andolamin> Well, even if it did I would have expected it to cause the build to fail on the pi2 as well...
Mar 15 16:57:41 <Herrie|Veer> Andolamin: Yes
Mar 15 16:58:55 <Herrie|Veer> See 901 in the bugtracker
Mar 15 17:00:28 <Herrie|Veer> Which is somewhat related to libluneos-keyboard-plugin.so when I was looking into the issue
Mar 15 17:02:58 * elvispre3 wishes all his issues were so easily resolved!
Mar 15 17:03:14 <Herrie|Veer> elvispre3: Yeah LOL
Mar 15 17:03:23 <Herrie|Veer> I'll just sort them for you LOL
Mar 15 17:03:43 <elvispre3> :-)
Mar 15 17:23:11 <Andolamin> Herrie|Veer:
Mar 15 17:24:02 <Andolamin> Herrie|Veer: Oops, fat thumbed enter instead of space. I updated https://github.com/webOS-ports/webos-ports-setup/pull/6 - should be good to go for pi2 builds. Will open a new one if necessary for pi3 builds.
Mar 15 17:55:28 <Herrie> Andolamin: merged, let me update testing & kick off some builds
Mar 15 17:58:03 <Andolamin> Woo!
Mar 15 17:58:31 <Herrie> OK picked to Testing
Mar 15 17:58:35 <Herrie> Will now kick off a build
Mar 15 17:59:27 <Herrie> Andolamin: Does it matter after which one I build RPi2?
Mar 15 17:59:46 <Herrie> We currently have: https://bpaste.net/show/b60e8eefecf4
Mar 15 18:00:15 <Andolamin> Shouldn't matter at all
Mar 15 18:00:23 <Herrie> I'll put it after Tenderloin
Mar 15 18:00:42 <Herrie> OK there goes nothing ;)
Mar 15 18:00:59 <Herrie> http://jenkins.nas-admin.org/job/luneos-testing_build/lastBuild/console
Mar 15 18:01:11 * Andolamin crosses fingers
Mar 15 18:01:43 <Herrie> Andolamin: Should be OK I guess
Mar 15 18:58:42 <Herrie> Andolamin: http://jenkins.nas-admin.org/job/luneos-testing_raspberrypi2/lastBuild/console
Mar 15 19:20:18 <Andolamin> Herrie: So far so good...
Mar 15 19:28:34 <Andolamin> Herrie: Might have a working raspberrypi 3 build in a few minutes, actually
Mar 15 19:30:21 <Andolamin> Got webos-keyboard building. Wasn't sure how to go about fixing the webos-keyboard recipe, but was 90%+ sure that it was the MACHINEOVERRIDES that was causing the issue, so removed the MACHINEOVERRIDES and just copied the *2* raspberrypi2 specific configs for raspberrypi3
Mar 15 19:30:48 <Andolamin> Not super happy with the solution, but it works for now
Mar 15 19:54:25 <Andolamin> Herrie: Odd that the total for the number of tasks is higher on the jenkins build than my local builds
Mar 15 19:54:38 <Andolamin> 6067 on jenkins, 5670 local
Mar 15 20:10:56 <Herrie|Laptop> Andolamin: For pi3 sounds like a reasonable workaround for now
Mar 15 20:11:04 <Herrie|Laptop> Not sure about Jenkins jobs tbh
Mar 15 20:11:10 <Herrie|Laptop> We'll see if it finishes OK
Mar 15 20:11:33 <Andolamin> Eventually they'd have to move away from the MACHINEOVERRIDES anyway. They won't be having identical pi2 and pi3 builds forever
Mar 15 20:11:50 <Andolamin> Right now the only difference should be some boot files
Mar 15 20:13:30 <Andolamin> RE: jenkins job, the build configuration looks correct, so I figure it will probably finish OK.
Mar 15 20:13:38 <Andolamin> Might be an Ubuntu 15.10 vs 14.04 thing
Mar 15 20:13:46 <Andolamin> My build machines are 15.10
Mar 15 20:14:21 <Andolamin> Had a successful pi3 build
Mar 15 20:14:24 <Andolamin> Flashing and testing now
Mar 15 20:16:10 <Andolamin> Wish I had a usb 3.0 port on this machine...
Mar 15 20:22:26 <Andolamin> It's alive!
Mar 15 20:23:02 <Andolamin> Except, I forgot the luna-platform.conf so all the scaling is really bad
Mar 15 20:23:28 <Andolamin> I'll open a PR for that
Mar 15 20:36:44 <Andolamin> Herrie: It boots! https://www.dropbox.com/s/0d9avr0bmobzn5z/2016-03-15%2013.26.40.jpg?dl=0
Mar 15 20:37:13 <Herrie|Laptop> Andolamin: NICE
Mar 15 20:37:20 <Andolamin> A couple of issues that I suspect are from switching the MACHINEOVERRIDES with a partially completed build.
Mar 15 20:37:53 <Andolamin> Doing a clean build to verify, then I think we should be ready to turn the Pi3 builds on tomorrow
Mar 15 20:38:02 <Andolamin> Its *fast*
Mar 15 20:38:20 <Andolamin> Scrolling the timezones list in first use is butter smooth
Mar 15 20:38:47 <Andolamin> On board WiFi didn't show up out of the box, but that should be fixable
Mar 15 20:42:43 <Herrie|Laptop> Andolamin: Could be connman somehow or first boot
Mar 15 20:42:50 <Herrie|Laptop> Can you try wifi on 2nd boot?
Mar 15 20:43:56 <Andolamin> Herrie|Laptop: Not at the moment - those couple of issues that I mentioned are actually that the cardshell renders as just a black screen -_-
Mar 15 20:44:01 <Tofe> Andolamin: yeah !
Mar 15 20:44:10 <Andolamin> I can interact with first boot no problem, but once I get past first boot, that's it
Mar 15 20:44:57 <Andolamin> Because first use works correctly, I suspect it's a sstate issue that cause the cardshell to not make it into the rootfs correctly
Mar 15 20:45:07 <Tofe> Cardshell is black ? That's weird
Mar 15 20:45:11 <Andolamin> I'm checking journalctl to verify
Mar 15 20:45:41 <Tofe> The cardshell is used also for displaying the first use app
Mar 15 20:45:47 <Tofe> even if it's only partially
Mar 15 20:47:36 <Tofe> But yes, if there's an issue you should see it in the luna-next logs
Mar 15 20:51:04 <Andolamin> Yeah - I strongly suspect that the issue was that bitbake didn't realize it needed to rebuild components once I remove the MACHINEOVERRIDES
Mar 15 20:51:27 <Andolamin> I had to manually bb -c cleanstate and rebuild about 6 components after that change just to get it to boot
Mar 15 20:52:10 <Tofe> ok I see
Mar 15 21:12:50 <Andolamin> Yeah - I think I found the issue. It's acting like qtbase was built with the wrong qpa
Mar 15 21:13:56 <Andolamin> Which is odd, because when that has happened in the past it wouldn't render anything. It would hang right where the LuneOS logo starts flashing.
Mar 15 21:14:05 <Andolamin> Here, the logo flashes, first use shows up, but nothing else...
Mar 15 21:26:50 <Herrie|Laptop> Tofe: I'll have a stab at upgrading our vkb stack + prediction (maliit, presage & hunspell), seems there have been quite some updates
Mar 15 21:27:03 <Herrie|Laptop> Will let you know how it goes and if I need some help
Mar 15 21:27:16 <Herrie|Laptop> But should be fairly straightforward it seems
Mar 15 21:27:28 <Tofe> ok; yes, I hope there won't be any surprise
Mar 15 21:27:46 <Herrie|Laptop> Tofe: Don't know, existing patches seem OK mostly
Mar 15 21:27:50 <Herrie|Laptop> So we'll see
Mar 15 21:27:53 <Tofe> yep
Mar 15 21:27:55 <Herrie|Laptop> Will take it a package at a time
Mar 15 21:28:12 <Herrie|Laptop> Might just help to solve some of the error messages in logs
Mar 15 21:28:14 <Tofe> on my side I've prepared the field for the qemu build of tomorrow
Mar 15 21:28:23 <Herrie|Laptop> If not were at least updated to latest :)
Mar 15 21:28:36 <Herrie|Laptop> Since it seems it's not really all maintained by OE anyway
Mar 15 21:28:45 <Herrie|Laptop> Tofe: Nice :)
Mar 15 21:29:01 <Herrie|Laptop> Anyway bedtime for now.. Been 5:30 again last couple of days :P
Mar 15 23:32:40 <Andolamin> Success! Writing build history now
Mar 15 23:32:55 <Andolamin> I'm going to download it and test is on one of my Pi 2 here in just a bit.
Mar 15 23:35:06 <Andolamin> Oh, doesn't look like the image was uploaded?
Mar 15 23:35:24 <Andolamin> http://build.webos-ports.org/luneos-testing/images/raspberrypi2/ doesn't have the rpi-sdimg
Mar 16 01:44:10 <Andolamin> So, fresh build of raspberrypi3 works perfectly. Cardshell displays properly and the UI is super snappy. Bad news: building pi2 in the same directory after a pi3 build produces all the same errors with webos-keyboard, so it was obviously not the MACHINEOVERRIDES change that fixed the issue but instead the cleansstate. Need to do some more work to figure out
Mar 16 01:44:10 <Andolamin> what is going wrong here...
**** ENDING LOGGING AT Wed Mar 16 02:59:58 2016