**** BEGIN LOGGING AT Sun Jul 05 03:01:09 2020 Jul 05 06:22:52 Morning! Jul 05 06:25:08 Tofe: The maliit-server and maliit-plugin we get from meta-qt5 is pretty old already quite some changes upstream from the looks of it. And they moved from QMake to CMake. I got the recipe updated yesterday evening for maliit-server just it was close to midnight so need to check that a bit still to make sure I didn't miss anything and clean it up a bit and then test. Jul 05 06:58:58 Morning! Jul 05 07:51:19 Herrie: it's the exact same stack :( Jul 05 08:07:29 Tofe: Would a bt from gdb be of any use? Jul 05 08:09:56 I'm already on it Jul 05 08:10:10 I do suspect some thread_local misbhevior, somehow Jul 05 08:14:05 Tofe: OK, so you have the same on Hammerhead as well? Jul 05 08:14:20 I saw this upstream which talks about QT 5.14+, but it seems also GTK which we don't use: https://github.com/maliit/framework/commit/99e41962d2ec74277357d886c2fb5eb10538cc9f Jul 05 08:14:31 yes, same on n5 Jul 05 08:15:16 I guess at some point updating maliit doesn't hurt anyway, will see where I get with it Jul 05 08:15:22 Herrie: I'll try a different version of the qtbase patch, this time avoiding using pthread_local at all Jul 05 08:20:46 Just the change from QMake to CMake will require some rework to recipe, but got something to build already, so I guess it's not going to be rocket science Jul 05 08:21:47 Herrie: it might also bring some other issues on the table, so I'll stick to the current version for this debug :p Jul 05 08:22:29 I now have my patch ready (it was quite easy, as Qt implemented actually both version uselessly), let's build and test Jul 05 08:23:12 I suspect some local pthread things from the android EGL driver is conflicting with Qt's TLS Jul 05 08:23:21 (Thread Local Storage) Jul 05 08:23:46 Tofe: Ah OK... Some funny edge cases again with Qt on Android :P Jul 05 08:23:52 Where it's not used very much... Jul 05 08:24:17 more like a libhybris side-effect I'd say Jul 05 08:24:44 but let's see if my fix makes things better... or worse Jul 05 08:24:53 write_ipk takes forever Jul 05 08:29:05 I broke the whole thing :p But maybe I see a mistake in my patch Jul 05 08:30:55 Tofe: Well I suspect issues with Mallit upgrade too, but I'll do some work on it anyway. Seems that there's quite some nice added things there so it would be good to have it upgraded at some point Jul 05 08:35:37 ah, better Jul 05 08:35:55 ... but no maliit :( Jul 05 08:37:14 this /usr/libexec/gstreamer-1.0/gst-plugin-scanner is a nasty thing, can't we disable it ?... Jul 05 08:38:35 \o/ maliit is there Jul 05 08:39:15 I think that's a good fix Jul 05 08:43:28 Herrie: https://bpa.st/I4BQ Jul 05 08:44:52 Tofe: It's the first time I hear about gst-plugin-scanner.... Jul 05 08:44:58 Why is it nasty? Jul 05 08:46:18 it's started by maliit-server (I don't know why), and then hangs (no idea why either) Jul 05 08:46:25 and it's quite useless for us Jul 05 08:47:32 Herrie: if you apply my new patch instead of the other one, things should work, now Jul 05 08:48:30 The only downside is that it's a bit less efficient, but we can afford it on old arm devices anyway Jul 05 08:50:02 And I think the issue was introduced with newer gcc Jul 05 08:51:16 now, I'll resume my camera work... :p Jul 05 08:52:30 Ah, yes, I wanted to get the Halium build running again Jul 05 08:52:41 was there an update on this? Jul 05 08:56:10 Tofe: Let me try the patch for Tenderloin Jul 05 08:56:21 Might be good to submit it to Qt? Jul 05 08:57:30 They'll probably refuse it -- it's related to libhybris, which probably isn't in their support scope Jul 05 08:57:59 if we reproduce it on mainline armv7, yes, but otherwise I don't think there's a chance. Jul 05 08:58:20 Plus, their code is fine Jul 05 09:00:16 It could also still be our old kernel somehow right? Jul 05 09:00:33 I guess we can just keep a patch at our end for it Jul 05 09:00:58 could be, but yes, we'll keep that on our end. And the new patch is simplier Jul 05 09:00:59 I saw some examples in OSE for specific arm archs so we could apply it just for armv7 I guess Jul 05 09:01:39 https://bpa.st/5V4A Jul 05 09:01:53 (the comment is incorrect now...) Jul 05 09:02:43 ideally it should only be applied for halium armv7 Jul 05 09:03:00 not sure if "_halium_arm" would work ? Jul 05 09:04:01 They have this in OSE: https://github.com/webosose/meta-webosose/blob/master/meta-webos/recipes-webos/chromium/webruntime.inc#L206 Jul 05 09:04:33 I don't think we have armv7ve at our end only armv7a right? Jul 05 09:05:18 Herrie: http://jenkins.nas-admin.org/job/LuneOS/view/halium/job/halium-luneos-7.1-build/40/console huh ? Jul 05 09:05:43 Herrie: I'm... not sure. I think you're correct. Jul 05 09:05:59 That's an issue on the builder Jul 05 09:06:13 We need JaMa, ka6sox or novaldex for that Jul 05 09:06:20 yep Jul 05 09:06:47 I wanted also to add a patch to disable the dupliacte service instances Jul 05 09:06:59 (the PR which was refused on Halium side) Jul 05 09:07:13 the two other PR were merged, iirc Jul 05 09:24:52 Tofe: Yeah that would be good Jul 05 09:28:53 Herrie: https://github.com/Halium/android_build/pull/18 do we merge it ? Jul 05 09:48:27 Tofe: saw you merged it good Jul 05 09:48:50 did I Jul 05 09:49:19 now, I did :) Jul 05 09:50:14 Following remarks on my patch for minimediaservice, I'll try the camera without starting that one. Maybe it's as simple as that. Jul 05 09:56:18 ooh I get some preview, über-zoomed, but still Jul 05 10:06:17 That's good Jul 05 10:06:31 yes, it's progress Jul 05 15:44:33 Tofe: Was out with the kids for a bit, now flashing Tenderloin with new image Jul 05 15:44:44 Then will try and see qtbase recipe for halium_armv7 :P Jul 05 15:49:29 Tofe: SRC_URI_append_halium_armv7a += seems to be OK for Tenderloin Jul 05 15:49:36 Now testing to see what it does for Mido ;) Jul 05 16:00:05 Tofe: Tenderloin seems to work OK as well now with Maliit Jul 05 16:00:45 mido build for qtbase is busy but needed to start build from scratch so will take a little bit before I have results Jul 05 16:07:33 Tofe: Seems we're OK with the recipe like that: https://paste.ubuntu.com/p/X9Zs8WXkVQ/ Jul 05 16:07:51 Note the fix001.patch was my patch filename Jul 05 16:08:09 Which only gets applied to Tenderloin and not to Mido when I use SRC_URI_append_halium_armv7a Jul 05 16:25:47 Tofe: You seen this? https://github.com/qt/qtbase/commit/81ce2d1d6fa741de4d27b939a378147a02019ec1 Jul 05 17:05:50 Seems it was committed yesterday only Jul 05 17:56:34 Herrie: ah, well, we'll have to rebase a bit my patch, but anyway it wouldn't solve anything for us Jul 05 18:02:56 Tofe: if you have rebased patch can you paste it somewhere? I have the rest ready for PR at my end Jul 05 18:03:15 Both meta-smartphone and meta-webos-ports Jul 05 18:03:38 Herrie: is meta-qt5 bumped already ? Jul 05 18:04:30 I meant, we'll have to rebase the patch... when meta-qt5 is updated Jul 05 18:08:34 Tofe:.ah true Jul 05 18:08:46 Ok will just PR then in a bit Jul 05 18:31:47 Tofe: https://github.com/shr-distribution/meta-smartphone/pull/116 and https://github.com/webOS-ports/meta-webos-ports/pull/441 Jul 05 18:32:02 I guess that should cover it right? **** BEGIN LOGGING AT Sun Jul 05 19:04:30 2020 Jul 05 19:10:11 OK good :) I'll probably toy a bit with the maliit upgrade to see if that might fix our regressions with spell check etc Jul 05 19:10:50 Also they forked ubuntu-keyboard on which ours was based, might be able to fork that back and add our bits will see Jul 05 19:12:38 This one I mean: https://github.com/maliit/keyboard Jul 05 19:17:03 Seems the maliit-plugin didn't really get updated so just server side, will toy a bit with that tomorrow then Jul 05 19:17:36 Mainly clean up recipe and add more CMake bits and verify patches that no longer apply Jul 05 20:22:57 ok :) **** BEGIN LOGGING AT Mon Jul 06 03:00:08 2020 **** ENDING LOGGING AT Mon Jul 06 03:00:56 2020