**** BEGIN LOGGING AT Fri Apr 29 02:59:56 2022 Apr 29 06:49:16 Morning! Apr 29 07:33:04 morning Apr 29 07:48:07 Morning! Apr 29 09:11:48 Tofe: Yeah it doesn't seem to build for me somehow Apr 29 09:11:50 Let me try again Apr 29 09:11:57 qtwayland-webos that is Apr 29 10:08:30 Tofe: I'm using https://github.com/webOS-ports/meta-webos-ports/tree/herrie/lsm-rebase Apr 29 10:09:05 Herrie: me too, it's strange Apr 29 10:09:09 I'll retry tonight Apr 29 10:09:35 Same error for qemux86-64 Apr 29 18:11:19 Herrie: "good" news, I could reproduce the build issue Apr 29 18:11:38 Tofe: Well that's a first step to a fix Apr 29 18:12:17 yep Apr 29 18:14:21 fixed Apr 29 18:18:51 I'll update the Qt 5.15 patch, one moment Apr 29 18:23:51 done Apr 29 18:30:57 OK let me try to build Apr 29 18:32:55 I've booted on PP, but no UI :( Apr 29 19:00:13 surface-manager LSM {} (null), file:///usr/lib/qml/WebOSCompositorBase/Views.qml:39:9: Type ViewsRoot unavailable Apr 29 19:00:30 file:///usr/lib/qml/WebOSCompositor/qml/LuneOSViewRoot.qml: Type LS unavailable Apr 29 19:03:20 fixing the error in LS, we get the same status as before the rebase Apr 29 19:08:51 Can you push that fix? Apr 29 19:09:20 well, it's more a workaround, but one moment Apr 29 19:10:14 I got a black screen on qemu Apr 29 19:10:22 yep Apr 29 19:10:54 LSM is actually the culprit here, it seems Apr 29 19:11:03 a bit weird Apr 29 19:11:23 Well maybe something weird in the rebase Apr 29 19:11:28 It could be I broke a patch somehow Apr 29 19:12:09 ah, I guess we also need a newer QML WebOSServices Apr 29 19:12:15 not sure what provides it Apr 29 19:12:53 qml-webos-bridge Apr 29 19:15:40 https://github.com/webosose/qml-webos-bridge/commit/196c82f58ac1352a6f1e22180ac3de4ad18e0770 we need that commit, at least Apr 29 19:16:58 let me push the SRCREV bump Apr 29 19:18:57 done Apr 29 19:19:20 I confirm this fixes UI Apr 29 19:20:02 but still issues with QML apps, and no maliit window :( Apr 29 19:22:01 still, if you don't see any regression in qemu or so, then we can merge the rebase Apr 29 19:37:14 Let me put kids to bed while this image with fix builds Apr 29 20:21:39 Tofe: Yeah I get UI here now Apr 29 20:21:50 Still the "invisible card" it seems with FirstUse Apr 29 20:27:33 I don't have the invisible card with PP Apr 29 20:28:27 Well I don't have it on Mido either, but with some clicking around I can get it if I reccall correctly Apr 29 20:32:41 Tofe: In /var/log/ there are some interesting debug logging with regards to resolution for QML Apr 29 20:32:52 https://bpa.st/JTOQ Apr 29 20:33:52 Mido has way too much logspam to do useful debugging on Apr 29 20:34:18 We might want to fix this one though in LSM: 2022-04-29T19:11:54.504424Z [20.479245652] kern.warning surface-manager [] surface-manager LSM {} (null), : appId is undefined. Adding an object with a undefined member does not create a role for it. Apr 29 20:34:20 Spams quite a bit Apr 29 21:20:20 ah, yes, WEBOS_WINDOW_BASE_GEOMETRY could be interesting, I'll investigate Apr 29 21:23:02 Herrie: https://github.com/webosose/qtwayland-webos/blob/master/src/plugins/platforms/webos-wayland-egl/webosscreen.cpp#L41= and I think it could be quite a good lead Apr 29 21:24:28 Isn't the DevicePixelRatio the same as the r value in the GEOMETRY string? Apr 29 21:25:00 not sure; but I'm going to try setting the geometry here to the correct value, and see where it gets me Apr 29 21:26:58 bingo ! Apr 29 21:27:50 and in the meantime I also got the invisible card while clicking in firstuse buttons Apr 29 21:28:46 so, how do we continue from there Apr 29 21:28:54 Tofe: Yeah that card is annoying Apr 29 21:29:08 Seems to be the volume notification app I guess? Apr 29 21:29:23 I'm just goign through the logs and see if I can kill some low hanging fruits here and there Apr 29 21:29:25 I don't know Apr 29 21:30:48 WebOSSurfaceItem(0x11d27b20, name="surfaceItem_default", parent=0x0, geometry=0,0 255x255) "" "_WEBOS_WINDOW_TYPE_CARD" Apr 29 21:30:57 ^ here the "" means it doesn't have any appId Apr 29 21:31:38 see for phone app: WebOSSurfaceItem(0x124c67e0, name="surfaceItem_org.webosports.app.phone_WEBOS_WINDOW_TYPE_CARD", parent=0x0, geometry=0,0 720x1440) "org.webosports.app.phone" "_WEBOS_WINDOW_TYPE_CARD" Apr 29 21:34:01 Interesting to see what would create a 255x255 window... Apr 29 21:34:06 Or that's a default value? Apr 29 21:34:16 My first guess was about a cursor texture Apr 29 21:34:24 But maybe I'm wrong Apr 29 21:35:50 https://github.com/webosose/qtwayland-webos/blob/master/src/plugins/platforms/webos-wayland-egl/weboscursor.cpp#L94= because of this Apr 29 21:39:20 Tofe: Well I've seen these in the logs: Apr 29 20:33:45 qemux86-64 webapp-mgr.sh[814]: [814:814:0429/203345.147431:ERROR:cursor_loader.cc(107)] Failed to load a platform cursor of type kNull Apr 29 21:39:24 Might be related or not Apr 29 21:39:57 I don't think it's related, because we don't have the issue with WAM (which is directly based on wayland, not qtwayland) Apr 29 21:40:55 it could also be a consequence of an incomplete patch for Qt 5.15 Apr 29 21:41:15 i.e. it tries to use a cursor with qtwayland, but not in the right way Apr 29 21:41:32 or maybe qtwayland was bugged before 6.0.0 regarding this Apr 29 21:51:31 disabling WebOSCursor::changeCursor seems to fix the issue on PP Apr 29 21:51:49 but not sure what it would do on qemu, where we actually want a cursor to appear Apr 29 21:54:38 Tofe: I'm happy to test if you have a paste Apr 29 21:55:45 https://paste2.org/zdhJEF6G here's a patch you can add to qtwayland-webos Apr 29 21:57:17 if it works on qemu, I can add it to the recipe, until we migrate to qt 6 (where I reasonably hope the issue will just disappear) Apr 29 21:58:52 Let me test Apr 29 22:01:43 Just it will want to compile a few packages, so will do a whole image Apr 29 22:01:46 Just as quick Apr 29 22:20:17 going to be, will see the result tomorrow ;) Apr 29 22:20:21 +d Apr 29 22:22:20 Tofe: Well it seems to work better Apr 29 22:22:29 FirstUse in qemu works, I can click etc Apr 29 22:22:33 Finish FirstUse Apr 29 22:23:08 Just it doesn't seem to finish FirstUse state, but that could be due to other changes I did Apr 29 22:23:11 I will check the loga Apr 29 22:23:29 The ghost window is gone and I can click at least Apr 29 22:30:45 Gn8! **** ENDING LOGGING AT Sat Apr 30 02:59:56 2022