**** BEGIN LOGGING AT Wed Feb 23 02:59:57 2022 Feb 23 09:40:22 hideki0[m]: are you following the instruction from https://github.com/webOS-ports/meta-pine64-luneos ? Feb 23 09:42:38 and, yes, you can also use "dd" with a regular wic image Feb 23 09:43:01 I just thought we still used the "bmap" wic image format, but maybe not Feb 23 09:43:58 Herrie: ^ did we change the default output image format ? Feb 23 09:45:01 https://github.com/webOS-ports/meta-pine64-luneos/blob/honister/conf/machine/pinephone.conf#L27 ah ok, we create both, but you uploaded the wic.gz; fine Feb 23 10:06:58 Tofe: Yeah that's all I use from my side usually Feb 23 10:07:19 Next time can upload both Feb 23 10:08:13 HerrieTP_: oh, don't bother, it's really fine Feb 23 10:24:37 Tofe: BTW seems SFOS is getting to a newer oFono state finally :) https://github.com/sailfishos/ofono/pulls Feb 23 10:24:43 piggz has been busy Feb 23 10:37:10 Tofe: do you have a sec to discuss lvgl? Feb 23 10:37:18 it's me :) Feb 23 10:37:50 Guest7223/JaMa: a bit later maybe ? I'm out for lunch in a couple of minutes Feb 23 10:39:26 sure Feb 23 11:30:47 JaMa: ping Feb 23 11:33:18 was the recipe in meta-pine64 imported from some other layer? Feb 23 11:34:52 the build issue I've seen yesterday with the meta-oe recipe was just the lv_conf.h location where your recipe correctly installs it directly in includedir while meta-oe recipe creates lv_conf.h from the template (with 2 sed replacements) but then installs it into includedir lvgl (where it's not found by lvgl) Feb 23 11:35:14 JaMa: good question, not sure about that... Feb 23 11:35:26 I think so Feb 23 11:35:52 and after fixing the location there are many more build error https:dpaste.comYT9R76QV possibly due to API changes in version 8 Feb 23 11:35:57 https://dpaste.com/3YT9R76QV Feb 23 11:36:22 sigh the other irssi client refuses to type '/' even when copy&pasting a link.. what's going on :/ Feb 23 11:37:25 as well as main.c:2:10: fatal error: lv_drivers/display/fbdev.h: No such file or directory Feb 23 11:37:47 the meta-oe recipe is using gitsm:// so in theory the drivers from submodule should be installed as well Feb 23 11:38:01 ah, could be lvgl dropped fbdev ? Feb 23 11:38:11 on the other hand I've noticed that your recovery-ui defines lvgl as submodules as well Feb 23 11:38:39 was it for easier development locally as I don't think it's checked out by bitbake during recovery-ui build Feb 23 11:39:12 Yes, it was much easier that way, and I could also play with it on desktop Feb 23 11:39:51 https://github.com/lvgl/lv_drivers/blob/master/display/fbdev.h still exists Feb 23 11:40:00 ah, good :) Feb 23 11:40:59 but the meta-oe recipe doesn't do anything to install these headers (I haven't checked if CMake is supposed to do it automatically when gitsm:// checks out the module) Feb 23 11:41:02 https://git.openembedded.org/meta-openembedded/commit/?id=21580d65af95a854cccc7d0bf5c31201c9c30ea9 Feb 23 11:42:30 and I don't know how much you customized lv_conf.h from the template as it was in version 6, the meta-oe recipe adds separate variable to control LV_MEM_CUSTOM Feb 23 11:42:42 hmm and there used to be EXTRA_OECMAKE += "-Dinstall:BOOL=ON" Feb 23 11:42:47 https://git.openembedded.org/meta-openembedded/commit/?id=b6738f86f3dd1d5208abdb455d10f9d16e198dd2 Feb 23 11:43:30 so long story short as you wrote recovery-ui, when you have some spare moments can you check the migration to lvgl 8? :) Feb 23 11:43:47 JaMa: yes, I'll do the migration :) Feb 23 11:44:03 as even if I get it building I won't be able to test it, so better you do both and I concentrate on some other build issues Feb 23 11:44:12 ok, thanks :) Feb 23 11:44:48 It isn't a very mature component yet, so lots of stuff is quite naïve Feb 23 11:45:26 but the documentation online looked much better than expected Feb 23 11:54:48 yes, it's a quite nice toolkit actually Feb 23 19:30:10 ahah, silly me: I tried to remove luna-next from the luneos-image, and succeeded, but now the cardshell can't import the QML modules LunaNext.Common and Shell :p Feb 23 19:31:22 I'll separate the luna-next recipe into two packages: compositor and qmlplugins Feb 23 19:34:03 or maybe luna-next and luna-next-qmlplugins; well, anyway. Feb 23 22:07:24 Herrie: I'd the very basics are now there for LSM + WAM on luneos Feb 23 22:07:44 Tofe: Nice Feb 23 22:08:00 I still have to fix WAM's systemd service, but otherwise the VirtualBox boot works Feb 23 22:08:21 I can start web apps (services are a bit off though) Feb 23 22:08:26 I asked in the Skype group if LGE has any examples or documentation for LSM and also for enhanced ACG Feb 23 22:08:49 Tofe: Well that's already a very good start Feb 23 22:09:09 HerrieTP_: I'll do the WAM change, and then PR it to your branch Feb 23 22:09:36 Tofe: Great! Feb 23 22:10:00 I'll be back in the weekend so will have some more time to test things at my end Feb 23 22:11:28 sure, no hurry Feb 23 22:14:19 How is LSM as a compositor from architecture side? Modular and easy to extend? Feb 23 22:14:28 Or you have any concerns there? Feb 23 22:15:12 HerrieTP_: so far I find it quite nice Feb 23 22:15:22 well organized, feels solid Feb 23 22:15:50 and yes, should be easy enough to extend, like auto-luna-surfacemanager did Feb 23 22:16:00 Though we won't need much Feb 23 22:16:18 HerrieTP_: could you give me write access to https://github.com/webOS-ports/wam ? Feb 23 22:16:29 will be much easier to modify the service on-site Feb 23 22:20:35 Tofe: OK Feb 23 22:23:21 Tofe: Done Feb 23 22:23:26 Do you need any repos created as well? Feb 23 22:24:32 no no, that's fine, I did the little change and now I'll bump the SRCREV Feb 23 22:24:34 For the split of the plugin? Feb 23 22:24:53 for luna-next ? no, I did that in the recipe Feb 23 22:29:07 https://github.com/webOS-ports/meta-webos-ports/pull/493 ok, here goes nothing, still a WIP of course Feb 23 22:30:46 Tofe: OK, I guess you didn't try this with Chromium 91 yet right? Feb 23 22:31:14 I did Feb 23 22:31:19 works fine Feb 23 22:31:58 Maliit doesn't show up yet, though it's started; not sure exactly where it fails Feb 23 22:33:52 Could be we need to create the surface somehow Feb 23 22:34:08 We might need to have a peek at OSE's Maliit Feb 23 22:34:25 Could be; and there are several changes in OSE's fork too Feb 23 22:35:10 Their changes seem to allow multiple input devices Feb 23 22:36:39 ah, I seem to have made a little mistake in luna-next.bb: I wrote RDEPENDS:${PN}-compositor = "..." but I didn't actually put files in ${PN}-compositor , or create that package Feb 23 22:38:35 ok, fixed Feb 23 22:40:18 So it looks like Enyo2 is going to be the way to go for new LuneOS app development -- lots of compromises with Enyo that Enyo2 solves. Feb 23 22:41:01 I've heavily modified the "bootplate" for Enyo2 to make it easy to create and deploy Enyo2 apps to LuneOS, webOS, Android and the web Feb 23 22:41:43 I'm still concerned about CORS. Enyo2 uses Cordova to run on Android (and an old version of Cordova to run on LuneOS and webOS) Feb 23 22:42:26 and Android CORS is a *huge* problem for Enyo+Cordova, because it uses the WebView control, which is not allowed to bypass CORS. Work-arounds out there are awful. Feb 23 22:43:18 Looking forward to seeing if CORS-hell doesn't exist on LuneOS Feb 23 22:43:44 Here's my new bootplate, btw. The build script is *fire* https://github.com/codepoet80/enyo2-bootplate Feb 23 22:45:02 Tofe: Our maliit fork is quite old, we might see if using OSE's fork with our changes ported might be an option Feb 23 22:45:57 HerrieTP_: yes, I'm seriously considering that too Feb 23 22:46:24 codepoet: isn't Enyo2 already 10 years old ? Feb 23 22:47:45 Tofe: Yeah and discontinued as well Feb 23 22:47:51 Replaced by EnactJS Feb 23 22:48:14 Which seems pretty good and solid, but needs mobile theme Feb 23 22:49:08 Yep Feb 23 23:06:04 and Enact won't run on legacy webOS Feb 23 23:06:21 so Enyo2 seems to be the sweet spot of running on old and new Feb 23 23:06:44 when LuneOS runs as well as webOS on my Touchpad, I'll consider switching to a newer Javascript framework ;-) Feb 23 23:07:54 funnily enough, the bootplate includes some LG-TV-style widgets and theme elements Feb 23 23:31:38 Enyo2 was used on TV's before EnactJS Feb 23 23:34:48 still is in some apps Feb 24 01:43:40 ya, i have the sampler running here: http://www.webosarchive.com/enyo2sampler/ Feb 24 01:43:53 if you choose the moonstone samples, it totally looks like my TV UI **** ENDING LOGGING AT Thu Feb 24 02:59:56 2022