**** BEGIN LOGGING AT Thu Jan 30 02:59:57 2020 Jan 30 08:42:27 is tg bridge ok? Jan 30 08:51:26 It was broken a while back Jan 30 08:51:30 Seems it stopped last week Jan 30 08:51:40 ohh Jan 30 08:52:21 rebuilding kernel for halium8 Jan 30 08:58:22 What did you change? Jan 30 08:58:49 disabled selinux and audit via param Jan 30 08:59:51 OK, that might help Jan 30 09:00:01 Still my kernel wouldn't panic so early on even with them enabled Jan 30 09:00:04 and deleted console Jan 30 09:13:16 nope nothing new Jan 30 09:31:39 Hmmz here it seems to be pretty much the same state Jan 30 09:31:50 Let me push the changes to android_build quickly at least Jan 30 09:36:25 https://github.com/Halium/android_build/commits/halium-8.1 Jan 30 11:28:07 So as I understand it, a Halium rootfs contains a "minimal" Android installation inside an LXC container, which seems like a very convenient state of affairs. Is it possible to run Android apps in this container? Jan 30 11:29:26 haha Jan 30 11:29:26 no Jan 30 11:29:36 dTal: it doens't include any java part of android Jan 30 11:30:12 I see. Is there any reason why it *couldn't*? Jan 30 11:31:23 dTal: Well why would you Jan 30 11:31:43 Because there are many irreplaceable Android apps. Jan 30 11:32:13 For this there are other solutions such as Anbox, Alien Dalvik etc Jan 30 11:32:48 dTal: basically you don't want to use the android you're using for drivers for apps, you want an android """emulator""", so that it still goes through halium compositor window manager & stuff Jan 30 11:34:19 That's a lot more layers on indirection, though. Jan 30 11:34:22 *of Jan 30 11:36:08 You need Android->Halium followed by Halium->Android, all working perfectly. Surely it's simpler and more reliable to use the Android subsystem you already have directly, instead of trying to run 2 simultaneously, plumbed through a non-android layer? Jan 30 11:36:45 dTal: Well SailfishOS does that to a certain extend I believe, not 100% sure Jan 30 11:36:53 They don't run Android in LXC Jan 30 11:37:00 it really depends on what you want to do/why you want halium Jan 30 11:37:06 UBPort, Plasma Mobile, LuneOS does it inside LXC Jan 30 11:37:08 if you keep everything from android, why use halium? Jan 30 11:37:20 if you just want a composer running as an android app... well just do that directly Jan 30 11:39:36 It sounds like what's needed is a surfaceflinger server, implemented as an x11/wayland client. Jan 30 11:40:25 What I want is a first class GNU/Linux experience and a first class Android experience, simultaneously. Jan 30 11:40:50 dTal for that you have Anbox I would say Jan 30 11:41:22 Does that actually work? Jan 30 11:41:35 Word on the street seems to tend towards 'no'. Jan 30 11:42:04 dTal: Well it's highly depending on the device & kernel version I'd say Jan 30 11:42:28 So 'no' then :p Jan 30 11:42:37 Having a device like PinePhone that doesn't rely on Android or other devices using mainline kernel makes it easier Jan 30 11:42:58 3.10+ kernel version should be OK-ish. Ideally 4.x kernel would be needed Jan 30 11:43:01 Of course. Would that all devices were like that. Jan 30 11:43:34 dTal: to me "first class gnu/linux experience" means that WM is a gnu/linux one like plasma Jan 30 11:43:39 you can't do that if you keep java android Jan 30 11:44:00 phh: yes that seems about right. Why not. Jan 30 11:44:41 dTal: because apps rely a lot of window manager apis implemented by android. like for instance chat heads or pip mode of apps Jan 30 11:45:22 dTal: and then they'll rely on android networking api to select between 4g and wifi, so that means you'll have to leave networking configuration to android, and that won't be gnu/linux networking Jan 30 11:46:54 and well pretty much everything in android is liike that Jan 30 11:55:25 Networking API? I thought that was just Linux API. Jan 30 11:56:10 Is there more to it than just Android rewriting routes from time to time? I thought all that was transparent to programs and apps. Jan 30 12:00:16 first joke of networking in android: Jan 30 12:00:38 when an app calls socket(2), the syscall is actually NOT done by the app, but by a specific daemon called netd Jan 30 12:02:02 this is done so that android can mark the packets as either going to wifi or mobile internet, without letting the app have access to mobile ims network Jan 30 12:02:16 (ims = volte) Jan 30 12:03:08 but yeah even configuring interfaces is a joke in android. good luck using network-manager with android. it'll run its own shitty dhcp client even if iface is already configured Jan 30 12:08:48 (to boot android over nfs i said fuck this shit and put android inside a netns/veth) Jan 30 18:38:59 vendor/qcom/proprietary/adsprpc/src/fastrpc_apps_user.c:1232: Error 57: open dev -1 for domain 0 faile Jan 30 18:46:53 * panioluka[m] sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/XWdYxySDheUElNTjQLqbRHOe > Jan 30 18:52:08 Panioluka[m]: hwservicemanager one I have too Jan 30 18:52:27 lol Jan 30 18:52:31 Unable to set property "hwservicemanager.ready" to "true": connection failed; errno=2 (No such file or directory) Jan 30 18:52:37 Which seems to be the main problem Jan 30 18:52:50 Not sure what causes it or how to sort it yet Jan 30 20:31:18 https://github.com/mer-hybris/libgbinder **** ENDING LOGGING AT Fri Jan 31 03:00:42 2020