**** BEGIN LOGGING AT Mon Feb 27 03:00:01 2017 Feb 27 06:59:36 marig: It would just do it like any kernel Feb 27 07:27:25 Morning! Feb 27 07:31:58 morphis_: ping ? Feb 27 07:33:15 Tofe: pong Feb 27 07:35:13 morphis_: (I just saw that I didn't have the log with me, silly me) I have a little permission issue with the camera with libcompat, maybe you already solved it on your side Feb 27 07:36:08 when I start test_camera, with camera_service started in the lxc container, I get this kind of message in logcat "Service: Waiting for permission android.hardware.CAMERA" Feb 27 07:36:22 sorry it's not the exact message, but I lost my logs this morning Feb 27 07:38:39 I guess you need to hack something inside the camera service Feb 27 07:38:48 we have some modifications there Feb 27 07:39:56 Well I think I have all the commits from frameworks/av -- I'll check again. Feb 27 07:40:24 It was just in case it reminded you something obvious, but if not, I'll dig in further Feb 27 09:05:24 Morning Feb 27 09:05:43 Tofe: YGM, probably you covered this but doesn't hurt to double check. Feb 27 09:29:36 Herrie|Pre3: thanks, yes, I saw this earlier, but the issue here seems to be that my Android userspace has no way of knowing if test_camera is allowed to use the resource or not Feb 27 09:30:18 Must be a missing permission file, or some missing piece of code in the Android servicemanager maybe Feb 27 09:31:08 Afaik, these permissions are set when you install a software that requires them (the famouse "allow permissions" dialog from Android) Feb 27 09:35:12 Ah, found the warning line: https://android.googlesource.com/platform/frameworks/native/+/android-5.1.1_r37/libs/binder/IServiceManager.cpp#110 Feb 27 09:39:07 ... and found the missing code: https://code-review.phablet.ubuntu.com/gitweb?p=opo/platform/frameworks/native.git;a=commitdiff;h=ee258d06e48f5504303afd000d3bc3a52104abd4 Feb 27 09:42:44 Tofe: So basically always return true :P Feb 27 09:42:49 That's good security LOL Feb 27 09:45:27 Well, they hook their own trust store on Ubuntu side; we'll have also to think of a way to restrict it a bit Feb 27 09:46:52 Tofe: Well we got the whole ls2 security bits. Feb 27 09:47:01 Where we can restrict it probably Feb 27 09:48:09 Herrie|Pre3: I'm quite annoyed by all these additional commits I have to cherry-pick on top of our Android code... First, it means we diverge from our upstream (here: mer-project/libhybris-12.1), and then, it also means we have to patch some Android system libs. And for the latter, didn't find yet a way to do it without directly patching /system... Feb 27 09:49:29 We'll see how it turns out, but right now it's not really a good point for the "overlay" approach Feb 27 09:51:00 Tofe: Yeah would be nice if we could integrate it as a patch in the bitbake build, but the ancient gcc prevents us from doing so it seems. Feb 27 09:55:18 Herrie|Pre3: the most annoying bit is that I didn't yet manage to start camera_service in lxc's init without actually modifying /system. I tried to make it load the .so from the hal-hybris directory, but unsuccessfully so far. Feb 27 10:10:47 Hmmz :s Yeah that's quite annoying Feb 27 11:19:04 Hi, i saw that you are searching for new JS framework for the feature development. I really like what the guys with AureliaJS are doing. For me it a better AngularJS, smaler, faster, uses modern JS and stands out of your way. Have a look at http://aurelia.io/ Feb 27 11:19:35 also here is a nice introduction of the framework and the concept and ideas behind: https://www.youtube.com/watch?v=fi33aDFKvxE Feb 27 11:36:04 MrTango: Looks pretty nice indeed. Feb 27 11:36:17 Just watching their 5 min video on aurelia.io Feb 27 11:36:27 Need to find some more time for the long YouTube one Feb 27 11:38:06 Herrie|Pre3, yes Eisenberg was also making some comperisons with React and Angular, i thing this might be even more interesting in the first. Feb 27 11:38:28 https://www.youtube.com/watch?v=gkbZ1LpWQB4 Feb 27 11:39:20 what i really like is, that they are really flexible, it's not the angular you have to dive into my universe way Feb 27 11:40:15 this is i think almost the same like the other video, but from anothe revent: https://www.youtube.com/watch?v=6I_GwgoGm1w Feb 27 11:48:56 We'd need to have a look if it features some of the things that Enyo does like swipe-to-delete etc Feb 27 11:49:25 Seems this is not very common in most frameworks or pretty hard to get working. Feb 27 11:54:06 Herrie|Pre3: I see Jolla are going to support Sony Xperia devices; and now I discover things like https://developer.sonymobile.com/open-devices/list-of-devices-and-resources/ which is indeed quite appealing Feb 27 11:55:10 Tofe: I've never been a big fan of Sony (dating back to all their proprietary stuff in 80's and 90's), but on Android and open source side of things they've been really good in recent years. Feb 27 11:55:22 If not the best right after Google probably. Feb 27 11:56:21 This raises even more the question about switching to AOSP, somehow. Feb 27 11:59:23 Tofe: Well AOSP should be fine for us in general. We only need a very stripped down Android base. AOSP would be fine for that imho? Feb 27 12:00:13 Yes; it's just that this would throw away the mer upstream we are using now Feb 27 12:01:51 Maybe it's a sign that I should try using Ubuntu touch's Android repos as upstream ;) Feb 27 12:02:17 (they are based on aosp) Feb 27 12:03:36 If the camera starts working, I'll give it a shot. Feb 27 12:27:34 Tofe: Well if Jolla's gonna support AOSP they might adjust their repos accordingly? Feb 27 12:28:42 Just Mer guys have been working more out in the open compared to Ubuntu I think with some thing. At least I remember the hybris guys not being happy about Ubuntu in the past, but that's years ago. Things probably changed. Feb 27 12:28:56 With morphis being on Ubuntu side :P Feb 27 12:30:19 Herrie|Pre3: well, things have changed a bit, but Ubuntu clearly also has its own agenda Feb 27 12:32:06 On the other hand, Ubuntu repositories are completely open, it's just that it's centralized on launchpad and we're not very used to it (not me, at least) Feb 27 12:39:56 I'm sure Launchpad has it's pro's but I prefer GH :P Feb 27 12:42:46 me too :) Feb 27 13:05:43 The 250M investment in Jolla by Chinese companies is nice too. Should hopefully help the OS development quite a bit. Feb 27 13:28:00 Oh, I didn't see the amount, quite nice yes! Feb 27 13:38:42 Herrie|Pre3: does the kernel auto-discover the amount of RAM when it's loaded? Or is it hard-coded for each device? Feb 27 13:41:47 marig: I don't really know, but my guess would be that there is an implementation for each kind of ARM board Feb 27 13:43:11 Tofe: I didn't see anything in the kernel config for the RAM amount. So I would expect it detects it? Feb 27 13:43:29 In meta-smartphone I mean Feb 27 13:44:52 Herrie|Pre3: oh that's definitely in the kernel itself, yes. It's either detected per-board, or maybe hard-coded on some boards Feb 27 13:51:31 marig: we can find things like this (i.e. C code, backed by asm code): https://github.com/shr-distribution/linux/blob/tuna/3.0/cm-12.1/arch/arm/mm/init.c#L135 Feb 27 13:55:15 marig: Why you want to know? You plan on modifying some legacy device :P ? Feb 27 14:31:52 Herrie|Pre3: Actually, yes, I am modifying a TouchPad with 2GB RAM. Feb 27 14:32:00 Herrie|Pre3: See - http://forums.webosnation.com/hp-touchpad/331349-2gb-ram-upgrade-touchpad-attempt.html Feb 27 14:32:50 I had to give up my attempts on this last year, but I'm trying to pick it up again. Feb 27 14:33:30 Where I left off was trying to understand if I need to change some config somewhere to have the extra RAM recognized, or not. Feb 27 15:47:10 hey guys, good job on the N5 port. i'll have to give that a shot soon. Feb 27 16:00:04 so it sounds like the BB KEYone will have the same locked bootloader as the priv Feb 27 16:00:13 there goes my hopes & dreams Feb 27 16:00:13 :( Feb 27 19:18:53 "I/ServiceManager(    0): Waiting for service SurfaceFlinger..." this is becoming a bit ridiculous :) Feb 27 19:23:59 marig: maybe you could share some boot kernel logs, e.g. klog in webOS? they might give some clues.. Feb 27 20:12:50 marig: that could very well be something like https://github.com/NewbyJE/android_kernel_hp_tenderloin/blob/cm-11.0_wb/arch/arm/mach-msm/hp/tenderloin/board-tenderloin.c#L169 ... or not :p Feb 27 21:12:06 Tofe: This is camera? Feb 27 21:12:29 Herrie: no, TP memory size (maybe) Feb 27 21:16:26 Tofe: Ah Feb 27 21:26:05 Herrie: No, *this* is camera. http://imgur.com/ytFEReQ Feb 27 21:26:55 :D Feb 27 21:27:12 That looks like solid progress :D Feb 27 21:27:34 yep, I'm quite happy with this :) Feb 27 21:28:15 This is actually just the test_camera program, from libhybris Feb 27 21:28:24 I've made it a wayland client, and tadaaa Feb 27 21:29:07 OK :) You're happy with the technical side of things as well? I.e. less hacky compared to earlier approach? Feb 27 21:29:20 ewww Feb 27 21:29:35 there has been quite some tweaking here Feb 27 21:31:21 I've imported a bunch of commits from ubuntu's android repos (frameworks/native and frameworks/av), I patched my native Android /system with some rebuilts libs, I've modified init.rc a bit (but through lxc hooks scripts, this one is clean) Feb 27 21:33:23 So I'm not sure how to package all this Feb 27 21:36:09 What I'm sure, however, is that libcompat is the way to go. When I see how easy it was (and generic, too, because none of this is device-specific), and when I see the struggles on SFOS ports to have a camera up and running, I think it was a good choice. Feb 27 21:37:41 Ah, and I forgot one annoying hack: I had to disable power.msm8974.so (i.e. rename the file) because it made test_camera hang. Feb 27 21:38:35 Tofe: Ah yeah libcompat makes sense Feb 27 21:41:55 I think there are two things still to solve: how do we apply these commits (do we fork libhybris' repos ? do we switch to using Ubuntu repos (or fork of them) ? do we switch to our own AOSP-based solution ? do we try to apply commits on-the-fly ?), and how do we make so that camera_service from hybris uses the rebuilt binaries instead of the vanilla ones. Feb 27 21:50:36 Well, that's all for tonight - it was a good evening :) Feb 27 21:53:13 Herrie, regarding AureliaJS and the swipe-to-delete, this is quite highlevel, it should be easy to build with it. Because the list ist just a javascript array and drag&drop or swipe is working in modern browsers. So one could build this specific feature as a aurelia component. Feb 27 21:54:13 and i thing for the rest things like the materialzed UI integration could work + some enyo/webos components on top of it. **** ENDING LOGGING AT Tue Feb 28 03:00:00 2017