**** BEGIN LOGGING AT Sat Aug 18 03:00:01 2018 Aug 18 07:45:59 @vanyasem, Its out Aug 18 08:02:07 @abhishek_01, Delete your message Aug 18 08:07:56 @abhishek_01, [Edit] Delete your message xD Aug 18 08:19:11 @Rafael, I am working on a Sony xperia z5 (sumire) port for ubports. The device manifest is already there.. Maybe you are more lucky if you just apply the diffconfig to the sumire_defconfig Aug 18 08:21:15 It is partially working.. The remaining issues are caf / xenial related it seems Aug 18 14:44:59 Hello again. I have received an answer from our potential ODM this week after I have asked for the source code of the Android implementation of Qualcomm, whose processors are used in their devices. … I thank the members of this Telegram group. I learned from you, that we need the ODM specific Android implementation to use Haliu Aug 18 14:44:59 m for the hardware adaption. For the members, who don't know me yet: My company plans a smartphone with a pre-installed alternative to Android and iOS. … The potential ODM wrote us, that we would have to pay for the Android source code and therefore we should specify which code we need. This is confusing, because we got the information from anothe Aug 18 14:44:59 r source, that Qualcomm has already published their source code at https://www.codeaurora.org. We also got the information that we need ODM specific drivers in addition. Can someone help us to specify the required source code we need for hardware adaption with Halium? … I also will ask this question in the Plasma Mobile group, because, this OS is o Aug 18 14:44:59 ne of our preferred candidates. Aug 18 14:53:39 @Marc Aurel, On top of paying for it they will also demand the signature of an NDA, so you will not be allowed to publish this source code Aug 18 14:55:13 So you might be able to use it for your own port, but you will not be allowed to publish this for others to use. Pretty much useless for Halium then. Aug 18 14:59:40 But they're still able to use Halium Aug 18 15:31:56 @Flohack, What's about the information, that Qualcomm already published their source code? … Anyway, if Qualcomm insists on keeping their source code private, we will do this. Why is Halium useless in this case? I expected, that we could use the open source project to adapt our selected Linux based to the hardware. … Our c Aug 18 15:31:57 andidates are Sailfish OS and Plasma Mobile and - taking recommendations of Telegram groups - Nemo Mobile and UBPorts. I learned, that we need a core Linux distribution preferably Arch for Halium, Halium with Libhybris that uses some components of Android, and an implemented UI Layer like Plasma Mobile, Lipstick (Sailfish OS, Nemo Mobile) or Uniti Aug 18 15:31:57 y8 shell with QtMir (UBPorts). … We need now a specification, which source code parts (or libraries?) of the vendor's Android we need to build Halium for the target hardware. … Please correct me, If I skipped important details for the ODM briefing. Aug 18 15:33:16 Of course we will need help from the community for hardware adaption. Maybe we'll find freelancers or developers we could hire for this process here in this community. Aug 18 17:01:58 DiMenar was added by: DiMenar Aug 18 17:16:43 On the topic of stack selection and distribution selection.. I would depend that on what kind user experience you are aiming for.. Aug 18 17:18:27 i. e. For an appliance of any kind I would never pick arch Aug 18 17:20:47 Wrt to source code those devices are full of programmable parts. You probably wont need all of it in source to get your os runnimg Aug 18 18:52:00 @Marc Aurel, Have you considered using NXP i.MX SoCs? Its drivers are open-sourced and pretty good supported in Linux kernel (though some components are currently out of mainline) and Purism contribute it for their Librem 5. It would be awesome if two GNU/Linux smartphones companies both collaborated on the same issues. In my op Aug 18 18:52:00 inion, Halium is inevitable measure for devices with nonfree firmware where you have no choice, thus there are many workarounds to get GNU/Linux working in a proprietary environment. Of course it's a great projects that deserves respect, but if you're building a new phone from scratch that is expected to be an alternative, I recommend you to consid Aug 18 18:52:00 er using pure GNU/Linux. Aug 18 18:53:36 @Marc Aurel, [Edit] Have you considered using NXP i.MX SoCs? Its drivers are open-sourced and pretty good supported in Linux kernel (though some components are currently out of mainline) and Purism contribute it for their Librem 5. It would be awesome if two GNU/Linux smartphones companies both collaborated on the same issues. I Aug 18 18:53:36 n my opinion, Halium is inevitable measure for devices with nonfree firmware where you have no choice, thus there are many workarounds to get GNU/Linux working in a proprietary environment. Of course it's a great project that deserves respect, but if you're building a new phone from scratch that is expected to be an alternative, I recommend you to Aug 18 18:53:37 consider using pure GNU/Linux. Aug 18 18:56:34 @Marc Aurel, I am not an insider when it comes to that but lets see how far you can go. Why I said its not working for Halium then is just bc Halium is an open source project. So we more or less prefer solutions that can be open-sourced Aug 18 18:57:58 By the way, the choice of the Linux flavor will be your last one. You got time until you need finally to decide. Plasma, UT and postmarket first need smth they can boot off. When you reached that point, you can decide what OS you really want. None of them has a better start bc all of them deal with the problems of closed source Andr Aug 18 18:57:58 oid hardware drivers. Aug 18 21:14:38 Hello there. What do I have to do after that? Aug 18 21:14:48 (Photo, 1024x768) https://irc.ubports.com/xPgO6nyd.png Aug 18 21:46:04 @ar7ch, I haven't. I just know, that Purism is working on GNOME Mobile as the core OS, but the information about GNOME Mobile are rare. … We would welcome collaboration with other companies working on an alternative OS. There are some ineresting projects, but most of them in an early stage or at least in a version, that is Aug 18 21:46:04 not ready for a commercial product. It's time for an alternative for Android and iOS, but this needs collaboration of companies, foundations and communities. … I'm wondering, why the hardware abstraction is not standardized yet. Even large copmanies are struggling. My hope is, that Halium could be this standardized abstraction layer. Many open sou Aug 18 21:46:04 rce projects are the base for commercial products. Even macOS is based of BSD. … My company would be happy to host a bootcamp or a similar event in Europe to bring parties together. If you know any interested contact, let me know. … Our potential ODM is willing to help us, but he also depents on his vendors like Qualcomm. Implementing drivers for Aug 18 21:46:04 a hardware seems to me a very expensive investment with man years of development, that is not our core business. … We also thought about the most simple way for our products using AOSP with a custom launcher like Blloc does. But it's not possible to implement an innovative UX wiht AOSP with system-wide gestures and a convergence use case for prof Aug 18 21:46:04 essional apps. There are great products like Sentio, but they are limited to Android apps. … Another alternative is Sailfish OS with professional support of Jolla. It's mature mobile OS with the flexibility to customize the UX with Lipstick. But this OS is far away from a convergence use case. … I you have any information or link about NXP i.MX S Aug 18 21:46:04 oCs, let me know. Aug 18 21:48:30 @Flohack, Looks like a long journey. Therefore I we start research for hardware adaption for our second prototype starge. This summer, we've started the first prototype stage by simulating the UX with apps for an evaluation with a focus group. But the convergence use case is difficult to simulare. Aug 18 21:58:03 @ar7ch, +1 … Marc , if you are developing a phone from scratch, why can't you request 2 versions of hardware drivers, 1 compiled for the libbionic systems, and another, for glibc systems. Halium(hybris) is just needed to run precompiled proprietary drivers on system with glibc. If you can get drivers working on the Aug 18 21:58:03 glibc, you do not need halium. Correct me, if I wrong Aug 18 22:00:52 @Marc Aurel, > I'm wondering, why the hardware abstraction is not standardized yet. Even large copmanies are struggling. My hope is, that Halium could be this standardized abstraction layer. … In overall, libhybris is de-facto standard for recent Linux mobile phones except Purism (Sailfish, Ubuntu Touch), but it's a huge hack d Aug 18 22:00:53 ue to the fact SoC vendors like Qualcomm don't do proper Linux drivers, so getting it as real standard is kinda bad thing. Aug 18 22:03:38 @Dzmitry Sankouski, He generally can't, at least without huge amount of money and direct connection to SoC vendor. … The situation is simular to the one with Gemini PDA. They have full source code of ODM, but there are libs that come in prebuilt state from SoC vendor, like Mali drivers, modem and so on. Aug 18 22:06:13 I'm walking away from haalium for a bit cause its starting to make me angry, hybris-boot isn't giving me a connection and dmesg says nothing about the device connecting Aug 18 22:06:51 @Tygerpro Tygerpro, Please wait until i get out of train, i promised to help you already Aug 18 22:06:55 No need to rage Aug 18 22:07:30 there are particular SoCs that currently have good mainline kernel support though, for example Snapdragon 820/845, but it's still going to require driver development for phone-specific components. Aug 18 22:08:06 [Edit] there are particular SoCs that currently have good mainline kernel support though, for example Snapdragon 820/845 (or NXP i.MX mentioned), but it's still going to require driver development for phone-specific components Aug 18 22:08:39 [Edit] there are particular SoCs that currently have good mainline kernel support though, for example Snapdragon 820/845 (or NXP i.MX mentioned + Rockchip/AllWinner), but it's still going to require driver development for phone-specific components Aug 18 22:09:00 I just need to walk away from it for a day or two before I teach my device how to fly Aug 18 22:29:54 @Dzmitry Sankouski, Soubd like a precise advice. … I can ask the ODM for these builds. In the end, I expect a system API for the next layer like ofono for the core smartphone features. Finally there must be a layer for the services and apps to access network, speaker, microphone, camera, sensors, usb, thouch gestures, … … Aug 18 22:29:55 Would these mentioned libraries enable these APIs for the next layer? Aug 18 22:33:24 Marc the issue is that ODM simply doesn't have ofono driver to build and so on Aug 18 22:33:36 [Edit] Marc the issue is that ODM simply doesn't have ofono driver which can be build and so on Aug 18 22:36:34 @NotKit, Our ODM suggest Snapdragon 825, but I would indeed prefer 845 for our use cases. Aug 18 22:37:43 I try to resume: … - Aug 18 22:51:35 I try to resume: … 1 … We need at least a device-specific kernel source code. OR … 2 … hardware drivers, one compiled for the libbionic systems and another for glibc systems. … The first option is used to build Halium, the second option is used directly by the secondary layer like Ofono? … Unfortunately I need to ask. My professi Aug 18 22:51:36 onal background are UI/UX concepts as well as app and service development with a focus on NLP, AI and ML. … In the end we need a stable core system with APIs to implement the high level services, the system UI and some main apps. … I hope, that we don't need to use the AOSP fall back, because, the customization of the UX is limited, allthough the Aug 18 22:51:36 launcher approachis very straight forward. Aug 18 23:02:26 Marc No, mine "at least a device-specific kernel source code" remark was about what we usually have when building Halium for existing devices released with Android (Google Nexus, Xiaomi, etc). This is accompanied by device tree used to built LineageOS, which is usually made from scratch by LineageOS community porters + extracted blob Aug 18 23:02:26 s from device. … In your case, there won't be LineageOS port for your device (unless it's made, of course), but you can use ODM's Android source code instead (which are how commercial devices with Sailfish/Ubuntu were done). … Hardware drivers in our discussions actually mean two different things: 1. kernel drivers 2. userspace libs, which talk to Aug 18 23:02:26 kernel drivers. Kernel drivers are open-source, while some userspace libs aren't. Libs generally are compiled for bionic. However this is not the only issue, as "common" hardware interfaces are different between Linux and Android. Aug 18 23:08:06 For example, on libhybris-based systems ofono generally talks to Android RIL daemon (there is an ofono driver which implements communication to RIL daemon as type of hardware for ofono). … But in case of "mainline" Linux ofono communicates with modem directly though interface provided by kernel. For example, on mainlined Nexus 5 qmim Aug 18 23:08:07 odem driver is used, which is part of oFono: https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/drivers/qmimodem. So in second case, ODM generally won't be able to provide you existing driver, and it's rather ofono code that needs to be adapted to work with your hardware. Aug 18 23:08:21 [Edit] For example, on libhybris-based systems ofono generally talks to Android RIL daemon (there is an ofono driver which implements communication to RIL daemon as type of hardware for ofono). … But in case of "mainline" Linux ofono communicates with modem directly though interface provided by kernel. For example, on mainlined Nexus Aug 18 23:08:21 5 qmimodem driver is used, which is part of oFono: https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/drivers/qmimodem. So in second case, ODM generally won't be able to provide you with existing driver, and it's rather ofono code that needs to be adapted to work with your hardware. Aug 18 23:17:33 (I suppose libhybris approach, while not particulary desirable by Linux community, is the most realistic if you do not want to go deep into hardware/drivers development due to ability to reuse existing solutions) Aug 18 23:24:08 @NotKit, So, potentially there are two options: One is using the ODM Android source code wiht implemented hardware drivers, the other is using directly hardware drivers implemented for Linux, which are not available for the most devices. … So we need to ask our ODM for the Android source code, needn't we? The ODM wrote us, Aug 18 23:24:08 that we should specify, which source code we need for hardware adoption, because the4 complete source cide would be very expensive. Aug 18 23:25:44 Hm, I wonder what they meant under complete source code. What you need to ask is Android source tree which can be used to produce booting firmware for your device, which is what you need even with AOSP launcher approach. Aug 18 23:25:56 [Edit] Hm, I wonder what they meant under complete source code. What you need to ask for is Android source tree which can be used to produce booting firmware for your device, which is what you need even with AOSP launcher approach. Aug 18 23:27:45 @NotKit, Indeed. I thought Halium is this kind of existing solution: We would get required libraries or source code from the chipset vendor compile a Halium build to provide the necessary system APIs. Aug 18 23:32:46 @NotKit, It was not my wording. I asked the ODM for the following: … "Source codes of the Android for the device, guidance how to build the source codes, the binary build of the Android that can be flashed to the device and the tools and instructions for flashing the image to the device." … The ODM answered: … "Could you p Aug 18 23:32:46 lease describe more specifically which source code parts are needed for which purposes?" … My intuition is a core headless AOSP with implemented drivers. Aug 18 23:35:36 your wording seems correct to me, you could ask them for Android source tree as if you wanted to compile custom Android build Aug 18 23:36:09 there is no core headless AOSP, getting normal firmware source code should be the easiest I suppose Aug 18 23:39:10 Thanks. I'll ask them for the Android source tree as a firmware source code. I appreciate your time for this great support. Aug 18 23:47:57 Android is broken and not configurable so when you build AOSP you generally get image that is booting to some kind of launcher, the day I see headless Android is the day I'm ripping my eyeballs out. Aug 18 23:50:18 @K31j0, Android Things? sorry for your eyes Aug 18 23:51:08 not really headless though Aug 18 23:51:09 the heck? Aug 18 23:51:32 They hacked it for zygote not to bring up surfaceflinger&co. on bootup? Aug 18 23:52:14 @K31j0, I thought you need an additional UI Layer to run AOSP, either the Google Layer with Google Apps or an alternative like LineageOS with alternative Apps and Launcher. Aug 18 23:53:03 @Marc Aurel, no, AOSP has basic launcher and some apps, which LineageOS were forked from Aug 18 23:53:07 Services are another thing Aug 18 23:53:21 @NotKit, LineageOS has their own apps as well Aug 18 23:53:23 @Marc Aurel, [Edit] no, AOSP has basic launcher and some apps, which LineageOS ones were forked from Aug 18 23:53:32 yeah Aug 18 23:53:35 for example AOSP music player hasn't been updated since froyo/gingerbread era Aug 18 23:53:46 same goes for FM Radio Aug 18 23:55:18 Does Halium support Kitkat Hybris? Aug 18 23:55:28 Kitkat-based Hybris adaptations, rather. Aug 18 23:56:14 currently branches of main manifest are 5.1 and 7.1 Aug 18 23:57:00 mer-hybris supports kitkat Aug 18 23:57:05 but dunno if it still works correctly Aug 18 23:57:18 I know it does work reasonably well Aug 18 23:58:32 there were times/devices that those old mer-hybris branches misbehaved Aug 18 23:58:43 like droidmedia on sony's shinano on 5.1 Aug 18 23:58:52 wouldn't work because of some codec issues Aug 19 00:00:43 Have there been issues with more recent hybris branches on the whole? Or is it better? Aug 19 00:01:42 guess it depends on the device, but I'd generally try the newest available first and if something doesn't work just move down the chain Aug 19 00:01:46 tried 7.1 on that same device, it is better, but the modem flopped now Aug 19 00:01:53 old kernels hurt Aug 19 00:02:04 (userland blobs even more) Aug 19 00:02:47 Well, the device I am porting to has the display powered by such a blob :) Aug 19 00:03:36 heck Aug 19 00:03:38 (Darn Texas Instruments and darn Barnes & Noble) Aug 19 00:03:46 nook? Aug 19 00:03:58 Yes Aug 19 00:04:15 I got mer-hybris to boot on it, for the hybris-11.0 branch. Aug 19 00:04:30 Which was mainly why I was interested in kitkat-based adaptations. Aug 19 00:04:30 never had it, but always seemed interesting to me Aug 19 00:04:51 They are extremely cheap and widely available on UK ebay Aug 19 00:05:12 Not from UK :P Aug 19 00:06:22 I need to revive my brother's Nexus 7 2012 Aug 19 00:06:37 It throws some libc exception after booting right after it gets to launcher Aug 19 00:06:45 I'm very scared it might be dying flash Aug 19 00:07:16 Those devices go "bad" after a while Aug 19 00:07:56 That's why they are so cheap, if I had a soldering station and another flash chip I could try to swap them Aug 19 00:35:36 6p **** ENDING LOGGING AT Sun Aug 19 03:00:01 2018