Re: Pushing Forward Debian Mobile BoF 12-08-17 (updated with last notes)

2017-08-15 Thread Rainer Dorsch
Am Samstag, 12. August 2017, 23:31:21 CEST schrieb Sebastian Reichel:
> Hi,
> 
> Your Thunderbird creates broken GPG signatures.
> 
> On Sat, Aug 12, 2017 at 03:28:57PM -0500, jathan wrote:
> > Pushing Forward Debian Mobile BoF
> 
> Even though I'm in Montreal now for a Collabora company meeting, I
> unfortunately couldn't make it. Somehow Debconf always ends up being
> in parallel to the big European hacker camps (i.e. SHA2017 this year).
> 
> > -Openmoco experience
> > Difficulties to have again Debian on mobiles
> > 
> > armel? (Doesn't affect Pyra, does affect zerophone because its core is
> > raspberry pi)
> RPi2+ support armhf, but RPi Zero seems to be based on RPi1 SoC?
> 
> > (going through the agenda from https://wiki.debian.org/Mobile/BoF201708):
> > 
> > * Motivation
> > 
> >   * Software Freedom
> >   * (Security) Updates
> >   * usability for little scripts (always needs a full app)
> >   * better flexibility (Android is not Java-or-nothing)
> >   * access to software body (but UI adaptability is a big issue); still no
> >   good ODF editor on Android
> >   
> >   Audience scope is an issue -- whatever we come up with might not
> >   be suitable for everyday users. Sandboxing of an Android might
> >   be a topic.
> 
> chroot might become a problem with the next release, due to Android
> using old kernels. Glibc from stretch needs 3.2, which is already
> breaks support for some older devices. If that is increased in buster
> sandboxing will get much harder.
> 
> >   Which form factors are we talking about? Just phones? Tablets?
> >   How can we get improvements in touch interfaces to convertibles?
> >   (By the way: how accessible are the graphics modules on ARM to
> >   Free Software? It's easy for Intel-based GPUs)
> 
> If you are talking about acceleration:
> 
>  * Qualcomm Adreno freedreno   good
>  * Broadcom Videocore  vc4 good
>  * Nvidia Tegranouveau ?
>  * ARM Malilimaneeds lots of work
>  * ImgTec PowerVR  -   incomple RE, no driver
> 
> Worth to be mentioned: On ARM the GPU is usually not taking
> care of refreshing the screen. This is done by another hardware
> components and most of those are supported mainline (so kernel
> console and unaccelerated X work).
> 
> > Other use case: Devices for employees when companies have
> > extensive security requirements.
> > 
> > * Hardware
> > 
> >   * What is based on armel is doomed (probably won't be official
> >   
> > release in Buster, maybe not even in debports). Affects
> > zerophone, openmoko
> >   
> >   * Devices that work well and are armhf based:
> > * PocketCHIP (a PDA?)
> > * Pyra (a game console with possibility of phone calls; to be ready by
> > October)> 
> >   suffers from GUIs not being adapted to the input situation --
> >   doesn't really need a "windowing" window manager or desktop
> >   environment
> This should list Nokia N900 (mainline/debian kernel support is
> really good. Basically only camera support is missing and that
> is being worked on).
> 
> Also it might be worth pointing out Droid 4, which I'm currently
> working on together with Tony Lindgren. It's progressing quite
> nicely: https://www.elektranox.org/droid4/
> 

I suggest to add also Openphonux http://projects.goldelico.com/p/gta04-main/ , 
built on top of the roots of Openmoko, but also runs Replicant. Hardware is 
expensive and (wrt. modern smartphones) somewhat outdated though

Rainer

-- 
Rainer Dorsch
http://bokomoko.de/



Re: Pushing Forward Debian Mobile BoF 12-08-17 (updated with last notes)

2017-08-12 Thread Sebastian Reichel
Hi,

Your Thunderbird creates broken GPG signatures.

On Sat, Aug 12, 2017 at 03:28:57PM -0500, jathan wrote:
> Pushing Forward Debian Mobile BoF

Even though I'm in Montreal now for a Collabora company meeting, I
unfortunately couldn't make it. Somehow Debconf always ends up being
in parallel to the big European hacker camps (i.e. SHA2017 this year).

> -Openmoco experience
> Difficulties to have again Debian on mobiles
> 
> armel? (Doesn't affect Pyra, does affect zerophone because its core is 
> raspberry pi)

RPi2+ support armhf, but RPi Zero seems to be based on RPi1 SoC?

> (going through the agenda from https://wiki.debian.org/Mobile/BoF201708):
> 
> * Motivation
>   * Software Freedom
>   * (Security) Updates
>   * usability for little scripts (always needs a full app)
>   * better flexibility (Android is not Java-or-nothing)
>   * access to software body (but UI adaptability is a big issue); still no 
> good ODF editor on Android
> 
>   Audience scope is an issue -- whatever we come up with might not
>   be suitable for everyday users. Sandboxing of an Android might
>   be a topic.

chroot might become a problem with the next release, due to Android
using old kernels. Glibc from stretch needs 3.2, which is already
breaks support for some older devices. If that is increased in buster
sandboxing will get much harder.

>   Which form factors are we talking about? Just phones? Tablets?
>   How can we get improvements in touch interfaces to convertibles?
>   (By the way: how accessible are the graphics modules on ARM to
>   Free Software? It's easy for Intel-based GPUs)

If you are talking about acceleration:

 * Qualcomm Adreno freedreno   good
 * Broadcom Videocore  vc4 good
 * Nvidia Tegranouveau ?
 * ARM Malilimaneeds lots of work
 * ImgTec PowerVR  -   incomple RE, no driver

Worth to be mentioned: On ARM the GPU is usually not taking
care of refreshing the screen. This is done by another hardware
components and most of those are supported mainline (so kernel
console and unaccelerated X work).

> Other use case: Devices for employees when companies have
> extensive security requirements.
> 
> * Hardware
>   * What is based on armel is doomed (probably won't be official
> release in Buster, maybe not even in debports). Affects
> zerophone, openmoko
>   * Devices that work well and are armhf based:
> * PocketCHIP (a PDA?)
> * Pyra (a game console with possibility of phone calls; to be ready by 
> October)
> 
>   suffers from GUIs not being adapted to the input situation -- doesn't 
> really need a "windowing" window manager or desktop environment

This should list Nokia N900 (mainline/debian kernel support is
really good. Basically only camera support is missing and that
is being worked on).

Also it might be worth pointing out Droid 4, which I'm currently
working on together with Tony Lindgren. It's progressing quite
nicely: https://www.elektranox.org/droid4/

>   An always-difficult topic is modems.

The defacto standard for modem support is ofono, which supports
quite a few of them.

>   * Android hardware has android kernels (old in terms of Linux)
>   that were ported to it. Running pure debian (but with android
>   kernel) can, in some situations, run X, but nothing radio-ish.
>   (See Cascardo's work). Linaro is working on mainstream support
>   for arm64 devices. Collabora has seen some parts of the Android
>   graphics system into Linux.

This won't work once we start to use kernel features, that are
not part of those old kernels.

-- Sebastian


signature.asc
Description: PGP signature


Pushing Forward Debian Mobile BoF 12-08-17 (updated with last notes)

2017-08-12 Thread jathan
Hi. I send the last notes of todays BoF. Regards.

Jathan

-- 
Por favor evita enviarme adjuntos en formato de word o powerpoint, si
quieres saber porque lee esto:
http://www.gnu.org/philosophy/no-word-attachments.es.html
¡Cámbiate a GNU/Linux! http://getgnulinux.org/es
Pushing Forward Debian Mobile BoF

-Openmoco experience
Difficulties to have again Debian on mobiles

armel? (Doesn't affect Pyra, does affect zerophone because its core is 
raspberry pi)

(going through the agenda from https://wiki.debian.org/Mobile/BoF201708):

* Motivation
  * Software Freedom
  * (Security) Updates
  * usability for little scripts (always needs a full app)
  * better flexibility (Android is not Java-or-nothing)
  * access to software body (but UI adaptability is a big issue); still no good 
ODF editor on Android

  Audience scope is an issue -- whatever we come up with might not be suitable 
for everyday users. Sandboxing of an Android might be a topic.

  Which form factors are we talking about? Just phones? Tablets? How can we get 
improvements in touch interfaces to convertibles? (By the way: how accessible 
are the graphics modules on ARM to Free Software? It's easy for Intel-based 
GPUs)

  Other use case: Devices for employees when companies have extensive security 
requirements.

* Hardware
  * What is based on armel is doomed (probably won't be official release in 
Buster, maybe not even in debports). Affects zerophone, openmoko
  * Devices that work well and are armhf based:
* PocketCHIP (a PDA?)
* Pyra (a game console with possibility of phone calls; to be ready by 
October)

  suffers from GUIs not being adapted to the input situation -- doesn't 
really need a "windowing" window manager or desktop environment

  An always-difficult topic is modems.

  * Android hardware has android kernels (old in terms of Linux) that were 
ported to it. Running pure debian (but with android kernel) can, in some 
situations, run X, but nothing radio-ish. (See Cascardo's work). Linaro is 
working on mainstream support for arm64 devices. Collabora has seen some parts 
of the Android graphics system into Linux.

There is the PostMarketOS project; some work might be reusable between them 
and Debian.

  * Replicant, LineageOS

* Installation

  * Document how to get root / how to unlock? LineageOS and Replicant already 
have that.

* User Interfaces

  * DEs are pretty adaptable (Gnome 3, KDE), but applications are unadapted
  
  * Problem fields:
  
* Small screen (usually easy, just use the keyboard, but sizes can be 
adapted)
  * often it's just drop minimum window sizes / make scrollable
* Touch (harder -- making buttons bigger is insufficient)

  * Openmoko apps are still available? Also from N900 era (QML), 
Foxtrot/TangoGPS

  * Talk to upstream for patching: Might need user base to argue; coming from 
Debian might help.

Openmoko applications still being around shows that devs are still here.

  * Having a few devices that are affordable and current (like OpenMoko was) 
would give as a good start for devs to hack on it.

  * Kirigami is KDE's approach to building applications that run on 
touch/smallscreen as well.

  * Could we have a Debian-style project that produces a Lineage-like 
distribution?

Wouldn't even need to be out-of-Debian; wouldn't need to be Debian releases 
(Debian libhybris/Android in parallel to Debian GNU/Linux), work together w/ 
FDroid

Android SDK is packaged for Debian and also there was one part for 
device-side
Debian installations running under Android versions of Linux. Packaging more
parts of Android device-side software could lead to running Android apps on 
Debian.

signature.asc
Description: OpenPGP digital signature