torios team mailing list archive

[ml@xxxxxxxxxxxx]

On Tue, July 21, 2020 7:24 am, Israel Dahl wrote:
> My main question is going to be, what architectures?
>
>
> Mobile is the newest low-spec ubiquitous computer everyone uses.  How do
> we support cheap computing easiest?  Mobile, is pretty much the only answer
> to that.  And most mobile OS are not focused on low-spec.

Also, most mobile solutions (such as Android) are for arm processors not
x86.  That would mean two copies of everything to support both platforms. 
Even in the mobile market, they're switching to 64 bit.  Google is pushing
to drop support for 32 bit and is no longer creating llvm cross-compilers
32 bit for Linux systems that build Android tools.  So, it'll be just as
hard to support low end arm systems as is it is to support low end x86
systems.

I could see us trying to leverage the Android-x86 project or libhybris and
use some of the Android apps that will run on x86 machines.  That would
give us access to more apps/applications.  I don't think dropping support
for x86 and moving to arm is the best solution for supporting older
machines at this point in time.  Supporting both platforms is going to
complicate things a great deal as well.

We can either support old computers well or low resource mobile devices
and devices like Raspberry Pi well.  It would be very hard to split
resources and support both.  They have different needs and constraints.

> What about compiling a mobile OS using musl (postmarketOS for example),
> or perhaps figuring out a way to get a really low-spec WM that works with
> Weston working as a good phone/tablet/laptop/desktop UI?

So that would probably be WIO at this point ( https://wio-project.org/ )
as far as the windowing goes.

This project is already doing what you've mentioned minus the mobile support:
https://github.com/tpimh/nenuzhnix
They're using musl and Wayland.  I mentioned WIO and they're looking into
that as well.

I'm guessing postmarketOS isn't going to be a good project to try to get
involved with because I haven't had any luck contacting the developers
about it.  I have e-mailed the developer of nenuzhnix before.  I think
he'd be welcome to sharing resources.

There are also some Puppy Linux distro developers that are working with
musl.  I think they're concentrating on trying to get tinyxlib working
instead of Wayland though.

Do we want to go with Wayland or do we want to go with a tiny X server
(like TinyCore Linux and a few of the Puppy spins)?  I was looking at some
of the tiny X ports and xserver-xsdl (used by Android apps).  There's an
interesting port of K-Drive (part of Xorg) that works on top of SDL.  If
we could build that successfully, SDL2 works with KMS and it could handle
low level functionality like keyboard/mouse/video support.  Then we
wouldn't have to worry about maintaining that part of things with Xorg
once X Server development stops.  As mentioned, the other way to go is
Wayland and a project like WIO.  Most popular GUI frameworks will have
support for X11 or Wayland at least for now.  One last option is nano-x
instead of X or Wayland.  It'll be interesting to see what happens with
BSD systems and what they choose since they can't adopt Wayland as easily
as a Linux system.

I guess I've raised more questions than answers.  What does everyone else
think?  What's our target audience?  Which processors (x86, arm, 64 bit,
32 bit, etc.) do we want to support?  Which processors do we have access
to (old computers, old laptops, old Android tablets, Raspberry Pis, etc.)?
 What do we want our desktop environment to look like (Wayland, Xorg,
KDrive, nano-x, framebuffer, maybe something like Jide had, something else
altogether)?