Re: If you were in charge of RISC OS...

David Bradforth wrote:
...what would you do with it?

Since I am just waiting for the result of an Eclipse "Recompile
The World" operation, I thought I could add my ideas...

Of course you have taken the fun out of it by providing
me with a limitless budget - the obvious use for such a
budget is to develop and manufacture a superfast ARM,
put it into cool and funky hardware, port RISC OS to it
and give it away for free.

So given a more restricted budget, these are the things
I would try to develop. You will find that I prefer
solutions from the technical side - I believe in building
the right technology first before selling shiny polished
products.

1.) A super-fast ARM emulator for the x86 platform. If you
have followed the developments in the Java world, and their
performance gains over the years with their HotSpot JIT,
you might see the potential. Nowadays, performance similar
to native code is often achievable.

ARM code (as used in RISC OS) is quite simple in structure,
similar to Java bytecode (in an abstract sense). So I think
that a really fast emulator would be possible. Which would
neatly solve one of RISC OS' biggest problems: being tied
to a low-performance processor architecture.

2.) Development tools. Look at what the GCCSDK developers
have achieved over the years with comparatively little
manpower. Think what would be possible with an always-up-to-date
GCC, a complete UnixLib, a complete Unix-like build environment,
shared libs, ready-to-use ports of commonly-used libraries etc.

3.) OS development. We all know the weak spots of RISC OS.
Many of them are solvable with relatively little (compared
to the development effort spent elsewhere) effort. We have
recently discussed the restructuring of the FS stack. Just do
it.

The "themed WIMP" developments are on the right track,
just a little more work and a capable designer to give
RISC OS a fresh new look.

Make RISC OS more portable. Port it to every piece of
ARM hardware you can lay your hands on to make it even more
portable. Porting RISC OS must be easier than porting Linux.

Give RISC OS the APIs it needs to create a common ground
for machine-specific implementations. Think USB, Firewire,
graphics, sound, scanning, printing, video, input devices.
This is essential to allow well-integrated solutions for
both native and emulated solutions. And easy porting of course.

4.) OS licencing. Give up the "shared source dual licencing"
scheme and use a liberal licence like MPL, MIT, BSD or Apache.

5.) Emulator development. RISC OS needs a "zero cost solution"
for people to try it out on their PC, no matter what hardware/OS
they run. And this solution has to be good. Well-integrated,
fast, stable.

6.) Hardware development. There are still people around who
don't want an emulated solution. Give them an IYONIX successor
with the IOP342. Should keep them happy for a few more years,
and would give enough speed potential to cater for today's
needs (browsing, video decoding).

Steffen

--
Steffen Huber
hubersn Software - http://www.hubersn-software.com/
.

Relevant Pages