[Virtualacorn-list] Enhancing the RISC OS experience

[ralph_valmai at ntlworld.com]

Thank you, David, for a very encouraging answer.

In message <60e5bef04f.druck at druck.freeuk.net>
          "David J. Ruck" <druck at druck.org.uk> wrote:

> On 18 Oct 2008 ralph_valmai at ntlworld.com wrote:
>> 1) Would it be possible to use the host to render images that RISC OS
>> cannot do? RAW images might be an example. Are there any other host
>> resources that might be used in this way?
> 
> Yes, lots. Native image decoders could easily incorporated in to
> Select's ImageRenedering API. Similarly native video codecs could
> be utilised.
> 
>> 2) Would it be possible to write some modules using the host
>> instruction set and simply have the emulator pass parameters to the
>> module and receive the results? The modules given this treatment would
>> have to be such that a significant time saving would result.
> 
> Yes, there is the Red Squirrel (VRPC's orginal name before it was
> commercialised) SDK, which allows you to write native DLL's which
> can easily communicate with RISC OS modules. RISC OS modules can also
> make direct calls in to host OS APIs. VRPC's HostFS and Networking
> code already employ this mechanism, as I assume does RO 6.10's VRPC
> video driver.

It's even more encouraging if ROL are taking some initiative here.

> 
>> 3) What other uses of the host are there that might enhance the
>> overall performance of the VA in a significant fashion?
> 
> Apart from many issues similar to images and video, removal of
> emulation of legacy Risc PC hardware, and altering the RISC OS code to
> talk directly to hardware on the host system.
> 
>> It is possible that some of those who have left RISC OS might be
>> tempted back if VA became known as the RISC OS you knew and loved but
>> with enhancements that RISC OS on the old 'native' machines never had
>> and never could have had.
> 
> I doubt it, but it would make VRPC far more useful for those that
> already use it.

I can't say that I am terribly optimistic myself but it could just do 
something to keep us going longer and it certainly should be tried. I 
would love a VA that had hardware floating point, access to display of 
images of all formats as well as access to a wide range of video 
displays, a much higher speed than the emulator could provide in some 
speed-critical operations by use of host code and, who knows what 
else?

I'm probably getting a bit over confident here since you seem to have 
endorsed most of what I guessed to be possible, but what about 
porting, say, the PC version of FireFox, keeping most of the host code 
intact and simply providing a RISC OS interface to communicate and 
render the results on screen?

> 
>> Greatly improved disc i/o via hostfs is a significant selling point at
>> the moment. The more enhancements of this kind that can be
>> implemented, the more attractive VA could become and the longer RISC
>> OS could be available and useful.
> 
> The big question is whether any developers consider it worth the
> effort to utilise the host OS from within RISC OS, and any users
> would use it in preference to the more sophisticated applications
> running on the host system.

Clearly developers need, not just enhancement of RISC OS performance 
in the way that we have been considering, but also a large enough user 
base to make their efforts commercially worthwhile. The user base has 
dropped below this point already I guess, and it is far from clear 
that gee whizz developments of VA will change that.

But, if PC versions of valuable Linux applications could be ported in 
the same way that I suggested for FireFox, i.e. keep most of the host 
code and suppling ARM code only for a RISC OS compliant user 
interface, together with i/o and screen rendering, then maybe more 
users could be attracted and the user base increased to the point 
where developers would become interested again.

One can dream.

Thanks again for your response.

Ralph




-- 

Ralph & Valmai Stevens
In leafy Epping