[Mep-dev] Tuesday Challenge for 14 April 2009

[Assi Friedman]

So instead of trying to tie the system to a specific architecture, why not shift the focus to developing the boundary interfaces of the system.  I.e. specify the behavior of the system rather than the operation. That way, you leave the implementation open for porting to many other systems.  Look at the approach crypto designers take. They design the behavior of the algorithm with considerations for a 8bit or 32bit processor and they leave the rest to the system implementers.  That way, to each his own...
Assi kk7kx


> -----Original Message-----
> From: mep-dev-bounces at uppermeadow.com [mailto:mep-dev-
> bounces at uppermeadow.com] On Behalf Of Michelle
> Sent: Thursday, April 16, 2009 7:25 AM
> To: mep
> Cc: Frank Brickle
> Subject: Re: [Mep-dev] Tuesday Challenge for 14 April 2009
> 
> 

> It's my understanding that developing MEP upon embedded linux means we
> must necessarily develop for one architecture at a time. For example,
> the TI OMAP.
> 
> 
> So, what are some reasons for considering Android for MEP?
> 
> My feeling is that porting MEP to as many processers as possible should
> be a very strong consideration, and here is why.
> 
>