Re: The future of Ada

[Gunther Dragoski <gunther.dragoski@nospam.vf.space.com>]

Marin David Condic wrote:

>
>
> We've got a very similar situation here, where we committed to the
> Motorola M68040 ten or so years ago only to find out years later that
> Motorola wasn't going to supply us with Mil Spec chips anymore.

I would think that wouldn't be too big of an issue with the use of industrial

components today. As a matter of fact Motorola still supports the 68040
up to 40 Mhz. Companies such as Thomson & Chip Supply can screen these
from -40 to +105 deg C. They also offer these services for PowerPC 603e
(check their roadmap website) in case you need to jump to next generation
embedded processor as 68K family is dead. Of course the real question is
how long the 603e will be supported before the G3 becomes the norm.

> . Actually, you're much more at risk long term for changes in
> hardware technology rather than software technology. As long as you've
> got the same hardware and a compiler, you can very cheaply keep someone
> maintaining it.

I agree here if you are using a higher order language (such as ADA) but keep
in mind that the processor upgrades due indeed result in software risks. For
example
if you upgrade from 68040 to lets say 603e processor how do you handle the L1

cache's. I find that timing analysis , task scheduling and code verification
to be complex
unless I turn off the caches. Its hard enough tracing 68040 code withs its
single , smaller
cache but trying dealing with both instruction and data caches of 16Kb.

> (Care to hear about our guys who are maintaining some
> _very_ old systems in an obscure assembler language because the program
> can't afford a hardware upgrade?)

Not really (just kidding !) I have been doing that for the last 5 years on a
20 year
old obscure assembler.  :)) However what I find is that the Software upgrade
and
certification is the prohibitive factor not the H/W design itself. Its cyclic
, every 2.5
years or so the same question keeps getting asked , how much to upgrade to a
68K
processor and guess what I pull out the same original detailed quote I did 10
years ago.
When they see the S/W conversion costs they turn and walk away (see you in
2.5 years)
thinking that somehow miraclously the cost will be reduced when they come
back to ask
the same question.


G.D.
Space Systems Engineering