Re: NOACE- End of the road for Ada?

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On Thu, 17 Mar 2005 13:25:25 GMT, Marin David Condic wrote:

> adaworks@sbcglobal.net wrote:
>> 
>> 
>> It would be really cool to have a standard interface between Ada and
>> Matlab.
>> 
> Probably, but it would be even MORE cool if there was an entire Ada
> equivalent of Matlab, et alia. The whole bundle of stuff produced by The 
> Mathworks is pretty cool for the modeling world, but I think it has 
> problems when people try to bring it into production. It is very 
> tempting to say "Well, now that you've defined the control in the model 
> world and conducted a bunch of tests on it, why don't we just compile it 
> and take that code into the production box..." But the model world likes 
> to play fast and loose with all the things that might cause problems in 
> a production world. (For example, everything tends to be just a "real" 
> number - not much type checking going on. Or not much support for 
> low-level representations of data that might matter when the zeros and 
> ones actually connect to the hardware. Or any number of other complaints.)
> 
> So people like to just build these models then take the code into 
> production, but might it not be BETTER if the modeling tool had 
> capabilities aimed at the production world? Might it not be BETTER if a 
> modeling tool took into consideration the needs of and proven techniques 
> of embedded software development? So if someone designed such a modeling 
> tool and incorporated Ada-isms into it and had it generating Ada code, 
> and it provided BETTER capabilities to the guy designing models, might 
> that not create an interesting marketplace for Ada?

It absolutely true.

I can confirm that in the automotive area it would be a great breakthrough,
if something like above existed. Simulink lacks multi-threaded middleware
for deploying the models. Considering a wide range of complex protocols
which need to be supported (from CAN to TCP/IP), clearly there is an abyss
between the code generated by Simulink to the code needed. More to the
point, there is no chance that people drawing diagrams would be able to get
this communication code right. An Ada platform with a middleware taking
care of this mess would be technically unbeatable.

However, there is a problem of acceptance. People want C. Even more they
want C#. Lots of projects are doomed to fail before they change their
minds...

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de