Re: Re-Marketing Ada (was "With and use")

[Marin David Condic <nobody@noplace.com>]

Warren W. Gay VE3WWG wrote:
> 
> 
> I have never had a problem with multi language projects. This
> only becomes a problem when impedance mismatches occur on
> interfaces and objects, in my experience.
> 
Yeah, but if your compiler is C and all the libraries in your problem 
domain are C and your operating system is C and all your de3velopment 
tools are in C and all your programming expertise is in C, it gets 
*really* tough to argue that it is economical to do some project in Ada. 
Everything that diverges back to some other language is another reason 
to avoid Ada. Hence, the effort should be to make Ada as capable as 
possible to remove the impediments to using it.

> 
> If all I had to do was add a Makefile command:
> 
>    adagen freds_fft.h
> 
> to generate a thin binding, in a consistent and
> portable way, then I'd have mighty little resistance
> to using Ada in a given project.
> 

Yeah, sure, there is nothing wrong with having an automatic translator 
to help you get a binding. I'm not against that. I just don't think it 
is a) the final answer, b) the best answer or c) the answer that is 
going to persuade C hackers to come over to Ada. It *still* comes down 
to "I've already got everything I need with C/C++ so why do I want to 
add effort and expense to my project by doing some part of it in Ada?" 
There has to be some really compelling reason to do that.

Consider that if Ada had some really spiffy library that did some 
wonderful things for a given problem domain and that this was a major, 
new advantage over C++ in some way. Then someone says "I want to use Ada 
to gain access to those features." C/C++ probably won't bind to Ada real 
well - or at least they'd need to develop a binding in Ada that reduced 
things to a level that C could handle. (Give the caller all sorts of 
pointers instead of the data, etc.) So now you've got a good economic 
reason why someone ought to do some part of their development in Ada. At 
that time, the ability to bind to C becomes an asset to them.



> The tough sell is when you have to report to management that
> it will add 3 weeks of work to build a binding to Fred's
> FFT routine, in order to use Ada. Project management will
> then ask, "can it be done in C/C++?" As soon as you answer
> affirmative to this, in the GP (non-embedded, non safety
> critical) world, the project management answer will be to
> use C/C++ instead. Bye bye Ada!
> 
> If instead, you don't have to add those 3 weeks to the
> project schedule, project management might even encourage
> the use of Ada, if it even cares. Alternatively, you
> might not even have to bring it up ;-)
> 
> For Open Source developers, I think this can make a difference
> too. Everyone has a limited free time, unless you have
> just won the lottery (but then would you still be coding? ;-)
> 

Ada has traditionally focused in on life-cycle costs and the problem is 
that for *most* development situations, the whole life cycle is 
irrelevant. Its the Time To Market that is critical. Development 
projects have to get done *Quickly* in most cases because getting there 
first means market dominance or (in the case of spare-time developers 
and some other projects) there is limited budget to get the project to 
completion. Hence, Ada ought to do something to try to add that 
leverage. Show your garden variety C++ programmer that he can get his 
job done in half the time (and have lower maintence thrown in as an 
added bonus) and maybe you've got something he is willing to look at.

MDC


-- 
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jsf.mil/NSFrames.htm

Send Replies To: m   o   d   c @ a   m   o   g
                    c   n   i       c   .   r

     "Trying is the first step towards failure."
         --  Homer Simpson

======================================================================