Re: Some questions about Ada.

[Dave Jones <davedave@io.com>]

Matthew M. Lih wrote:
> 
> Dave Jones wrote:
> 
> > My opinion:  Ada95 is the best all-around language for software engineering.
> > C++ is awful.  Nevertheless, if I were starting a project today, I would
> > probably choose to use C++.  Why?:  More (and better and cheaper) tools are
> > available for C++, more programmers are trained to program in C++, et cetera.
> 
> Your statement perplexes me. If Ada95 is the best language, I would hope
> that means it's more "cost-effective" than other languages. 

By itself, Ada is more "cost-effective" than most other languages (by themselves).
However, there are some other factors to consider:  Is the institution funding your
project going to give you the time and money necessary to train programmers *properly*
in Ada?  Probably not.  Will the institution funding your project be happy about 
paying $3000 for an Ada development tool which is not as good as a C++ development 
tool that costs $500?  Probably not.  

I agree with those who say that, in the *long* run, using Ada would save a company 
money.  However, the bean counters that you rely on for your funding aren't 
thinking that way.  They want to save money *now*  (In fact, at most companies,
they *need* to save money now, or they could lose their jobs.  Sadly, most companies 
punish managers for thinking ahead, and reward them for "living for the day".)

> (Use C++
> for argument's sake.) But your statement effectively says this is outweighed
> by the tool factor and training factor, and other minor issues. I could
> argue against this, but I guess my question is really "How good does
> Ada have to be to outweigh the popularity of C++?"

An interesting question.  First of all, we have to realize that C++ and Ada actually
fulfill slightly different roles:  C++ is an improved version of C which was designed
to be a portable replacement for Assembler.  C and C++ have, in fact, fulfilled 
their design goals:  They are now the most popular languages for applications which
used to be developed in Assembler (shrink-wrap software, system-level programming, 
high-speed routines, etc.).   

Ada, on the other hand, was designed to replace languages like Jovial, Pascal and COBOL. 
 Ada has succeeded fairly well in more "technical" areas like the aerospace/defense 
industry.  However, in the area of business applications, Ada has largely lost out to the 
4GLs (here in the U.S. -- things are different in Europe).  The same can be said of other 
well-designed languages like Modula-3 and Eiffel.

For Ada to become as popular as C/C++, one of two things needs to happen:    

1) A "paradigm shift" will have to take place in the programming community:  Developers 
will have to realize that improved compilers and high-speed CPUs mean that there is no 
longer a need for a language to offer them the same power/flexibility that Assembler 
does.  They will then be more likely to concentrate on things like reliability and 
maintainability.  If this happens, then Ada will have a chance to compete directly with 
C/C++ on its home turf. 

2) Someone needs to develop a Delphi-like application based on Ada.  They will have to 
sell it at a reasonable price (less than $300).  This will allow Ada to compete with the 
4GLs.

-- Dave Jones
davedave@io.com