Re: How to make Ada a dominant language

[Ed Falis <efalis@mediaone.net>]

Marin David Condic wrote:
> 
> I'll bet that using some version of this kind of iterative prototyping to
> approaching an ultimate product is done a lot more often than many of us may
> want to admit. It probably is also capable of producing good designs and
> reliable products in less time than more formal methods for some classes of
> problems. I doubt it is the sort of thing one would want to use to develop a
> major banking network system or a complex rocket launch control system (I've
> seen such things organically grown) but for some kinds of smaller, self
> contained problems, this is probably a good design methodology for the
> experienced developer. Lets face it - most of us think better in Ada than we
> do in UML or Mil-Std-2167a documents. That, and we like to experiment with
> different approaches to gain enlightment. (Think about it a little, hack a
> little code, run some tests, fix some bugs, go back to step one...) We don't
> generally want to draw a diagram, start writing the code, discover there is
> a better way to do it and be forced to decide to either redraw the diagram
> or stick with it and have a sub-optimal answer.
> 
> I'd consider the technique valid, but before declaring "Victory", I'd want
> to review the prototype(s) and take a decision about using it "as is" or
> using it as a model for a more robust design/implementation.

You might want to check out "extreme programming".  It's a method that
tries to add discipline to such an iterative process by combining
various practices synergistically.  It's not intended for large programs
involving hundreds of programmers, but has been used successfully in
efforts involving 20 or less.  And AUnit, the GPL-ed unit test framework
for Ada (at adapower) supports one of the key practices of the method -
test first programming.  You can start at www.xprogramming.com

- Ed