At 10:18 AM 12/6/2002 -0700, Susan Bodily of Sandia Nat'l Lab wrote:
>The environment that you mention (uncertain and/or changing requirements)
>is one that I work in every day.  I have successfully and productively
>used Ada in that environment and find it much superior to C++, which I
>have also used.  It sounds like there may be a misunderstanding of OO
>and/or Ada.

I like to confirm the writer's sense of value of Ada when requirements are
not fully cooked.

A report that I prepared for next week's SigAda conference remarks:

"The context of development of a large experimental facility such as NIF
has distinct challenges for software engineers. It is not possible to
construct a complete set of requirements a priori: the facility itself is
being designed concurrently with the control system that will operate it.
This observation applies at every level of the organization of the software

At the hardware interface, numerous components of the laser are being
invented as the controls for those components are being constructed, so
only during integration testing of first units does the software developer
refine the details of device control.  The same condition occurs in
large-scale integration.  Since the techniques for operating the laser are
refined while the first beamlines are being activated, high-level
requirements emerge as preliminary operations elaborate the conditions for
successful exploitation of the innovative design."

I think the presentation (next Tuesday) by my colleague Bob Carey on the
operational experience using Ada in the National Ignition Facility will
encourage Ada-philes of the value of Ada in our very large controls project.