Re: Do we need "Mission-Critical" software? Was: What to Do?

[n_brunot@my-deja.com]

In article <9345qu$8r76@news.kvaerner.com>,
  "Tarjei T. Jensen" <tarjei.jensen@kvaerner.com> wrote:
>
> n_brunot@my-deja.com wrote
> >A very huge part of those packages is simply writing specifications, and
> >pragma import ...
> >Which requires some work in Ada, while you would have absolutely nothing
> >to do in C.
> >This is of very high cost.
>
> Which you recouperate during development and debugging. It looks like
overhead,
> but is really an investment which pays off handsomely when you use them.

That's usually right, and especially for big projects.
But nevertheless, this is a lengthy and tedious work, where mistakes can
be made.
For a lot of small applications, the cost keep most programmers away
from Ada and the investment is globally negative.

Like in any field, you should be able to choose an extremely easy but
less reliable way to program (this wouldn't be worse than C) and a more
complex and more reliable way to program.
The problem is that the easy way is today impossible.

Don't forget that if the basic component is in C, you won't make it more
reliable because you interfaced it in Ada ...
A lot of C components are extremely useful, and an immediate way to use
them would not make any harm.

Anyway nothing can justify that this work is not at least extremely easy
and made with a standard tool or feature of the language.
The benefits would be the same.
The cost would nearly disappear.

This shows up as soon as a great part of your work is interfacing
existing C libraries, and this is an enormous problem for the spreading
of Ada.
Most available components are in C, and it would be a nonsense to wait
for their Ada version before you can use them easily.



Sent via Deja.com
http://www.deja.com/