Re: Ada student homework ?

[Robert Dewar <robert_dewar@my-deja.com>]

In article <3A0C0C98.EB19F7E8@cadwin.com>,
  Nicolas Brunot <n.brunot@cadwin.com> wrote:

> Unfortunately, I also think that notion of good is not
> necessarily the same for a software engineer and a software
> user.

Good is much too vague a term to use in the marketplace. Instead
you need to look at meeting various needs. If you sell model
trains, they need to work reasonably well, but if every now
and then they bounce off the tracks that's annoying but not
disastrous, and if it only happens once a week, it is not
clear that model train buyers will pay 10 times as much for
complete reliability (as an Marklin Z scale enthusiast, I
certainly wouldn't that stuff is expensive enough already :-)

SO model trains can be "good" without being 100% reliable

But if you now decide to build *real* high speed trains, the
criterion changes, and they had better be orders of magnitude
more reliable. So real trains definitely cannot be "good"
without going to the expense of achieving something close to
absolute reliability.

Some things have even less reliability requirements than
my model trains, somethings more than real trains (e.g. a
nuclear reactor control program).

So "good" has to be qualified by requirements. Ada obviously
has its greatest strengths where high realibility IS a
requirement. There is nothing wrong in Nicolas attempt to
duplicate Microsoft's success in the world of relatively
low reliability requirements, where (at least in Microsofts
case) having thousands of known problems does not inhibit you
from releasing software, but I don't think this is where Ada
can make its biggest strides.


Sent via Deja.com http://www.deja.com/
Before you buy.