Re: Ada student homework ?

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

In article <3A0C03BE.C3216454@cadwin.com>,
  Nicolas Brunot <n.brunot@cadwin.com> wrote:
> If you want companies to use Ada, don't tell them that
> Microsoft or Oracle are failures
> (Unless you prove you can make better than them, but in that
> case you would be as famous as them ...)


No, Microsoft is not successful in terms of quality, but in
terms of quantity. This is a perfectly valid goal, and indeed
is the road to riches. Ford was much more successful than
Bugatti, but I do not think people confuse this with the
quality of the cars.

Just because BMW does not have a 90% market share does not
mean that Ford makes better cars today.

If your goal is to have companies like Microsoft use Ada,
then fine, you can emphasize that Ada can hold its own just
fine in environments like this from a technical point of view,
but realistically, it will be non-technical factors that
govern the choice in an environment like Microsoft, where
software reliability is not a primary goal (time to market
and fancy features are much more important in that market).

Sometimes I wonder whether typical PC software is not a repeat
of the red delicious apple debacle (where the suppliers thought
that what consumers wanted was a nice looking apple and they
succeeded, unfortunately, people are not actually buying them
because they were also looking for a nice tasting apple. You
and I as tax payers are about to lauch one of the biggest
bailouts of its kind in history to rescue that industry).

In the PC World, people are simply accustomed to unrealiable
software that does strange things and breaks all the time. One
of the most remarkable bugs of all time was the bug in Windows
95 that caused it to realiable crash after a few days, but no
one noticed, because this crash just appeared to be normal
operation in the unlikely event that you could keep your
Windows 95 machine running that long.

Where Ada *is* succesful is in large scale projects requiring
high reliability. This is NOT an area in which the supposed
success of Microsoft in the software field is necessarily
such a clear plus :-)

A very interesting view of the future is that high reliability
software will become more important. For example, the ebay
application is certainly not safety critical, but it is worth
literally billions of dollars, in that if it fails, the stock
of ebay can take a hit at this level (and indeed has on more
than one occasion). The resources certainly exist at ebay to
create a completely reliable system. What does not exist is
an awareness that this is achievable.

Now of course Ada is not a magic bullet for achieving
reliability, sufficiently incompetent programmers can create
junk in any language, and sufficiently competent programmers can
create reliable software in absolute machine language (e.g. the
inner core executive of the ICL 1900 series machines).

But Ada, and the philosophy that surrounds Ada *can* be
significant aids in achieving realiability.

I think it is ultimately a more successful path for pushing
Ada use to emphasize the role that Ada can play in improving
reliability, rather than trying to make the point that it can
work well in the current world of get-it-out-the-door,
working-more-or-less-is-all-that-matters software delivery.
(a recent thread on whether "works OK" is the most important
aspect of software performance is relevant here).

Robert Dewar


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