Fri, 18 Dec 1998 01:25:13 EST
[log in to unmask] quoted and then wrote:
>From: [log in to unmask] (Samuel Mize)
>Sender: [log in to unmask] (Team Ada: Ada Advocacy Issues (83 & 95))
>Reply-to: [log in to unmask] (Samuel Mize)
>To: [log in to unmask]
>Jerry van Dijk wrote:
>> > Safety-critical may be optional, but I have no interest in attracting
>> > Ada advocates who are not interested in high-integrity. It is quite
>> > possible to write lousy software in Ada, and the reputation of Ada is
>> > best preserved by not attracting those who want to go in that direction.
>> Are you _really_ saying that people like me should resign from this list
>> and stop using Ada and leave you high and mighty people who do the
>> really important work alone ?
>I can't speak for Larry, but I can tell you what *I* think.
I can :-).
>I think he overstated a very valid point.
>One normally thinks of high-integrity systems as being things like
>medical systems, flight controls, banking systems -- things that
>absolutely HAVE to work right all the time.
Whereas I think of a high-integrity system as one designed and constructed
to have those qualities to make it work right all the time. Whether it is
applied to controlling a nuclear reactor or to operating a video game, the
system can still be constructed in a high-integrity fashion.
Now those two extreme environments attach different priorities to whether
the system be constructed in a high integrity fashion. However there is
a continuum of applications in between the two end points. And for many
of those intermediate positions, different attitudes may be adopted by
organizations that might otherwise be judged to be in the exact same
I have a non-technical friend who noted that her bank's ATM machines had
breakdowns most often on Friday afternoon when demand is highest. That
a system should fail when operating at extremes is no surprise. Probably
the bank can do just fine with the US bank regulators so long as no money
is lost. So high-integrity is optional when it comes to the availability
of cash machine services.
Thus I believe that high-integrity is a fundamental property of the
and while different domains may have differing requirements for
there can be cases where one organization strives for higher integrity
Consider the case of developing commercial software products. To a
extent, the winner is the one who gets the fewest trouble calls. Those
can originate from unpredictable behavior, non-intuitive behavior or lousy
documentation. A high-integrity system is just part of the commercial
equation, but it can be the difference between profit and loss.
>Some developers take a fast-fast-fast development road: slap something
>together, debug it into running, test out the most obvious bugs and
>promise that the others will magically disappear for the next version.
>That isn't the kind of developer, or development approach, that Ada is
>really suited for.
>I think it's these kind of developers that Larry doesn't feel will add
>much to a discussion of promoting or improving Ada.
And I believe the discussion was about who we should attempt to _attract_
to using Ada. A campaign that was highly successful at attracting to Ada
the folks who are going to build sloppy systems anyway does nothing to
those people, and does harm to the reputation of Ada.