> You make a good point about the Ada world being the way it is "because of
> the DoD". However, I consider the notion of "validation", perhaps in even
> a stronger form than we know it for Ada, to be a natural, necessary
> consequence of sound Software Engineering principles. I also note that
> virtually every other society-critical industry has, industry-wide, at
> LEAST as strong a notion of "validation" (translated into whatever makes
> sense in their domain). This is closely related to, but distinctly
> different from, the licensing issue. Society (unconsciously) demands that
> software, across the board, be validate before brought to market. The day
> will come when this "demand" will become VERY conscious, and heads will
> start to roll. (I, personally, would hope that software types will be in
> the driver's seat of whatever mechanisms are put into place to identify
> which heads. :)
But let's not be naive about how "validation" comes about. For example,
medical devices are "validated", but I surmise that any standardization
in this area comes (in the US) straight from the Food and Drug
Administration, that is, the US government.
Underwriters Laboratories (again, in the US) "validates" electrical
appliances. This is not the government, but the fire insurance
companies ("underwriters" are insurance companies), which definitely
have an interest in trying to assure that your toasters meet
"safety-critical" standards, that is, they don't burn your house down.
The point is that _standardizing_ testing/certification/validation
is only interesting to an industry group when it's more expensive
not to do it than to do it. A toaster wouldn't sell much in the US
without that UL label - people do seem to understand its meaning.
> It is an observable fact that software developers, in general, are just
> professionally irresponsible when it comes to things like "subjecting their
> products to someone else's validation suite". Although I don't pretend to
> fully understand this phenomena, I do understand quite a bit of it. There
> are many "sources" of this attitude, of course. But a lot of the
> underlying "psyche" here is precisely the same one by which so many
> software types become completely blind to the inadequacies of C-class
> languages, and are more concerned to "be a geek", "do their own thing",
> etc., than to be able to do good Math, Science, or Engineering.
You're being harder on software people than on other techies. I think
it's more realistic to say that we'll focus on this only when not
doing so either gets us in trouble with the government (the Ada
case...) or could really hurt us in the pocketbook (the UL case).
> This circumstance is entirely intolerable.
Well, yes, but not much worse for software than for (say) Ford/Firestone.
> Anyway, enough of the soap box . . . for now. :)
Right - stay tuned.:-)
> S. Ron Oliver, semi-retired professor of Computer Science and Computer
> Engineering. www.csc.calpoly.edu/~sroliver