Re: No Go To's Forever!

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

In article <sfke9p2pivv17@corp.supernews.com>,
  ka@exit109.com (Kenneth Almquist) wrote:
> Robert Dewar wrote:
> > But there really is a very fundamental point here. If you
want
> > to study it further, try generating the full denotational
> > semantics and in fact try generating the full proof
conditions.
> > That will make things clearer I think.


Incidentally, one more point, to construct the denotational
semantics of the original formulation with the goto is not
quite trivial. Obviously you don't want to describe the goto
as a general goto (the denotational semantics of general gotos
are gruesome, see the original Ada/Ed implementation for some
indication of what I mean). This goto is really a special kind
of loop construct, and indeed in a language with a bit more
general control over loops (e.g. SETL) you would not use a goto
at all for the outer loop, since it can be formulated more
clearly. Basically the goto here is a "repeat loop q" construct,
and that has a fairly simple denotational semantics. Indeed
in this particular case, you can of course get rid of the
goto by replacing it with a simple tail recursion, and the
denotational semantics should reflect such a replacement.

I sometimes wonder if people who have a hard time with gotos
in general also have a hard time with recursion. For me, the
goto that replaces a tail recursive call is a very special
case, and I regard it as simply a minor syntactic variation
on the tail recursive call (a good compiler will convert such
a tail recursion to the goto in any case). Yes, general gotos
can be a horrible mess, but the special case used here of a goto
that simply reflects a normal tail recursive structure seems
quite straightforward to me, and I see no more reason for
forbidding it than I would see for forbidding tail recursion :-)



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