Fri, 24 May 2002 14:41:59 -0500
> OK, I'll agree with you that it's not always meaningless.
> It may be determinable at compile time by code analysis,
> but it can't be trusted at run time. I still don't see
> much use for it!
Well, it has to exist as all non-limited types have an "=" operation. It if
didn't exist, an access-to-subprogram type couldn't be used to match a
generic formal private type.
So to get rid of it, access-to-subprogram types would have had to be
limited. But there are no (really) limited elementary types, so that could
be a semantic problem.
So, it has to exist, and some reasonable definition has to be made. It
doesn't have to be useful. (We have the same set of issues with unchecked
unions, many of the operations are very meaningful, but have to be provided