Van Snyder wrote:
> I would prefer to introduce at least an integer-bounded interval
> type, with constructor i:j:k meaning exactly what it does now for
> subscripts, allow variables and define intrinsic assignment for the
> type, and allow variables (as well as constructors) of the type to be
> subscripts and arguments.
While I am not going to sign on the idea of a whole new type for this
purpose (we already have enough pains with our type system), the
general idea is great, in the sense of making 1:n:stride an object of
its own, or at least an entity of some kind that can appear in
different contexts, and in particular subscripts, arguments, rhs, and
even DO loop ranges.
Best,
Aleks
|