I believe I now understand the history and current issues of having
zero-dimensional intervals and otherwise empty intervals.

I feel that multi-dimensional intervals with possible nonzero lower
bounds are different from the domains of vectors and strings, which are
one-dimensional and have a zero lower bound (so zero-length strings and
vectors are well-defined).

Because I cannot come up with semantics for either zero-dimensional
arrays or empty arrays that make sense for me, I propose to rename this
SRFI as "Nonempty Intervals and Generalized Arrays".

That way, an implementation can add zero-dimensional intervals and/or
empty intervals as a compatible extension to this SRFI.

Comments?

Brad