You appear to have very neatly followed up on code presented by M-J. Dominus in The Perl Journal, Autumn 1997, Infinite lists in Perl. The article is an excellent primer on the subject. He starts from no math knowledge of streams, and builds to a brief discussion of data flow programming with lots of applications of streams (such as pseudo-random numbers, and "The Sieve of Eratosthenes," which is the oldest algorithm for computing primes).

His conclusions, and I quote:

Other Directions

The implementation of streams in Stream.pm is wasteful of space and time, because it uses an entire two-element hash to store each element of the stream, and because finding the n'th element of a stream requires following a chain of n references. A better implementation would cache all the memoized stream elements in a single array where they could be accessed conveniently. Our Most assiduous Reader might like to construct such an implementation.

A better programming interface for streams would be to tie the Stream package to a list with the tie function, so that the stream could be treated like a regular Perl array.

---
-DA
[download]