series queries

["Richard C. Waters" <dick@ai.mit.edu>]

Rob,

I am very pleased to hear of your interest in Series.

It is definitely a mistake (probably for the reason you state) that
generators are declared to be functions.  The only place in the code
that I can find generators being declared as functions is in
ADD-PHYSICAL-INTERFACE where the line
    (push (list generator 'function) (aux frag))
should be
    (push (list generator T) (aux frag))

Unfortunately, I have only a very weak implementation of generators
because I consider them to be a peripheral aspect of Series.  You
could experiment with a function-based implementation, however, I
would not put that at the top of the list of things to do.  You could
look at the XEROX implementation which is more that way, however, it
is different from and incompatible with the description in CLtLII.

With regard to good integration, have you seen my MIT AI memo
describing Series, it contains some specific suggestions.  If you give
me your surface mail address, I will send you a copy.  With regard to
Let etc., I suggest just replacing the current definition rather than
shadowing it.  Alternatively, the handling of series could be moved
into a separate phase of the evaluator and compiler.

A final point: I am in the midst of adding a couple of important new
features to Series.  These are upward compatible from what it says in
CLtLII, but very useful.  The latest version of the code has some of
these features partially supported.  I hope to have this work
completed in a couple of months and suggest that you use that version
when it is completed.  It should stil be profitable for you to
explore how to entegrate the current version now, because the method
of integration will not change.

Feel free to send me mail or call me (617-258-6017) about any
questions you may have in the future.

		Dick Waters