[alicebot-archcomm] gone but not forgotten

[Noel Bush]

Kim, Tom and I agreed that my cleanup job would include excising all
unused code from Program D, and that task has nearly been accomplished.

In the process of going through all of this code in such detail, not
only have I gained a tremendously deeper appreciation for all of the
work Jon did to transform Program B, but I have also taken note of many
good ideas that were sitting in the code but not complete or not in use.
Many of these ideas were Jon's, and many were methods constructed by
predecessors in days gone by that were supplanted by newer methods.

All of this effort by Tom & Pedro (and now me) on the Program D code has
been with the aim of producing not only a clear and streamlined
reference implementation, but also a suitable base upon which developers
can collaboratively add new functionality or enhancements without
wasting time in following dead-ends, and with a better understanding of
what constitutes the Program D API.  In a week or so more I believe
we'll be at a stable place.

But it is a shame to think of the portions that were removed as "dead
ends".  Even though a number of the ideas now excluded were linked to
some heated debates on this committee, I think that we ought now to
think about drawing up a list of the "rejects" that might indeed have a
place in a reference implementation (and/or in the AIML spec) if they're
thoroughly thought out.

Really I think that the most fruitful of these ideas will be those that
concerned interfaces of the Alicebot engine to various other programs,
like email servers, application servers, P2P systems, etc.  During the
heyday of Jon's stewardship of this code, new capabilities and
interfaces were appearing at a dizzying pace.  I think most of us would
agree that the vast majority of those experiments were interesting and
promising.  The oft-times unpleasant process of "interrogating into
submission or reification", however, left many on the cutting-room
floor.  I do think, though, that the change in the landscape now is for
the better, as there is an experimental platform making itself known
(TinyAlice), which should provide an ideal prototype view of some ideas,
without impacting the stability of the reference code, giving all of us
a better basis on which to make judgments about what to include in the
reference implementation and/or AIML spec.

So I would like to know: which ideas do different members feel were lost
and should now be retrieved?  Can we construct a tentative Program D
roadmap that includes some of these ideas?