re: front & back end communication / was GUI ...

I wonder if the front end could perform selections etc. by using
commands like those from MAC-ED?

 (See: http://citeseer.nj.nec.com/kajler95survey.html pp. 4-5)

I gather (have not read code yet!) that what TeXmacs does is keep the
expressions in it's own internal document-tree form, then translate them
to Maxima form for I/O.  I'm not clear on whether it communicates using
Lisp s-expressions or Maxima command-line commands.  I would imagine
that s-expressions would allow more flexibility and control.

Just musing as to whether it's better in any way to have it just gather
what parts of the expressions it's referencing, then build selection
commands, vs. sending a copy of a selected subtree --- the only reason I
see for this would be to maintain "eq"-ness of those selected
subexpressions internally to Maxima... ?

I need to learn more about actually _working_ _with_ CAS.  What sort of
(real or fantasy) manipulations would the Ultimate CAS User Interface
support?  Only after listing those, deciding on how it should be
implemented can proceed.

Symaxx has an interesting concept.  So does "calc" embedded mode.  It
might be cool to have a Maxima session (restartable, notebook style)
there, but hidden or folded, with only a few selected expression nodes
visible in the final document, as display mode equations.

It might be useful to have the ability, as discussed in the above survey
of CAS user interfaces, to drag a term around, having the CAS adjust the
form according to correct algebraic laws.  It would also be useful to
drag variables into position, and have it maintain them in that order
for future results involving a particular expression.

Another useful feature would be the trace and root finding found on
graphing calculators.

-- 
Karl M. Hegbloom <hegbloom@pdx.edu>