Page 249

match

We'll use the unifier in unify.lisp. It handles circular bindings and uses a list of binding lists, the same as the prover in ddr.lisp,

var?

Always be nervous when code assumes some sequence is non-empty. Can you create a symbol that will cause var? to blow up?

Page 252

<-

Making <- a macro is cute but not very useful. It means there's no easy to load a set of rules stored in a variable.

prove

This function would be better if it were data-driven, with a table linking symbols to proof procedures. That way new special logical forms could be added without editing the basic definition of prove.

vars-in

Using union on the CAR and CDR is CONSful. Look at the definition of pattern-variables in ddr.lisp for a more efficient approach.

with-answer

The value returned for a variable with binding is not a very satisfactory if the variable is bound to another variable. See replace-variables in ddr.lisp for an recursive solution that gets to the bottom of a value chain, if possible.

Comments? Send mail to Chris Riesbeck.