(DEFINE (EVAL EXP ENV)
(COND ((ATOM EXP)
(COND ((NUMBERP EXP) EXP)
(T (VALUE EXP ENV))))
((EQ (CAR EXP) 'QUOTE)
(CADR EXP))
((EQ (CAR EXP) 'COND)
(EVCOND (CDR EXP) ENV))
(T (APPLY (VALUE (CAR EXP) ENV)
(EVLIS (CDR EXP) ENV)
ENV))))
(DEFINE (APPLY FUN ARGS ENV)
(COND ((PRIMOP FUN) (PRIMOP-APPLY FUN ARGS))
((EQ (CAR FUN) '&PROCEDURE)
(EVAL (CADDR FUN)
(BIND (CADR FUN) ARGS ENV)))
(T (ERROR))))
(DEFINE (EVCOND CLAUSES ENV)
(COND ((NULL CLAUSES) (ERROR))
((EVAL (CAAR CLAUSES) ENV)
(EVAL (CADAR CLAUSES) ENV))
(T (EVCOND (CDR CLAUSES) ENV))))
(DEFINE (EVLIS ARGLIST ENV)
(COND ((NULL ARGLIST) '())
(T (CONS (EVAL (CAR ARGLIST) ENV)
(EVLIS (CDR ARGLIST) ENV)))))
For VALUE and BIND see Figure 3.
Figure 5
Evaluator for Treating Procedures as Objects
Another good thing about this version of the interpreter is that the gross non-modularity of the scattered occurrences of PROCEDURES has disappeared. The problem has not been solved, of course, but we certainly feel relieved that the particular manifestation has been removed!
By the way, we also eliminated the explicit tests for T and NIL in EVAL, assuming that we can simply put their initial values in the initial environment provided by DRIVER.
An interesting property of this interpreter is that free variables now have been given a meaning, though we originally did not intend this. Indeed, in the original recursion equations interpreter, there were free variables in a sense: all procedural variables were free (but they could be used only in operator position in a combination). In our new
