REPL hangs on some special mit-scheme code

[oiovoyo]

; filename : cons.scm
(define (cons x y)
  (lambda (m) (m x y)))

(define (car z)
  (z (lambda (p q) p)))

(define (cdr z)
  (z (lambda (p q) q)))

(car (cons 1 2))

(cdr (cons 2 (cons 3 4)))

I redefine cons car and cdr
type ,e for each procedure
REPL never return at (car (cons 1 2))
But it works while I run in command line as scheme < cons.scm

[kovisoft]

I tried to find out what's going on here. If I modified your definitions so that instead of redefining cons, car, cdr I defined a new cons2, car2, cdr2, then everything was okay. Then I played a little bit with different redefinitions for car and it seems to be that this also modifies the definition of car as used in the swank server (slime/contrib/swank-mit-scheme.scm). For example when using this definition:

(define (car x)
  (display x)
  (first x))

then whenever the swank server executed a car form, it also printed its argument in the scheme command line window (the xterm window that is running the scheme swank server in the background).

So I believe that redefining any form used by the swank server will have serious consequences and I don't think I can do anything about that. This seems to be a limitation of the MIT Scheme swank server. Also please note that the swank server code is not written by me, it is (lisp, scheme, clojure) is "borrowed" from SLIME.

[oiovoyo]

much appreciate that you response in detail.

I'v already avoided this issue by renaming primitive keyword in the following code.

Thanks a lot