Comments on: SICP section 3.3.4

By: luca

luca — Sat, 16 Aug 2008 03:51:17 +0000

all right figured it out completely sorry for the useless posts

By: lukasjob

lukasjob — Wed, 06 Aug 2008 14:39:39 +0000

i’m sorry i got it now
only thing that (push c lcout) shouldn’t be (push c lcin)?

By: lukasjob

lukasjob — Wed, 06 Aug 2008 14:00:20 +0000

(defun ripple-carry-adder (la lb ls c)
(let ((n (length la)))
(unless (= n (length lb) (length ls))
(error “Expecting all lists of same length”))
(labels (
(ripple-build (la lb lcin ls lcout)
(unless (null la)
(full-adder (car la) (car lb) (car lcin)
(car ls) (car lcout))
(ripple-build (cdr la) (cdr lb) (cdr lcin)
(cdr ls) (cdr lcout)))))

(let ((lcin ‘()) (lcout ‘()))
(dotimes (i n)
(let ((ci (make-wire)))
(push ci lcin)
(push ci lcout)))
(push c lcout)
(ripple-build la lb lcin ls lcout)))))

excuse me sir i’m struggling to go through this book and sometimes it’s not easy I was trying to understand your function for ex 3.30 I was expecting something like the wire passed as c-out to full-adder used as the wire passed as c-in to the next full-adder in the sequence … something like a swap between recursive calls but couldn’t find anything like that … mumble mumble … *___*

luca

By: 3fen

3fen — Sat, 19 Jul 2008 12:36:44 +0000

The change in queue from LILO to LIFO can’t be simulated that way, I think.

The procedures are like this:
(define x (make-wire))
(define y (make-wire))
(define z (make-wire))
(and-gate x y z)
(set-signal! x 0)
(set-signal! y 1)
(propagate)

(set-signal! x 1) ; s1
(set-signal! y 0) ; s2
(propagate)

set-signal! will add some segments to the-agenda, and the signal of z will change only after (propagate) is done. So that after s1, something like (set-signal! z 1) will be added to the-agenda, and s2 will add (set-signal! z 0). Here is the difference: if the LIFO queue is used, (propagate) will run (set-signal! z 0) first instead of (set-signal! z 1). Therefor, the signal of z will be 1 while x=1 and y=0 at last. It’s not only a hazard, I guess.