Re: string vs list processing

[Sascha Ziemann]

Sascha Ziemann <address@hidden> writes:

| The following example compares string-ref directly with list-ref:
| 
| (let* ((str (string-append (list->string (make-list 1000 #\A))))
|        (lst (string->list str))
|        (len (1- (length lst)))
|        (len/2 (quotient len 2)))
|   (compare-run-time 100000
|     (string-ref str 0)
|     (list-ref lst 0)
|     (string-ref str len/2)
|     (list-ref lst len/2)
|     (string-ref str len)
|     (list-ref lst len)))
| ;;   1.21 seconds for 100000 times (string-ref str 0)
| ;;   1.24 seconds for 100000 times (list-ref lst 0)
| ;;   1.24 seconds for 100000 times (string-ref str len/2)
| ;;   1.88 seconds for 100000 times (list-ref lst len/2)
| ;;   1.22 seconds for 100000 times (string-ref str len)
| ;;   2.58 seconds for 100000 times (list-ref lst len)

The calculation was not fair.  I removed the time spend for the
iterations and now the result looks different:

;;   0.35 seconds for 100000 times (string-ref str 0)
;;   0.34 seconds for 100000 times (list-ref lst 0)
;;   0.37 seconds for 100000 times (string-ref str len/2)
;;   1.03 seconds for 100000 times (list-ref lst len/2)
;;   0.36 seconds for 100000 times (string-ref str len)
;;   1.74 seconds for 100000 times (list-ref lst len)

But I think avoiding string functions is still a good idea.

bis später...
Sascha

-- 
Ein Hoch auf den 7. Juni 2000