Am Mi., 2. Sept. 2020 um 16:50 Uhr schrieb Wolfgang Corcoran-Mathe
<xxxxxx@sigwinch.xyz>:
>
> On 2020-09-02 15:14 +0200, Marc Nieper-Wißkirchen wrote:
> > In this spirit, I would also want to suggest
> >
> > (string-range STRING)
> >
> > a range constructor that takes a string and does the obvious thing. As
> > with vector-range, the argument mustn't be modified later.
> >
> > [snip]
> >
> > The nice thing is it is transparent for the user who just wants to
> > examine a string character-wise. He or she just has to work with
> > (string-range s) instead of s directly.
>
> Excellent points, I like it.  Added to my repository.

Excellent.

Would (range->string RANGE) also make sense then? It would be an error
if not every element of the range were a character.

(define (range->string range)
  (let ((s (make-string (range-length range))))
    (range-fold (lambda (i ch) (string-set! s i ch) (+ 1 i)) 0 range)
    s))

(No assume necessary because string-set! should have a type test for CH.)

In case, an implementation (with O(1) access to random string
elements) implements string ranges in a special way, the restriction
that the result of range->string mustn't be modified should be added
(as we have it for range->vector).

Marc