srfi-13 string-any tail call

srfi-13 string-any tail call

[Kevin Ryde]

I'm thinking of the following wrapper around the string-any C code to
get a tail call on the last pred test, per the srfi-13 spec.  It's not
too pretty, but at least it'll be right.  I guess in the head it could
tie into eval now that string-any is in the core.

(define (string-any char_pred s . rest)
  (let ((start (if (null? rest)
                   0 (car rest)))
        (end   (if (or (null? rest) (null? (cdr rest)))
                   (string-length s) (cadr rest))))
    (if (and (procedure? char_pred)
             (> end start)
             (<= end (string-length s))) ;; let c-code handle range
             error
        (or (string-any-c-code char_pred s start (1- end))
            (char_pred (string-ref s (1- end))))
        (string-any-c-code char_pred s start end))))


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: srfi-13 string-any tail call

[Marius Vollmer]

Kevin Ryde <user42@zip.com.au> writes:

> I guess in the head it could tie into eval now that string-any is in
> the core.

Yes.  It would be nice in general to have a new kind of 'trampoline'
subr that returns a function+args to call in a tail-recursive way.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: srfi-13 string-any tail call

[Kevin Ryde]

Marius Vollmer <marius.vollmer@uni-dortmund.de> writes:
>
> Yes.  It would be nice in general to have a new kind of 'trampoline'
> subr that returns a function+args to call in a tail-recursive way.

Or func+args in some new magic type, so a subr can return either a
value or a new place to goto.

I was wanting `catch' to do a tail call to its handler so the handler
could recurse to continue on the rest of some data.  `catch' would be
a case of either returning a value (from the body) or the handler to
jump to.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: srfi-13 string-any tail call

[Marius Vollmer]

Kevin Ryde <user42@zip.com.au> writes:

> Marius Vollmer <marius.vollmer@uni-dortmund.de> writes:
> >
> > Yes.  It would be nice in general to have a new kind of 'trampoline'
> > subr that returns a function+args to call in a tail-recursive way.
> 
> Or func+args in some new magic type, so a subr can return either a
> value or a new place to goto.

Hmm, a value can be returned by using 'identity' as the function to
tail-call, but which method is more efficient is probably hard to say.
 
> I was wanting `catch' to do a tail call to its handler so the handler
> could recurse to continue on the rest of some data.

That would be very nice.

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel