[Jscheme-user] list* function (was: apply/2 vs apply/n ...)

[Hoehle, Joerg-C. <Joe...@t-...>]

Hi,

>>(define list* (lambda R (if (null? (rest R)) (first R) (cons (first =
R)=20
>>(apply list* (rest R)))))))
>>BTW, Tim, your definition of list* relies on (null? (rest '()) -> #t
>>which it is in Jscheme, but it is an error in Scheme in general.
>I think the only time you will get (null? (rest R)) throwing=20
>an error is if
>you call this as (list*) which maybe should throw an error???

I feel it would be The Right Thing when adding list* is to implement =
the Common Lisp behaviour:
http://www.xanalys.com/software_tools/reference/HyperSpec/Body/f_list_.h=
tm#listST

o list* takes at least one argument.
o The last one need not be a list, i.e. (list* x y) =3D=3D (cons x y)

That means, null? is not the correct end-iteration test for list*. =
pair? is the correct test to continue iteration (there doesn't seem to =
be the equivalent of the ATOM predicate in Scheme, which would be =
needed here).

I therefore propose:

(define (list* elt . more)
  (if (pair? more)
      (cons elt (apply list* more))
    elt))
{jsint.Closure list*[1,n] (elt . more)} : documents it needs at least =
one element.

> (list* 1 2 '(3 4 5))
(1 2 3 4 5)
> (list* 1 3)
(1 . 3)
> (list* 1)
1

>But, I wonder if we should make (rest '()) and (first '()) throw =
errors
>instead of returning ()????
I wouldn't object. I thought the Scheme standard mandated this ("it is =
an error in R5RS"). I was surprised to discover that it works in =
Jscheme (I believe some of the elf/ code uses this).

Regards,
	J=F6rg H=F6hle.
BTW, I didn't check whether list* is in some SRFI on list =
manipulations.