Some remarks to the reply of Erik Naggum concerning the destructive nature
of sort:

> this is indeed a strange way to think.  just because the side effect is
> different from your expectations, doesn't mean you can doubt the return
> value.  Lisp functions are usually defined in terms of their return values,
> and not in terms of their side effects.  (other languages have other ways.)

There are many Lisp-Functions which are only used because of their
side-effects and for which sometimes even the return-value is left open to the
Lisp-implementation (many I/O functions, but also destructive operations).
Apparently there are two kinds of destructive operations:
1) those, where the destructive side-effects are well defined and reliable for 
the user (example: nconc)
2) those, where  the destructive modification is only allowed (for
example for efficiency reasons) but left open to the Lisp-Implementation
and in this case the user cannot be sure about what is really happening

My 'strange way to think' might be that I thought that sort is of 
type 1 and not 2. The description of sort in CLTL2 is not very clear for me 
with respect to this question but apparently sort is considered to be of
type 2. 

> note that `sort' is unable to know where to store the return value, and
> that if `sort' is called on a list, there is no way to store a pointer to
> the first element.  if `sort' is called on a vector, it will obviously be
> able to return its value by side effect to the vector, but that is an
> accident.

Then why do other Lisp-Dialects (unfortunately I was only able to test it with
CLISP) don't have problems with my small example?
Perhaps there are other members of the mailing-list who are able to test
it on other Lisp-dialects.

Bjoern

*******************************************************************************
Dipl.-Inform. Bj"orn H"ofling	* Otto-von-Guericke Universit"at Magdeburg
Phone: +49 391 67 18055		* FIN/IIK	
Fax:   +49 391 67 12018		* Universit"atsplatz 2
				* D - 39106 Magdeburg (Germany)
*******************************************************************************
E-mail: <iik.cs.uni-magdeburg.de at hoefling>
WWW: http://www-ai.cs.uni-magdeburg.de/~hoefling
*******************************************************************************