segmentation fault with Maxima/ECL



Oliver Kullmann wrote:
> On Wed, Sep 03, 2008 at 11:37:49AM -0400, Raymond Toy wrote:
>> Stavros Macrakis wrote:
>>> Hi, Ray.  Do they support an arbitrary number of arguments, or do they 
>>> also have some arbitrary lowish limit (e.g. 1023 or 64k arguments)?
>> They all have limits, at least according to the value of 
>> call-arguments-limit.
>>
>> As mentioned in a previous email call-arguments-limit is:
>>
>> gcl	64
>> ecl	65536
>> clisp	4096
>> cmucl	most-positive-fixnum (536870911)
>>
> 
> just to mention: according to my experimentation,
> for "apply(union,...)" with CLisp the bound, if there
> is any, must be rather high (10^7 arguments worked
> on all machines I have).

This could be caused by the fact that union is defined like so:

(defun $union (&rest a) ...)

So $union is never given all 10^7 args separately.  It receives a list 
of 10^7 elements.  May also imply some smarts in (Lisp) apply or the 
compiler as well.  Not sure.

> 
> By the way, according to my experimentation the bound for
> CLisp is 4095 (not 4096).

call-arguments-limit is the exclusive upper bound.

Ray