Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!decvax!bellcore!petrus!scherzo!allegra!princeton!orsvax1!pyrnj!caip!topaz!harvard!h-sc1!breuel
From: breuel@h-sc1.UUCP (thomas breuel)
Newsgroups: net.micro.amiga
Subject: Re: shopping frustration
Message-ID: <1029@h-sc1.UUCP>
Date: Thu, 3-Apr-86 23:13:35 EST
Article-I.D.: h-sc1.1029
Posted: Thu Apr  3 23:13:35 1986
Date-Received: Sat, 5-Apr-86 07:47:19 EST
References: <5232@alice.uUCp>
Reply-To: breuel@h-sc1.UUCP (thomas breuel)
Organization: Harvard Univ. Science Center
Lines: 45

|||one of the demos we did try was amigalisp 1.0.  it at least could handle
|||(cons 'a 'b), but had an infinite recursive loop when we attempted to
|||execute (fib 2) (or (fib N) for N>1 for that matter) with fib defined
|||as
||| 
|||(df fib (n) (cond ((zerop n) 1)
|||                  ((eq n 1) 1)
|||                  (t (plus (fib (sub1 n)) (fib (difference n 2))))))
||
||I have not played with AmigaLisp, but you should probably write
||'(equal n 1)' rather than '(eq n 1)'. Most Lisp implementations
||only guarantee atoms to be unique (your code would, however, work
||in Franz Lisp -- by what I would call a coincidence :-).
|
|until a few days ago i had not encountered a lisp in which eq did not
|work on numbers.  of maclisp, interlisp, uci-lisp, franz lisp, etc, and
|amigalisp, i would say that amigalisp is broken.
|
|will someone out there who has amigalisp verify whether or not this indeed
|is the problem?

Well, if you type to your favourite, friendly Franz Lisp interpreter:

> (eq 99999 99999)

you'll get a 'nil' as an answer (at least I do, and that is what
the documentation says should happen).

The reason is that for the sake of efficiency, small integers
in Franz are represented uniquely. If I recall correcly, the
manual points out something like: if you know what you are doing,
you can use 'eq' on numbers, otherwise use 'equal'. I would
give you the same advice.

'eq' exists for two purposes: two compare symbols (NOT atoms),
and to check whether two objects are identical (i.e. eq :-).
For numbers, you don't want to know whether they are identical,
(i.e. whether they are located at the same memory location)
since you rarely alter the value of a number by setting the
number itself! You want to know whether they are equal. Therefore,
use 'equal'. It is also portable, and works.

Altogether, AMIGALisp is NOT broken if (eq 1 1) fails.

					Thomas.