Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!mcvax!cernvax!ethz!ceb
From: ceb@ethz.UUCP (Charles Buckley)
Newsgroups: comp.lang.lisp
Subject: Re: A small brush fire ...
Message-ID: <412@ethz.UUCP>
Date: Sat, 29-Nov-86 22:05:16 EST
Article-I.D.: ethz.412
Posted: Sat Nov 29 22:05:16 1986
Date-Received: Mon, 24-Nov-86 22:36:45 EST
Organization: CS Department, ETH Zuerich, Switzerland
Lines: 85

In his posting of 21 November (our number 535 - the numbering schemes
seem to play a time zone style game as the messages cross the Atlantic), Mr.
Steele responds to  Mr. Sloan's challenge (recalled here) 

   >	So how would you answer the following:  Since "good"
   >programmers know better than using free (== global) variables,
   >the argument between lexical and dynamic scoping is really a
   >mute point.  The real issue this: Can you exhibit a piece of code
   >that depends on the scoping rules that can't be coded AS WELL
   >in a way that doesn't depend on the scoping rules?
   >

with the following redefinition of mapcar, and an application function
using it:

   (defun mymapcar (myfunction list)
     (cond ((null list) nil)
	   (t (cons (myfunction (car list))		;or funcall, etc.
		    (mymapcar myfunction (cdr list))))))

   (defun consall (l list)
     (mymapcar '(lambda (x) (cons x list))
	       l))

   lexical:
   (consall '(a b c) '(d)) ==> ((a d) (b d) (c d))

   dynamic:
   (consall '(a b c) '(d)) ==> ((a a b c) (b b c) (c c))

The code presented certainly depends on the scoping rules, but the
second part of the challenge remains unanswered, i. e. "can't be coded
AS WELL in a way that doesn't depend on the scoping rules."  

In fact, in the context of the (perforce) interpreted
lambda-expression which makes up the function pased as an argument,
list is a special variable which must be resolved by name at runtime,
an expensive process.  Therefore, many would not consider the above an
example of good code at all.

One rule independent way of coding the above (I use Zetalisp syntax) is

(defun mynewmapcar (function list-to-be-indexed &rest fixed-args)
   (cond ((null list-to-be-indexed) nil)
         (t (cons (lexpr-funcall function (car list-to-be-indexed) fixed-args)
                  (lexpr-funcall 'mynewmapcar function (cdr list-to-be-indexed)
                                 fixed-args)))))

(defun consall (l list)
   (mynewmapcar 'cons l list))

[lexpr-funcall is the same as funcall, except the last argument is
used to terminate the argument list passed to the funcalled function,
instead of being the last element in a nil terminated list.]

The above behaves in the manner implied by Mr. Steele's example
regardless of scoping rules, and supports Mr. Sloan's point about free
(special) variables.  Note I called the workhorse function 'new'
mapcar, because the issue lurking behind al of this is that if the
meta-functions which we have grown accustomed to were semantically
more powerful, the brush fire would have remained as someones
smoldering cigarette ashes.

Dr. Charles E. Buckley			uucp: mcvax!ethz!jungfrau!ceb
					earn: BUCKLEY@CZHETH5A

Institut fuer Integrierte Systeme	
ETH-Zentrum
CH-8092 Zuerich (Suisse)
Tel: 01/256 5245 (national)
   +411 256 5245 (international)


-- 

Dr. Charles E. Buckley			uucp: mcvax!ethz!jungfrau!ceb
					earn: BUCKLEY@CZHETH5A

Institut fuer Integrierte Systeme	
ETH-Zentrum
CH-8092 Zuerich (Suisse)
Tel: 01/256 5245 (national)
   +411 256 5245 (international)