Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!biar!trebor
From: trebor@biar.UUCP (Robert J Woodhead)
Newsgroups: comp.lang.c
Subject: Re: Adding two pointers
Message-ID: <542@biar.UUCP>
Date: 7 May 89 18:42:33 GMT
References: <2765@buengc.BU.EDU> <563@lzaz.ATT.COM> <4093@ficc.uu.net> <17340@mimsy.UUCP> <17357@mimsy.UUCP>
Reply-To: trebor@biar.UUCP (Robert J Woodhead)
Organization: Biar Games, Inc.
Lines: 31

In article <17357@mimsy.UUCP> chris@mimsy.UUCP (Chris Torek) writes:
>>	sub	base,low,r0
>>	sub	base,high,r1
>>	add	r0,r1,r0
>>	div	r0,$2,r0
>	and	r0,$~3,r0
>>	add	r0,base,mid

Gee, if you define things as an array of pointers, you not only get
clearer source code, but more efficient machine code:

	ptr=&ptrArray[(high+low)/2]

becomes (pardon my assembly language, I'm extrapolating from your example)

	add	high,low,r0	; add high and low
	shr	r0,$1,r0	; divide by 2 (gee, a smart compiler!)
	shl	r0,$2,r0	; adjust to wordsize
	add	ptrArray,r0,r0	; add in the base address
	move	r0,ptr		; all done.

Seems to me this whole argument is spurious.  You can't add/mul/div/sub two
pointers together because the result is semantically meaningless.  You want
to redefine the language to ``hack in'' an operation that is already there,
and your hack will result in code that is harder for humans to read, and
harder for compilers to optimize.

-- 
Robert J Woodhead, Biar Games, Inc.  !uunet!biar!trebor | trebor@biar.UUCP
"The lamb will lie down with the lion, but the lamb won't get much sleep."
     -- Woody Allen.