Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!ames!lll-winken!uunet!mcvax!hp4nl!botter!star.cs.vu.nl!sater
From: sater@cs.vu.nl (Hans van Staveren)
Newsgroups: comp.lang.c
Subject: Re: Another silly question
Keywords: C, history, anecdotal
Message-ID: <2425@sater.cs.vu.nl>
Date: 1 May 89 10:41:46 GMT
References: <pointer> <array> <reference> <access> <element> <2459@nmtsun.nmt.edu> <1513@auspex.auspex.com>
Reply-To: sater@cs.vu.nl (Hans van Staveren)
Organization: VU Informatica, Amsterdam
Lines: 32

In article <1513@auspex.auspex.com> guy@auspex.auspex.com (Guy Harris) writes:
>
> >My CS instructor and I disagree about a certain moot point.  I have a text
> >book which says that
> >
> >	*(a + i)	and	a[i]
>	/* LOTS OF STUFF DELETED */
>You are correct; he is incorrect.  Perhaps he does not understand how
>pointer addition works in C?  If you add an integral value N to a pointer,
>it doesn't increment the address in that pointer by N storage units
>(bytes on byte addressible machine, etc.), it can be thought of as
>incrementing the address by N objects of the type to which that pointer
>points.  In C, pointers have types, and those types are significant.

Just to show how old I am, let me tell you that in Unix V6 on the PDP 11,
the only machine it ran on, the expressions
	a + i
and
	i + a
with a a pointer and i an integer were not equivalent.
	a + i worked as it does nowadays
while	i + a worked as this guys instructor fears.

I am even willing to admit I used this trick, but then in those days the way 
to get an unsigned was to declare it as a char* and casts were not invented
yet.

Language historians, take note!

	Hans van Staveren
	Vrije Universiteit
	Amsterdam, Holland