Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ncar!tank!mimsy!chris
From: chris@mimsy.UUCP (Chris Torek)
Newsgroups: comp.lang.c
Subject: Re: Contiguous Arrays
Message-ID: <16080@mimsy.UUCP>
Date: 22 Feb 89 23:48:22 GMT
References: <2508@ssc-vax.UUCP> <8943@alice.UUCP> <1828@valhalla.ee.rochester.edu> <1831@valhalla.ee.rochester.edu>
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 22

In article <1831@valhalla.ee.rochester.edu> badri@valhalla.ee.rochester.edu
(Badri Lokanathan) writes:
>... But Henry's statement implies that on some architectures
>
>(x + i) - i != x + (i - i)
/* where x is a pointer */

This is correct: on some architectures, if `x' is one of the integral
types except the unsigned types, or if x is a pointer type, x + i may
cause an overflow trap, iff the address computed by x+i is not within
the object denoted by x.  (That is: on some machines, there exists an
integer $i$ such that $x+i = \error$.)

The trick described is, however, portable whenever the `zero point' of
the derived array is contained within the array, because in that case
we know that 0 <= i <= N (where N is the number of elements in the array)
and the pANS guarantees that this address is computable (does not cause a
range error).  (That is: $\forall i \elem [0,N], x+i$ is computable, and
$(x+i)-i = x$.)
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163)
Domain:	chris@mimsy.umd.edu	Path:	uunet!mimsy!chris