Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!husc6!think!ames!pasteur!ucbvax!decwrl!hplabs!sdcrdcf!trwrb!cadovax!gryphon!sarima
From: sarima@gryphon.CTS.COM (Stan Friesen)
Newsgroups: comp.lang.c++
Subject: Re: overloading array indexing
Keywords: [] bugs? array indexing
Message-ID: <4023@gryphon.CTS.COM>
Date: 15 May 88 14:10:30 GMT
References: <451@ethz.UUCP>
Reply-To: sarima@gryphon.CTS.COM (Stan Friesen)
Organization: Trailing Edge Technology, Redondo Beach, CA
Lines: 35

In article <451@ethz.UUCP> peter@ethz.UUCP (Peter Beadle) writes:
>
>I have been trying to overload the array indexing operation as part of
>a simulator I am writing. ...
>
>The translator produces a _tiny__vec function when I don't inline the
>overload.  This function contains the correct code but it never gets
>called in main.  ...
>
>/* test program */
>#include    <stream.h>
>class tiny{
>    int v;
>public:
>    tiny() {v=0;}
>    void operator[](int i) { cerr << "in overloaded []\n"; }
>};
>
>main() {
>    tiny    fred[5];
>
>    fred[2];
>}
>
	AH, I see the problem. The declaration creates a REAL array of 5
"tiny" objects called fred. Thus the expression "fred[2]" later indexes
into this array, generating a reference to a tiny object . You would have
to say "fred[2][2]" to get tiny::operator[], since that is only applied to
a "tiny" object, not an array of "tiny" objects. Look on page 182 of
Stroustrup's book and note that he declaes an instance of his "assoc" class,
which has a [] operator, as "assoc vec(512)" not "assoc vec[512]".
-- 
Sarima Cardolandion			sarima@gryphon.CTS.COM
aka Stanley Friesen			rutgers!marque!gryphon!sarima
					Sherman Oaks, CA