Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!gatech!hubcap!beede
From: beede@hubcap.UUCP (Mike Beede)
Newsgroups: comp.lang.c
Subject: Re: Two dimensional arrays in C (BLAST!!)
Message-ID: <254@hubcap.UUCP>
Date: Thu, 2-Jul-87 20:34:46 EDT
Article-I.D.: hubcap.254
Posted: Thu Jul  2 20:34:46 1987
Date-Received: Sat, 4-Jul-87 08:29:58 EDT
References: <245@hubcap.UUCP]
Organization: Clemson University, Clemson, SC
Lines: 35

in article <245@hubcap.UUCP], beede@hubcap.UUCP (ME AGAIN) says:
] in article <238@hubcap.UUCP>, beede@hubcap.UUCP (Me) says:
]> #define matrix_el(array,i,j) (array[y*sizeof(double)+j])
]> 
] which of course is completely bogus.  I'm not sure _where_ that
] sizeof(double) popped in, but anyway the correct version is
] 
] #define matrix_el(array,i,j) (array[i*y+j]) ,
] 
Well--now the macro is correct if you don't use anything but single constants
or variables as the indexing arguments.  If we do something like

 matrix_el(A,k+5,m)

it expands to 

 (A[k+5*y+m])

which is surely bogus, since this is k+(5*y)+m instead of (k+5)*y+m.
The moral of the story, as Mark Brader was kind enough to point out
(twice!), is to always parenthesize the arguments of a macro when they
occur in the rhs, like the (REALLY) correct version:

#define matrix_el(array,i,j) (array[(i)*y+(j)]) ,

The sad part is, not only did Mark include the correct version in his
mail to me (which I discarded), but I myself was once burned by the
lack of parenthesis in a macro.

Sorry this took so long to iron out.
-- 
Mike Beede                      UUCP: . . . !gatech!hubcap!beede
Computer Science Dept.          ARPA: BEEDE@TECNET-CLEMSON.ARPA
Clemson University              INET: beede@hubcap.clemson.edu
Clemson SC 29631-1906           PH: (803)656-{2845,3444}