Path: utzoo!mnetor!uunet!steinmetz!ge-dab!codas!pdn!alan
From: alan@pdn.UUCP (Alan Lovejoy)
Newsgroups: comp.lang.modula2
Subject: Re: 0-based/1-based arrays
Message-ID: <2597@pdn.UUCP>
Date: 21 Mar 88 17:14:11 GMT
References: <7161@sol.ARPA> <2740@mmintl.UUCP> <4343@june.cs.washington.edu> <kW=wqIy00Uo507dUV3@andrew.cmu.edu> <8496@reed.UUCP>
Reply-To: alan@pdn.UUCP (0000-Alan Lovejoy)
Organization: Paradyne Corporation, Largo, Florida
Lines: 22

In article <8496@reed.UUCP> douglas@reed.UUCP (P Douglas Reeder) writes:
>I like being able to make the lower bound of an array  whichever I prefer
>and is the sensible one.  Allowing this requires only a trivial addition to
>array element evaluation:  the lower bound is subtracted from the index first.

That's way too complicated.  The right way is to bias the base address
of the array so that (Rn, Dn*ElementSize) points to the first element of the 
array when Dn's value is the lower index bound.   

Thus for 'a: ARRAY [m..n] OF Element', the compiler's (linker's)
internal base address for the array 'a' would be 

  ADR(firstElement) - (m * TSIZE(Element))

so that for positive m the base address is less than the address of the
first element by m * TSIZE(Element) size units, and for negative m it
is greater than the address of the first element by that amount.

This is just removing an invariant computation from run-time to compile-time
where it belongs.

--alann@pdn