From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.5-pre1 (2020-06-20) on
	ip-172-31-74-118.ec2.internal
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=3.0 tests=BAYES_00 autolearn=ham
	autolearn_force=no version=3.4.5-pre1
Date: 20 Apr 92 14:00:38 GMT
From: firth@sei.cmu.edu  (Robert Firth)
Subject: Re: loop indices
Message-ID: <43874@as0c.sei.cmu.edu>
List-Id: <comp.lang.ada>

In article <70482@ut-emx.uucp> hasan@emx.utexas.edu (David A. Hasan) writes:

>      v1 : Vector( p1 .. (p1+size-1) );
>      v2 : Vector( p2 .. (p2+size-1) );
>   BEGIN
>      FOR i IN 1..size LOOP
>         v2(i) := v1(i);      -------------------(!)
>      END LOOP;
>   END;

Well, I tend to be guided by style rather than efficiency, but in this
case I think one can have both.  The loop I'd write is

	for I in 0 .. SIZE-1 loop
	  V2(I+V2'FIRST) := V1(I+V1'FIRST);
	end loop;

I might alos revise the declarations, but that's a minor point.

Now, any reasonably good compiler should be able to work out that the
address expressions in the loop contain invariant subexpressions, and
essentially hoist the addition of V'FIRST out of the loop.  Even better,
many compilers generate an array root address that already designates
the first true element (rather than the mythical element zero), so the
addition isn't needed at all.  Finally, if the compiler can do simple
algebra, it can also see that no range check is needed on the array
accesses.