Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!hp4nl!eurtrx!euraiv1!reino
From: reino@cs.eur.nl (Reino de Boer)
Newsgroups: comp.lang.pascal
Subject: Re: Is this not a bug?
Message-ID: <1991Feb22.111536.13080@cs.eur.nl>
Date: 22 Feb 91 11:15:36 GMT
References: <1991Feb20.170152.5401@ux1.cso.uiuc.edu> <5777@husc6.harvard.edu>
Reply-To: reino@cs.eur.nl
Organization: Erasmus University Rotterdam
Lines: 53

In <5777@husc6.harvard.edu> rind@popvax.uucp (747707@d.rind) writes:

>I don't see this as a bug.  It's not ABS that's not functioning as you
>expect.  Any other function would have behaved the same way.  When
>you took the difference of two WORD variables inside a function call,
>the result was a WORD.  Since a WORD is unsigned, it could never
>go negative, and so the difference had to be a number between
>0 and 65535.  The function call ABS(w2-w1) is equivalent to creating
>a third word variable w3 and doing w3 := w2-w1 followed by ABS(w3).
>I can't see the point in ever taking the absolute value of a WORD
>type variable, but perhaps someone else has some thoughts on this.

You are ABSolutely right, but when the original poster really wants the
equivalent of

	if w1 > w2 then          (hope you see the point)
	  w3 := w1 - w2
	else
	  w3 := w2 - w1

to get the absolute difference between two words, the following code
seems to work:

	w3 := word(abs(integer(w1 - w2)))

but, as anyone who understands the problem knows, this fails for any
values of w1 and w2 with:

	w1 < w2    and    w2 - w1 > 32768

since that large a difference cannot be represented as an

	abs(integer(i))

So, your best bet would be to write something like

	function abs_diff( w1, w2 : word ) : word;

	  begin
	    if w1 > w2 then
	      abs_diff := w1 - w2
	    else
	      abs_diff := w2 - w1
	  end;

Hope this helps -- Reino

>David Rind
>rind@popvax.harvard.edu
-- 
Reino R. A. de Boer     "We want to build the right product right, right?"
Erasmus University Rotterdam ( Informatica )
e-mail: reino@cs.eur.nl