Path: utzoo!yunexus!ists!helios.physics.utoronto.ca!jarvis.csri.toronto.edu!mailrus!uwm.edu!cs.utexas.edu!tut.cis.ohio-state.edu!n8emr!uncle!oink!jep
From: jep@oink.UUCP (James E. Prior)
Newsgroups: comp.graphics
Subject: Re: Integral Square Root
Message-ID: <52@oink.UUCP>
Date: 31 Dec 89 17:16:46 GMT
Article-I.D.: oink.52
References: <9170@cbmvax.commodore.com>
Reply-To: jep@oink.UUCP (James E. Prior)
Organization: Random Prime Research Institute  Columbus, Ohio
Lines: 28

In article <9170@cbmvax.commodore.com> mitchell@cbmvax.UUCP (Fred Mitchell - PA) writes:
>...
>   for (i = 15, r = 0; i >= 0; --i)
>       if ( (rb = r+(b=1<<i)) * rb <= n)
              ^^                 ^^
              !!                 !!
>           r += b;
>   return (ULONG) r;
>...

Beware, C doesn't guarantee the order of evaluation.  Fred's code will 
work on some compilers, but not on others.

The following code fragment should overcome that particular problem.  
   for (i=15,r=0;i>=0;--i)
      if ((rb=r+(b=1<<i)) , rb*rb<=n)
         r+=b;
   return (ULONG)=r;

And the following code should be a little quicker.  
   for (i=15,r=0,b=1<<i;i>=0;i--,b>>=1) {
      rb=r+b;
      if (rb*rb<=n)
         r+=b;
   return (ULONG)r;

-- 
Jim Prior    jep@oink    osu-cis!n8emr!oink!jep    N8KSM