Re: Strange result with exponents...

[Hans Boldt <boldt@xxxxxxxxxx>]

Scott Klement wrote:
> 
> Someone posted this question on the iSeries Network's RPG forum, and it
> stumped me, so I thought I'd bring it here and see what you guys think.
> 
> I'm running the following code.   The two numbers, VAR1 and VAR2 should
> have the same results, shouldn't they?
> 
>      d VAR2            S             30p 0
>      d VAR1            s             30p 0
> 
>      c                   eval      VAR2 = (16**15)
>      c     VAR2          dsply
> 
>      c                   eval      VAR1 = (16**14)
>      c                   eval      VAR1 = VAR1 * 16
>      c                   dsply                   VAR1
> 
>      c                   eval      *inlr = *on
> 
> The result of running this program looks like this:
> 
> DSPLY             1152921504606846980
> DSPLY             1152921504606846976
> 
> So, my question is, what's going on here?   And, why would there be a
> difference of 4?!
> 
> As far as I can tell, this only happens on V5R2.  Can anyone confirm that
> it doesn't happen on older releases?
> 

Scott:  I haven't been able to monitor this list lately since this week I've 
been downtown waiting to be selected for jury duty. But today they let us 
out a bit early. 

What's happening here? It's simple, really. The exponentiation operator 
returns a float numeric value. When you're dealing with double precision 
IEEE float numeric, you're limited to about 16 digits of precision. Any 
digits past the 16th or 17th cannot be trusted. Period. (Note in the values 
listed that the difference shows up in the 18th digit and beyond.)

Cheers! Hans