Re: MIDRANGE-L Digest V2 #1707

[Jim Langston <jlangston@xxxxxxxxxxxxxxxx>]

Floating point numbers are using mostly for scientific type
equations, and by their very nature loose quite a bit of
accuracy at any significant number of digits.

You must remember, a floating point number taking up, say,
4 bytes can express any number between 1 and infinity.
Obviously something is going to loose ground.  I used to
play around in QBasic on a PC with floating point numbers
seeing how large the discrepancy built over large numbers.
If I remember correctly on those old machines using 2 or
4 bytes (I can't remember which) the greatest a number could
reach was 99 decimal places wither way.  Something line
2.143983923E+99 or something like that.

If you want to calculate the distance to the sun, the circumference
of the earth in millimeters, or the U.S.'s national debt, floating
point is the way to go. But if you want to know the financial
balance of your company, stay as far away from them as humanly
possible.

Regards,

Jim Langston

Jon.Paris@halinfo.it wrote:

>  >> I use key word VARYING for simplicity.  I also encounter another similar
> issue of using %float and %dec
>
> I would describe this as the expected behavior. Floating point values are not
> exact and will not produce the same results as decimal based operations under
> all circumstances.  In this case if you half adjusted the result into a field
> with 5 decimal places you'd get the same result. Always use decimal values 
>when
> accuracy is important - unless of course you're trying to make the results 
>match
> what the accountant sees on his pocket calculator <grin>

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---