Re: fixed point vs floating point (When do you use it ?)

[dewar@merv.cs.nyu.edu (Robert Dewar)]

tmoran says

<<A stocks & commodities database I know of uses fixed point.  Stocks have a
delta of 1/8 (though that is supposed to change to decimal soon) and
different commodities have a bunch of different deltas.
>>

The fact that stocks have a delta of 1/8 is NOT a good reason to choose
fixed-point over floating-point for the representation of these values.
In practice all stock values will be model numbers of any reasonable
floating-point type, so the issue of accurate representation does not 
arise. What is important is the computations that you propose to perform
on these values. I would guess that for most purposes, it is better to
use floating-point for stock prices than fixed-point (of course if you
are in COBOL, you have little choice, and indeed stock prices don't fit
nicely into COBOL, which has no provision for binary fixed-point).