From: Robert McGovern <robertm@...>
Date: 2002-12-10T21:27:57+09:00
Subject: Re: warnings -w

> The problem is that two-tenths (0.2) cannot be exactly represented as a
> binary floating point number.  When you look at the details, the number
> 0.2 is just really ...
> 
>   >> "%.20f" % 0.2
>   => "0.20000000000000001110"
> 
> ... or just a shade larger than two-tenths.  Therefore, when you divide
> 2.4 by something just a bit larger that two-tenths, you get something
> just a bit smaller than 12...
> 
>   >> "%.20f" % (2.4 / 0.2)
>   => "11.99999999999999822364"
> 
> This is a consequence of representing numbers in floating point and is
> common whenever dealing with floating point numbers.  Last week at work,
> I came across a situation where a sum of monetary amounts were off by a
> penny ... Yes, they were using floating point to represent dollar
> amounts.  Never a good idea.

Duh forgot about that, wish I'd recieved this before I sent my mail out.

*sigh*