RE: DecimalFormat

As you probably know, float and double numbers usually do not have an exact
decimal representation.  So the chances are that 5.345 cannot be exactly
represented as a float number, and it is stored internally as something that
approximates to 5.344999999.  And this number then rounds to 5.34.  Doubles
are more precise than float, so this is less likely to happen with double,
but I don't think it is impossible.  Personally, instead of fiddling with
this I would just use BigDecimal, where decimal rounding is well-defined and
controllable.

PC2

-----Original Message-----
From: Joe Pluta [mailto:joepluta@PlutaBrothers.com]
Sent: December 20, 2001 07:43
To: java400-l@midrange.com
Subject: RE: DecimalFormat


> From: Xu, Weining
>
> I run the code.  Yes, by using the numbers you listed, I got the same
> results as yours: n1 = 1.25, n2 = 2.32, n3 = 1.12.
>
> But, if I change the input numbers as: n1 = 5.345F; n2 = 1.145F; n3 =
> 2.645F;
> The results:
>       n1 = 5.34, n2 = 1.14, n3 = 2.64
> All three are wrong!

Xu, I got the same results.  I don't have a good answer.

        n1 = 5.345F; n2 = 1.145F; n3 = 2.645F;
        n1 = 5.34, n2 = 1.14, n3 = 2.64

        n1 = 5.345001F; n2 = 1.145001F; n3 = 2.645001F;
        n1 = 5.35, n2 = 1.15, n3 = 2.65

This doesn't make sense to me.  Anybody else?

Joe