Fixed-Point Arithmetic?

Have there been any proposals (or, indeed, does anyone have any private libraries) for fixed-point arithmetic? I can imagine uses for these 64-bit types: * unsigned, with all bits after the decimal point i.e. 0 <= x < 1 * unsigned, with 32 bits before the decimal point and 32 after * signed, with 32 bits before the decimal point and 32 after -- Ashley Yakeley, Seattle WA