IEEE 754 Floating point numbers should only be used where you can tolerate rounding errors. Even if you have enough precision to do base conversions, actually

IEEE 754 Floating point numbers should only be used where you can tolerate rounding errors. Even if you have enough precision to do base conversions, actually doing math will frequently introduce rounding errors. You can either be very careful to avoid the many cases where this can happen, choose a different number format (fixed-point, decimal floating point, numerator-denominator, arbitrary precision mantissa binary floating point, which one depends on your problem domain), or use units that let you work in integers (equivalent to fixed-point, but easier to explain).