Fixed-Point Arithmetic Types for C++

摘要

Computers cannot exactly represent the value of fractions such as 8/9, where the divisor is not a multiple of 2. In fact, the value of the integer expression 8/9 is zero, even though the mathematical quantity is much closer to 1. For computations involving fractions, most developers use floatingpoint arithmetic, which approximates the value of 8/9 as the decimal fraction 0.88889, a more sensible and accurate choice than the unintuitive result of the integer expression.