refactor：

首先魔鬼数字100并没有体现出精确到小数点后两位这一知识。在这里我们需要将精度这一知识显性化。

Rectangle::Rectangle(double width, double height)
{
    const unsigned int PRECISION = 2;
    double factor = pow(10, PRECISION);
    m_width = (unsigned int)(width * factor)/factor;
    m_height = (unsigned int)(height * factor)/factor;
}

double Rectangle::area() const
{
    const unsigned int PRECISION = 2;
    double factor = pow(10, PRECISION);
    return (unsigned int)((m_height * m_width)*factor + 0.5)/factor;
}

在sprint2中提及到：

长和宽的精度保持为0.01,且不按照四舍五入，例如1.236 => 1.23。

这其实就是向下取整，当前实现中使用了强制类型转换来达到同样的效果，但是没有直接体现出向下取整这一知识。在这里我们直接使用库函数floor来以明确：

Rectangle::Rectangle(double width, double height)
{
    const unsigned int PRECISION = 2;
    double factor = pow(10, PRECISION);
    m_width = floor(width * factor)/factor;
    m_height = floor(height * factor)/factor;
}

double Rectangle::area() const
{
    const unsigned int PRECISION = 2;
    double factor = pow(10, PRECISION);
    return floor((m_height * m_width)*factor + 0.5)/factor;
}

现在代码比之前的清楚多了，但是仍然存在重复，接下来我们消除重复的代码：

namespace
{
    double processPrecision(double value, const unsigned int precision, double compensation)
    {
        double factor = pow(10, precision);
        return floor(value * factor + compensation)/factor;
    }
}


Rectangle::Rectangle(double width, double height)
{
    m_width = processPrecision(width, 2, 0);
    m_height = processPrecision(height, 2, 0);
}

double Rectangle::area() const
{
    return processPrecision(m_height * m_width, 2, 0.5);
}