double 计算时，什么时候会丢失精度？

public class Precision {
    public static void main(String[] args) {
        double num1 = 1.9;
        double num2 = 0.5;
        double num3 = 1.4;
        double num4 = (num1 - num2)/num3;
        System.out.println("num4 = " + num4);
        double num5 = 1.4;
        double num6 = 0.5;
        double num7 = 0.9;
        double num8 = (num5 - num6)/num7;
        System.out.println("num8 = " + num8);
    }
}

 

两种结果的区别是什么？