longとfloatの足し算の落とし穴

Java

longとfloatの足し算を行なうと精度が落ちるので注意が必要だ。
コンパイラが警告してくれてもいいのに。

ソースコード

public class Test {
	public static void main(String[] args) throws Exception {
		System.out.println("1000000000000L + 1 - 1000000000000L = " + (1000000000000L + 1 - 1000000000000L));
		System.out.println("1000000000000L + 1 = " + (1000000000000L + 1));
		System.out.println("1000000000000L + 1f - 1000000000000L = " + (1000000000000L + 1f - 1000000000000L));
		System.out.println("1000000000000L + 1f = " + (1000000000000L + 1f));
		System.out.println("1000000000000L + 1d - 1000000000000L = " + (1000000000000L + 1d - 1000000000000L));
		System.out.println("1000000000000L + 1d = " + (1000000000000L + 1d));
		System.out.println("1000000000000L + 1L - 1000000000000L = " + (1000000000000L + 1L - 1000000000000L));
		System.out.println("1000000000000L + 1L = " + (1000000000000L + 1L));
	}
}

結果

1000000000000L + 1 - 1000000000000L = 1
1000000000000L + 1 = 1000000000001
1000000000000L + 1f - 1000000000000L = 0.0
1000000000000L + 1f = 1.0E12
1000000000000L + 1d - 1000000000000L = 1.0
1000000000000L + 1d = 1.000000000001E12
1000000000000L + 1L - 1000000000000L = 1
1000000000000L + 1L = 1000000000001