Properties of Double.NaN and why does Integer wrapper not have an NaN data memeber?

Question

i=Double.NaN
while(i==i)
{
//some code
}

what is the output? Why don't we have a Integer.NaN ?

Answer 1

IEEE floating points have a "Not a Number" representation by spec. Integral types do not have such a state. Every possible binary representation of an integer is a real number.

Answer 2

what is the output?

There is no output because NaN != NaN as per the IEEE 754 standard, so the loop will never be entered.

Why don't we have a Integer.NaN?

Because Integers are based on a two's complement binary representation where every bit pattern is a valid integer, and none have any special meaning.

Answer 3

Double.NaN == x始终为false ，无论x是什么。

Answer 4

I cannot tell you why but you can work around it with this:

(int)Double.NaN;

so my guess is that there is not a good reason.

Answer 5

For float and double NaN is not equal to anything even itself.

For int and long types I have used MIN_VALUE as a value like NaN, but you have to code this yourself, if you want it to work this way.

BTW: There is a puzzler, when is the following an infinite loop.

while(x != x + 0);

There are three types for x when this is an infinite loop.

Another is when is this an infinite loop.

while(x == -x);

There is 16 type/value combinations for this, much more than you might expect. ;)