I have a code piece that I'm analysing but I don't understand this lines:
double level = 0;
...
if (!(level == level) && !(level < level) && !(level > level))
exit(0);
I think it is about double comparison subject. Any idea what it is?
Asked
Active
Viewed 249 times
0
Volkan Ozyilmaz
- 470
- 4
- 19
-
See NAN (Not A Number) which would make this expression return true. – Yam Marcovic Jan 19 '16 at 14:42
-
4Who has written this? – i486 Jan 19 '16 at 14:42
-
My first thought would be that they're trying to see if `level` is NaN. I assume there is some intervening code that you've elided that changes the value of `level`? Where did you get this code? What does it do? – Cody Gray - on strike Jan 19 '16 at 14:42
-
1The test with `==` tests for NaN. The tests using `<` and `>` seem pointless. – interjay Jan 19 '16 at 14:44
-
1[Related: Answer to "Checking if a double (or float) is NaN in C++"](http://stackoverflow.com/a/570694/96780) – Daniel Daranas Jan 19 '16 at 14:48
-
@CodyGray Yes, it calculates power of the values and accumulate in it. for (int i = 0; i < len; i++) level += values[i] * values[i]; – Volkan Ozyilmaz Jan 19 '16 at 14:57
-
@i486 I don't know who wrote it but is it important? – Volkan Ozyilmaz Jan 19 '16 at 14:58
1 Answers
5
It's a rather over-engineered test for not a number (NaN).
NaN is a special floating point value. (Not all floating point types have it but the most common one - IEEE754 - does.)
It's defined to compare not equal to all other numbers, including itself.
It's also defined to be not greater than or not less than any other number, including itself. (Although this subsequent test is superfluous).
From C++11, you can use std::isnan(level) instead. Before then, my favourite way of testing was to use the considerably simpler expression level != level.
Bathsheba
- 231,907
- 34
- 361
- 483