I am currently working on writing a small math related library in php. I know there exists some kind of an analogue of javascript's NaN in php, it is a constant NAN of float type.
So far I have not ever come across this constant apart from php standard functions like asin() and alike, which will return this NAN value if you pass them an invalid parameter (>1 in this exact case).
This value is not detected by is_float() and is_numeric() since it is of float type, both of these functions return true and the only way to check it is is_nan(). Also, casting it to (int) will result in 0;
$x = asin(3); // $x = NAN
is_float($x); // true
is_numeric($x); // true
$d = (int)$x // $d = 0;
is_nan(0.5) // false
is_nan($x) // true
But none of existing libs whereof I know uses this behaviour, instead throwing an error is a common choice, as it seems.
So my questions are: which approach ought I to choose? why nowhere, apart from standard php functions, the NAN is used?
