Inconsistencies with Javascript isNaN()

Question

I've just been reading that there are apparently some inconsistencies with how isNaN handles things and is therefore not completely reliable. Strange.

In ES6 Number.isNaN() has some improvements over the built in isNaN() in that regard. However, I did a few quick tests:

var a = Number.isNaN(1); // false
var b = isNaN("hi"); // true
var c = Number.isNaN("hi"); // false <<< should be true!
var d = isNaN(NaN); // true
var e = Number.isNaN(NaN); // true
console.log(a, b, c, d, e)

It seems that there are still inconsistencies. So, is there some other solution to test for numbers? I've thought of these things but I wonder if there are some other methods.

Possible Solution to inconsistencies with isNaN:
- use parseFloat(). It will result in NaN if not a number. Question: is it fast?
- all numbers should be divisible by one. A `string` or `NaN` will result in `NaN` when divided by 1.

Answer 1

There is more.
Returns false only on number.

Number.isNaN(parseInt("hi")) //returns true
Number.isNaN(parseInt(NaN)) //returns true
Number.isNaN(parseInt(10)) //returns false

Answer 2

It's not issue or inconsistency, it's specific behaviour ^^

isNaN(arg) global function test if argument converted to Number is NaN;
Number.isNaN(arg) method test if argument is NaN, and nothing else...

So, contrarly to what you wrote in your question:

Number.isNaN("hi"); // shouldn't be true, as "hi" is a String and is not equal to the NaN special value ^^

Answer 3

@toni

Looks like a good solution!