"comparing identical expressions" discourages using un-broken isNaN() substitute about jslint

[douglascrockford]

No. We are looking at a better alternative to isNaN in a later edition. In the meantime, I cannot recommend the false equality trick.

isNaN and isFinite are inadequate in their current forms: not only can they be implemented in terms of the standard equality operators, but their automatic coercion of their arguments to numbers is unnecessary at best and unexpected at worst.

As @douglascrockford noted, the spec recognizes that a better alternative is needed; see the note in section 15.1.2.4:

A reliable way for ECMAScript code to test if a value X is a NaN is an expression of the form X !== X. The result will be true if and only if X is a NaN.

@yahelc In the meantime, I would recommend overwriting the global isNaN and isFinite functions with the following:

function isNaN(value) {
  return value != value;
}

function isFinite(value) {
  return value == value && value != 1 / 0 && value != -1 / 0;
}

[douglascrockford]

That is terrible advice.

[GarrettS]

Replacing the global isNaN with your own could make for some confusing and surprising code, for sure. Strongly agree not to do that.

The inequality check clearly works as specified; what's your reason for recommending against it?

isNumberNaN = (typeof x == "number" && x !== x);

[douglascrockford]

typeof x === 'number' && isNaN(x)

[GarrettS]

Yep. Clearer, less thinking req'd, no comment needed. Can fail if somebody messes with global isNaN.