Chrome and Node have a method called “”

Update: I found that this is part of the ECMAScript 6 standard. Firefox 22 supports this too. Some of the stuff in this post is wrong.

In short: is an undocumented method only supported by V8 JavaScript engines, so don't use it in the browser., b) is identical to a === b, except that, NaN) is true where NaN === NaN is not.

I'm the kind of person that goes into Chrome's Web Inspector, types window into the console, and pokes through the massive list to see what he can find. Today, I found an interesting method in Object: a method called is. I'd never heard of it before and it's on a very important object in JavaScript, so I did some sleuthing.

The first thing I found (or, rather, didn't find) was that it's undocumented. It's not in MDN's reference, it's not in the ECMAScript 5.1 spec, it's not in the ECMAScript 6 draft, nor could I find anything when searching around.

The second thing I found was that it's nonstandard. It's in Chrome and in Node, but not in Firefox or Opera or IE8 (I didn't try anything higher than IE8). It's not just some seemingly-undiscovered feature. It also won't work a lot of the time.

I started poking around with how it works, and I realized pretty quickly that it looked a lot like the === operator. The function would return true given 5 and 5, or "hello" and "hello". It would return false given two arrays, or two objects, or 5 and 8. It looked a lot like a functional equivalent of ===, so I decided to test it.

I started by writing a function that would test my hypothesis. It looked like this:

// Test my hypothesis that, b) is the same as a === b.
// Log an error if they're different.
function test(a, b) {
  if ((, b)) !== (a === b)) {
    console.error(' is different from === for ' + a + ' and ' + b);

I then hit it with 4,009,043 tests. I tried throwing pairs of numbers at it, random strings, booleans, objects, arrays, and the "weird" types: undefined, null, Infinity, and NaN. The output was just this: is different from === for NaN and NaN
Ran 4009043 tests.

Only one difference in over four million tests. In hindsight, it's kind of the one to expect. Here's the difference:

NaN === NaN;          // => false, NaN);  // => true

From my findings, that's the only difference. If you'd like grab my tests and give them a run.

So...when should you use Almost never.

You should use if you're only using Chrome or Node and you need to check the equivalence of two variables and they could be NaN. But because nobody's ever heard of this function (it seems), you should find a different solution.

Of course, I could be wrong about a lot of this; after all, it's undocumented!