Why you should use my new way to convert to an array

new-convert-to-array.js

// The data
var items = document.getElementsByTagName('p');

// The old way
var arr = Array.prototype.slice.call(items);

// Pros:
// - Doesn't need a helper method that loops through stuff
//
// Cons:
// - Verbose
// - Doesn't work in IE 8 or lower

// The new way
var arr = Array.apply([], items);

// Pros:
// - Doesn't need a helper method that loops through stuff
// - Reasonably concise
// - Works in IE 7 and 8 at least
//
// Cons:
// - None that I can think of! Take that, old way! (jsPerf test pending.)

// Awesome new way brought to you by Ryan O'Hara (minitech)

@pimvdb Oh, that's true. Array.apply([], document.getElementById('id')) returns an empty array for me though.
I expected it to return an array containing the DOM node.

Edit: Array.prototype.slice.call(document.getElementById('id')) returns an empty array, too.

@js-coder: .apply expects an array, so:

Array.apply([], [ document.getElementById('id') ]);

But since getElementById returns at most one element, the whole .apply trick is not necessary I think. You could just do [ elem ] to put the element in an array. As I understand it, the gist is about converting an array-like object to a real array, which is a problem you don't encounter with getElementById.

@pimvdb Yeah, I had an error in reasoning. :)

---

[].slice.call(items); seems to work, too, and it's a bit shorter than the usual way. (It creates a new array though, but I don't think you'll see any real performance difference.)

Argh. Right. And yes, null is better, [] is left over from other experiments.

@js-coder: Still doesn't work in IE, though - that's what I'm trying to fix in a non-frustrating way.