love this

I agree. Also, in Node.js there is URL constructor available where you can parse the host easily then in JS we also have location.host. So why reinvent the wheel.

great detaild explanation！

@pepkin88: It would be nice if we had benchmarks on different engines, because I'm hesistant to declare one solution or another "best practices" without seeing some performance measures...

[].concat looks for property concat in the newly created object []. It doesn't find it, so it looks for that property in the prototype of that new array object. Array.prototype.concat also makes 2 lookups, first for prototype property of Array function, second for concat property of Array.prototype object. So the difference is creation an empty array object [].

All of that is only in theory, because the real behavior depends on the engine‘s implementation. It could be optimized for such situations and could skip some steps (eg. would skip creating a new, later unused, array object, or would translate Array.prototype directly to that object, without a lookup).

Thank you for the detailed explanation xcthulhu! Your response has helped to clear up some of my confussion with apply().

In the comments for this post on SO, SLaks explanation of [] versus Array.prototype was:

"[] is shorter, but it creates an extra array instance. Array.prototype is longer but faster."

I'm not sure if that's correct, so maybe somoene here can verify that is the case and provide a citation to documentation that indicates the veracity of the statement.

cool solution!!!!

you win, sir.