Javascript performance with closure

Question

var name = function(n) {
    var digits = ['one','two','three','four'];
    return digits[n];
}

var namenew = (function() {
    digits = ['one','two','three','four'];
    return function(n) {
        return digits[n];
    }
}());

Both the versions result in the same output, however it is said that the second version is much faster than the first version.

As I understand, first version executes the function everytime where as the second version stores the result of execution. That is what confuses me as a functional/regular OOPS programmer.

How can one save a function with its inner context? What is happening under the hood? Can some one pls clarify?

Answer 1

The real answer to that question would be about 3 pages long. But I try to make it as short as possible. ECMA-/Javascript is all about Execution Contexts and Object. There are three basic types of Context in ECMAscript: Global context, Function contexts and eval contexts.

Every time you call a function, your engine will spawn it in it's own function context. Also, there is a such called Activation object created. This mystic object is part of a function context which consists out of at least:

• [[Scope chain]]
• Activation object
• "this" context value

There may be more properties on different engines, but these three are required for any implementation of ES. However, back to the topic. If a function context is invoked, all parent contexts (or more precisly, the Activation objects from the parent contexts) are copied over into the [[Scope]] property. You can think of this property like an array, which holds (Activation-) Objects. Now, any function related information is stored in the Activation object (formal parameters, variables, function declarations).

In your example, the digits variable is stored in the Activation object for namenew. The second when the inner anonymous function is created, it adds that Activation object into its [[Scope]] propertys. When you call digits[n] there, Javascript first tries to find that variable in its own Activation object. If that fails, the search goes into the Scopechain. And voila, there we found the variable because we copied the AO from the outer function.

I already wrote too much for a short answer, but to really give a good answer to such a question you have to explain some basic knowledge about ES here. I guess that is just enough to give you an idea what really happens "under the hood" (there is a lot more to know, if you want to read more I'll give you some references).

---

You asked for it, you get it:

http://dmitrysoshnikov.com/ecmascript/javascript-the-core/

Answer 2

The first function recreates digits every time it's executed. If it's a large array this is needlessly expensive.

The second function stores digits in a context shared only with namenew. Every time namenew is executed it only performs a single operation: return digits[n].

An example like this wont show any noticeable gains in performance, but with very large arrays/objects/function calls performance will be improved significantly.

In an OOP perspective using a closure in this manner is similar to storing data in a static variable.

---

Don't forget that namenew is receiving the result of the closure function. The closure itself is only executed once.