views:

810

answers:

3

I've read that rather than simply writing a bunch of functions, I should use object literal.

Can someone explain what the advantages of object literal are with examples, because I don't understand thus far.

Thanks

+2  A: 

Using an object literal (a.k.a. object literal pattern) will not pollute the global namespace as severely as using many functions declared globally will, and also helps to organise code in a logical fashion

For example, this object literal

var obj = {
              find : function(elem) { /* find code */ },
              doSomething: function() { /* doSomething code */ },
              doSomethingElse: function() { /* doSomethingElse code */ }
          }

compared to

function find(elem) { /* find code */ },
function doSomething() { /* doSomething code */ },
function doSomethingElse() { /* doSomethingElse code */ }

will create only one property on the global object compared to three. You can then easily use the functions like so

obj.doSomething();
Russ Cam
you could just write many function in one normal function and not pollute your code. object literal doesn't give any added value.I myself use the prototype method..
vsync
+1  A: 

Rebecca Murphey did a talk on Object Literals at this year's jQuery Conference. One of the best reasons to use them is simply good code organization.

Here is Rebecca's write up on the Object Literal Pattern : http://blog.rebeccamurphey.com/2009/10/15/using-objects-to-organize-your-code/

Alex Sexton
Another great article on object literals: http://www.wait-till-i.com/2006/02/16/show-love-to-the-object-literal/
Alex Sexton
+5  A: 

As Russ Cam said, you avoid polluting the global namespace, which is very important in these days of combining scripts from multiple locations (TinyMCE, etc.).

As Alex Sexton said, it makes for good code organisation as well.

If you're using this technique, I'd suggest using the module pattern. This still uses object literals, but as the return value from a scoping function:

var MyThingy = (function() {

    function doSomethingCool() {
        ...
    }

    function internalSomething() {
        ....
    }

    function anotherNiftyThing() {
        // Note that within the scoping function, functions can
        // call each other direct.
        doSomethingCool();
        internalSomething();
    }

    return {
        doSomethingCool: doSomethingCool,
        anothrNiftyThing: anotherNiftyThing
    };
})();

External use:

MyThingy.doSomethingCool();

The scoping function is wrapped around all of your functions, and then you call it immediately and store its return value. Advantages:

  • Functions are declared normally and therefore have names. (Whereas with the {name: function() { ... }} format, all of your functions are anonymous, even though the properties referencing them have names.) Names help tools help you, from showing call stacks in a debugger, to telling you what function threw an exception.
  • Gives you the freedom of having private functions only used by your module (such as my internalSomething above). No other code on the page can call those functions; they're truly private. Only the ones you export at the end, in the return statement, are visible outside the scoping function.
  • Makes it easy to return different functions depending on environment, if the implementation just changes completely (such as IE-vs-W3C stuff, or SVG vs. Canvas, etc.).

Example of returning different functions:

var MyUtils = (function() {
    function hookViaAttach(element, eventName, handler) {
        element.attachEvent('on' + eventName, handler);
    }

    function hookViaListener(element, eventName, handler) {
        element.addEventListener(eventName, handler, false);
    }

    return {
        hook: window.attachEvent ? hookViaAttach : hookViaListener
    };
})();

MyUtils.hook(document.getElementById('foo'), 'click', /* handler goes here */);
T.J. Crowder
When you instantiate a function in the "anonymous" way, you can still give it a name (var x = function x() { ... }). When you do that, the name is bound such that it's available for recursive references within the function.
Pointy
@Pointy: You can't do that (use a function name within an assignment) cross-browser, it doesn't work correctly on either IE or Safari; details: http://yura.thinkweb2.com/named-function-expressions/ And you don't need to, the function's proper name (the `foo` in `function foo`) is in-scope in the entire scope where it's declared, including within the function itself, so `foo` can call itself via the symbol `foo`, no need to assign the function reference to anything (at that point).
T.J. Crowder
(Continuing) Granted it would be *nice* to be able to use both assignment and proper name at the same time because it would make it easier to export functions from the scoping function. The spec certainly allows it, but sadly, practicalities (implementation bugs) enter into it.
T.J. Crowder