Make your Backbone.js apps dance with a composite application architecture!

Backbone.Marionette is a composite application library for Backbone.js that aims to simplify the construction of large scale JavaScript applications. It is largely a collection of common design and implementation patterns found in the applications that I (Derick Bailey) have been building with Backbone, and includes various pieces inspired by composite application architectures, such as Microsoft's "Prism" framework.

Backbone.Marionette is designed to be a lightweight and flexible library of tools that you can use when you want to. Like Backbone.js itself, you're not required to use all of Backbone.Marionette just because you want to use some of it.

You can download the raw source code from the "src" folder above, or grab one of the many builds from the "lib" folder.

To get the latest stable release, use these links which point to the 'master' branch's builds:

Development: backbone.marionette.js

Production: backbone.marionette.min.js

Development: backbone.marionette.js

Production: backbone.marionette.min.js

Marionette is unofficially available from various package management systems, such as RubyGems, Node Package Manager, Nuget, etc. These packages are maintained by the community and are not part of the core Backbone.Marionette code.

Available Packages

This readme represents the majority of documentation for Marionette. However, this file has become quite large and can be difficult to wade through. There's a better version of it available, though.

A wiki is an important aspect of a thriving community, as it provides a place for the community to contribute ideas, examples, answer frequently asked questions, and more. If you're looking for community-driven information, examples that go beyond the dry technical documentation, or want to contribute your own ideas and examples to the community, please see the wiki page.

In addition to this readme, I've commented the source code quite heavily and run it through Docco as part of my build process. This produces a nicely formatted, annotated source code as documenation file.

You can read the annotated for all the detail of how Marionette works, and advice on which methods to override when.

Lastly, I blog about Marionette on a regular basis, at my LosTechies.com blog.

There are many ways to support Marionette, ranging from submitting bugs and patches, to helping answer questions in the issue list and on StackOverflow, to providing financial support for maintaining Marionette.

If you're interested in helping with code and questions, please see the issues list and stack overflow tag here:

• Github Issues
• StackOverflow Tag

If you have an idea to improve Marionette or want to report a bug, please use the issues list.

I understand that not everyone can offer assitance with code, bug submissions, and answer questions. That's fine. Everyone has their own strengths.

If you're using Marionette and you're finding that it is saving you as much time and effort as I believe it does, then please consider financial support for the project.

(You would be surprised at how little financial support is given, in spite of the financial savings that a framework like this creates.)

These are the strings that you can pull to make your puppet dance:

• Backbone.Marionette.Application: An application object that starts your app via initializers, and more
• Backbone.Marionette.Application.module: Create modules and sub-modules within the application
• Backbone.Marionette.AppRouter: Reduce your routers to nothing more then configuration
• Backbone.Marionette.View: The base View type that other Marionette views extend from (not intended to be used directly)
• Backbone.Marionette.ItemView: A view that renders a single item
• Backbone.Marionette.CollectionView: A view that iterates over a collection, and renders individual ItemView instances for each model
• Backbone.Marionette.CompositeView: A collection view and item view, for rendering leaf-branch/composite model hierarchies
• Backbone.Marionette.Layout: A view that renders a layout and creates region managers to manage areas within it
• Backbone.Marionette.Region: Manage visual regions of your application, including display and removal of content
• Backbone.Marionette.EventAggregator: An extension of Backbone.Events, to be used as an event-driven or pub-sub tool
• Backbone.Marionette.BindTo: An event binding manager, to facilitate binding and unbinding of events
• Backbone.Marionette.Renderer: Render templates with or without data, in a consistent and common manner
• Backbone.Marionette.TemplateCache: Cache templates that are stored in <script> blocks, for faster subsequent access
• Backbone.Marionette.Callbacks: Manage a collection of callback methods, and execute them as needed

The Application, Region, ItemView and CollectionView use the extend syntax and functionality from Backbone, allowing you to define new versions of these objects with custom behavior.

The Backbone.Marionette.Application object is the hub of your composite application. It organizes, initializes and coordinate the various pieces of your app. It also provides a starting point for you to call into, from your HTML script block or from your JavaScript files directly if you prefer to go that route.

The Application is meant to be instantiated directly, although you can extend it to add your own functionality.

MyApp = new Backbone.Marionette.Application();

Your application needs to do useful things, like displaying content in your regions, starting up your routers, and more. To accomplish these tasks and ensure that your Application is fully configured, you can add initializer callbacks to the application.

MyApp.addInitializer(function(options){
  // do useful stuff here
  var myView = new MyView({
    model: options.someModel
  });
  MyApp.mainRegion.show(myView);
});

MyApp.addInitializer(function(options){
  new MyAppRouter();
  Backbone.history.start();
});

These callbacks will be executed when you start your application, and are bound to the application object as the context for the callback. In other words, this is the MyApp object, inside of the initializer function.

The options parameters is passed from the start method (see below).

Initializer callbacks are guaranteed to run, no matter when you add them to the app object. If you add them before the app is started, they will run when the start method is called. If you add them after the app is started, they will run immediately.

The Application object raises a few events during its lifecycle. These events can be used to do additional processing of your application. For example, you may want to pre-process some data just before initialization happens. Or you may want to wait until your entire application is initialized to start the Backbone.history.

The events that are currently triggered, are:

• "initialize:before": fired just before the initializers kick off
• "initialize:after": fires just after the initializers have finished
• "start": fires after all initializers and after the initializer events

MyApp.bind("initialize:before", function(options){
  options.moreData = "Yo dawg, I heard you like options so I put some options in your options!"
});

MyApp.bind("initialize:after", function(options){
  if (Backbone.history){
    Backbone.history.start();
  }
});

The options parameter is passed through the start method of the application object (see below).

Once you have your application configured, you can kick everything off by calling: MyApp.start(options).

This function takes a single optional parameter. This parameter will be passed to each of your initializer functions, as well as the initialize events. This allows you to provide extra configuration for various parts of your app, at initialization/start of the app, instead of just at definition.

var options = {
  something: "some value",
  another: "#some-selector"
};

MyApp.start(options);

Every application instance comes with an instance of Marionette.EventAggregator called app.vent.

MyApp = new Backbone.Marionette.Application();

MyApp.vent.on("foo", function(){
  alert("bar");
});

MyApp.vent.trigger("foo"); // => alert box "bar"

See the Marionette.EventAggregator documentation below, for more details.

Marionette allows you to define a module within your application, including sub-modules hanging from that module.

This is useful for creating modular, encapsulated applications that are be split apart in to multiple files.

By default, every module that you define and add to your application, including sub-modules, is a new Application object instance. This allows you to have unlimited sub-modules and nesting. It also means you have all of the features and functionality of an Application object for each module.

A module is defined directly from an Application object, and created another Application object as the specified name:

var MyApp = new Backbone.Marionette.Application();

var myModule = MyApp.module("MyModule");

MyApp.MyModule; // => a new Marionette.Application object

myModule === MyApp.MyModule; // => true

If you specify the same module name more than once, the first instance of the module will be retained and a new instance will not be created.

You can specify a callback function to provide a definition for the module. Module definitions are invoked immediately on calling module method.

The module definition callback will receive 6 parameters:

• The module itself
• The Parent module or Application object
• Backbone
• Backbone.Marionette
• jQuery
• Underscore

You can add functions and data directly to your module to make them publicly accessible. You can also add private functions and data by using locally scoped variables.

MyApp.module("MyModule", function(MyModule, MyApp, Backbone, Marionette, $, _){

  // Private Data And Functions
  // --------------------------

  var myData = "this is private data";
 
  var myFunction = function(){
    console.log(myData);
  }


  // Public Data And Functions
  // -------------------------

  MyModule.someData = "public data";

  MyModule.someFunction = function(){
    console.log(MyModule.someData);
  }
});

console.log(MyApp.MyModule.someData); //=> public data
MyApp.MyModule.someFunction(); //=> public data

Sub-modules can be defined in a number of ways.

You can define a module, and then later use that module to define sub-modules:

MyModule = MyApp.module("MyModule");
MyModule.module("SubModule");

MyApp.MyModule.SubModule; // => a valid module object

Or you can define an entire hierarchy of modules and sub-modules all at once:

MyApp.module("Parent.Child.GrandChild");

MyApp.Parent.Child.GrandChild; // => a valid module object

When defining sub-modules using the dot-notation, the parent modules do not need to exist. They will be created for you if they don't exist. If they do exist, though, the existing module will be used instead of creating a new one.

As each module is an Application object instance, initializer functions can be registerd on your modules. There is no automatic starting of these modules, though. If you wish to use the module's initializer functions, you must manually call the module's start method.

MyApp.module("MyModule", function(MyModule){

  MyModule.addInitializer(function(){
    console.log("I'm a module initializer!");
  });

});

MyApp.MyModule.start(); // => "I'm a module initializer!"

If you want to override the module object with your own object, you can do that by returning an object from the module definition:

MyApp.module("MyModule", function(){
  var customModule = {};


  return customModule;
});

MyApp.MyModule; // => the 'customModule' object

Note: When you override a module with a custom object, you will not be able to add sub-modules to it using the module fuction, as this function will not exist on your module (unless you add it yourself).

Reduce the boilerplate code of handling route events and then calling a single method on another object. Have your routers configured to call the method on your object, directly.

Configure an AppRouter with appRoutes. The route definition is passed on to Backbone's standard routing handlers. This means that you define routes like you normally would. Instead of providing a callback method that exists on the router, though, you provide a callback method that exists on the controller that you specify for the router instance (see below).

MyRouter = Backbone.Marionette.AppRouter.extend({
  appRoutes: {
    "some/route": "someMethod"
  }
});

You can also add standard routes to an AppRouter, with methods on the router.

App routers can only use one controller object. You can either specify this directly in the router definition:

someController = {
  someMethod: function(){ /*...*/ }
};

Backbone.Marionette.AppRouter.extend({
  controller: someController
});

Or in a parameter to the contructor:

myObj = {
  someMethod: function(){ /*...*/ }
};

new MyRouter({
  controller: myObj
});

Or

The object that is used as the controller has no requirements, other than it will contain the methods that you specified in the appRoutes.

It is recommended that you divide your controller objects into smaller pieces of related functionality and have multiple routers / controllers, instead of just one giant router and controller.

Region managers provide a consistent way to manage your views and show / close them in your application. They use a jQuery selector to show your views in the correct place. They also call extra methods on your views to facilitate additional functionality.

Regions can be added to the application by calling the addRegions method on your application instance. This method expects a single hash parameter, with named regions and either jQuery selectors or Region objects. You may call this method as many times as you like, and it will continue adding regions to the app.

MyApp.addRegions({
  mainRegion: "#main-content",
  navigationRegion: "#navigation"
});

As soon as you call addRegions, your region managers are available on your app object. In the above, example MyApp.mainRegion and MyApp.navigationRegion would be available for use immediately.

If you specify the same region name twice, the last one in wins.

You can specify an el for the region manager to manage at the time that the region manager is instantiated:

var mgr = new Backbone.Marionette.Region({
  el: "#someElement"
});

Once a region manager has been defined, you can call the show and close methods on it to render and display a view, and then to close that view:

var myView = new MyView();

// render and display the view
MyApp.mainRegion.show(myView);

// closes the current view
MyApp.mainRegion.close();

If you replace the current view with a new view by calling show, it will automatically close the previous view.

// show the first view
var myView = new MyView();
MyApp.mainRegion.show(myView);

// replace view with another. the
// `close` method is called for you
var anotherView = new AnotherView();
MyApp.mainRegion.show(anotherView);

You can specify a second parameter to the show method, which will be used to determine how the HTML from the view's el is attached to the DOM region that is being managed.

The options include any valid jQuery DOM object method, such as html, text, append, etc.

MyApp.mainRegion.show(myView, "append");

This example will use jQuery's $.append function to append the new view to the current HTML.

WARNING: Be careful when using this feature, as the view you are replacing may not be managed / closed correctly as a result. This can cause unexpected behavior, memory leaks or other problems. Use At Your Own Risk

There are some scenarios where it's desirable to attach an existing view to a region manager, without rendering or showing the view, and without replacing the HTML content of the region. For example, SEO and accessibiliy often need HTML to be generated by the server, and progressive enhancement of the HTML.

There are two ways to accomplish this:

• set the currentView in the region manager's constructor
• call attachView on the region manager instance

var myView = new MyView({
  el: $("#existing-view-stuff")
});

var manager = new Backbone.Marionette.Region({
  el: "#content",
  currentView: myView
});

MyApp.addRegions({
  someRegion: "#content"
});

var myView = new MyView({
  el: $("#existing-view-stuff")
});

MyApp.someRegion.attachView(myView);

A region manager will raise a few events during it's showing and closing of views:

• "view:show" - when the view has been rendered and displayed
• "onShow" - called on the region when the view has been rendered
• "view:closed" - when the view has been closed

You can bind to these events and add code that needs to run with your region manager, opening and closing views.

MyApp.mainRegion.on("view:show", function(view){
  // manipulate the `view` or do something extra
  // with the region manager via `this`
});

MyApp.mainRegion.on("view:closed", function(view){
  // manipulate the `view` or do something extra
  // with the region manager via `this`
});

MyRegion = Backbone.Marionette.Region.extend({
  // ...

  onShow: function(view){
    // the `view` has been shown
  }
});

The region manager will call an onShow method on the view that was displayed. It will also trigger a "show" event from the view:

MyView = Backbone.View.extend({
  onShow: function(){
    // the view has been shown
  }
});

view = new MyView();

view.on("show", function(){
  // the view has been shown.
});

MyApp.mainRegion.show(view);

You can define a custom region manager by extending from Region. This allows you to create new functionality, or provide a base set of functionality for your app.

Once you define a region manager type, you can still call the addRegions method. Specify the region manager type as the value - not an instance of it, but the actual constructor function.

var FooterRegion = Backbone.Marionette.Region.extend({
  el: "#footer"
});

MyApp.addRegions({footerRegion: FooterRegion});

Note that if you define your own Region object, you must provide an el for it. If you don't, you will receive an runtime exception saying that an el is required.

There may be times when you want to add a region manager to your application after your app is up and running. To do this, you'll need to extend from Region as shown above and then use that constructor function on your own:

var SomeRegion = Backbone.Marionette.Region.extend({
  el: "#some-div",

  initialize: function(options){
    // your init code, here
  }
});

MyApp.someRegion = new SomeRegion();

MyApp.someRegion.show(someView);

You can optionally add an initialize function to your Region definition as shown in this example. It receives the options that were passed to the constructor of the Region, similar to a Backbone.View.

The region manager show method takes advantage of jQuery's deferred cababilities, allowing for some very advanced techniques to be used for rendering views.

To use a deferred, a view that is shown via a region manager must return a jQuery deferred object from the render method:

DeferredView = Backbone.View.extend({
  render: function(){
    var that = this;
    var data = this.serializeData();
    var dfd = jQuery.Deferred();

    this.getTemplate(function(template){
      var html = that.renderTemplate(template, data);

      that.$el.html(html);

      if (that.onRender){
        that.onRender();
      }

      dfd.resolve();
    });

    return dfd.promise();
  }
});

var view = new DeferredView();
MyApp.mainRegion.show(view);

The region manager will wait until the deferred object is resolved before it attached the view's el to the DOM and displays it.

Marionette has a base Marionette.View type that other views extend from. This base view provides some common and core functionality for other views to take advantage of.

Note: The Marionette.View type is not intended to be used directly. It exists as a base view for other view types to be extended from, and to provide a common location for behaviors that are shared across all views.

Views can define a set of triggers as a hash, which will convert a DOM event in to a view.trigger event.

The left side of the hash is a standard Backbone.View DOM event configuration, while the right side of the hash is the view event that you want to trigger from the view.

MyView = Backbone.Marionette.ItemView.extend({
  // ...

  triggers: {
    "click .do-something": "something:do:it"
  }
});

view = new MyView();
view.render();

view.on("something:do:it", function(){
  alert("I DID IT!");
});

// "click" the 'do-something' DOM element to 
// demonstrate the DOM event conversion
view.$(".do-something").trigger("click");

The result of this is an alert box that says, "I DID IT!"

You can also specify the triggers as a function that returns a hash of trigger configurations

Backbone.Marionette.CompositeView.extend({
  triggers: function(){
    return {
      "click .that-thing": "that:i:sent:you"
    };
  }
});

Triggers work with all View types that extend from the base Marionette.View.

The serializeData method will serialize a view's model or collection - with precedence given to collections. That is, if you have both a collection and a model in a view, calling the serializeData method will return the serialized collection.

There are times when a view's template needs to have some logic in it, and the view engine itself will not provide an easy way to accomplish this. For example, Underscore templates do not provide a helper method mechanism while Handlebars templates do.

A templateHelpers attribute can be applied to any View object that uses the serializeData method - including ItemViews, Layouts and CompositeViews. When this attribute is present, it's contents will be mixed in to the data object that comes back from the serializeData method for you. This will allow you to create helper methods that can be called from within your templates.

<script id="my-template" type="text/html">
  I think that <%= showMessage() %>
</script>

MyView = Backbone.Marionette.ItemView.extend({
  template: "#my-template",

  templateHelpers: {
    showMessage: function(){
      return this.name + " is the coolest!"
    }
  }

});

model = new Backbone.Model({name: "Backbone.Marionette"});
view = new MyView();
view.render(); //=> "I think that Backbone.Marionette is the coolest!";

In order to access data from within the helper methods, you need to prefix the data you need with this. Doing that will give you all of the methods and attributes of the serialized data object, including the other helper methods.

templateHelpers: {
  something: function(){
    return "Do stuff with " + this.name + " because it's awesome.";
  }
}

You can specify an object literal (as shown above), a reference to an object literal, or a function as the templateHelpers.

If you specify a function, the function will be invoked with the current view instance as the context of the function. The function must return an object that can be mixed in to the data for the view.

Backbone.Marionette.ItemView.extend({
  templateHelpers: function(){
    return {
      foo: function(){ /* ... */ }
    }
  }
});

Formerly known as CompositeRegion.

A Layout is a specialized hybrid between an ItemView and a collection of Region objects, used for rendering an application layout with multiple sub-regions to be managed by specified region managers.

A layout manager can also be used as a composite-view to aggregate multiple views and sub-application areas of the screen where multiple region managers need to be attached to dynamically rendered HTML.

The Layout extends directly from ItemView and adds the ability to specify regions which become Region instances that are attached to the layout.

<script id="layout-template" type="text/template">
  <section>
    <navigation id="menu">...</navigation>
    <article id="content">...</navigation>
  </section>
</script>

AppLayout = Backbone.Marionette.Layout.extend({
  template: "#layout-template",

  regions: {
    menu: "#menu",
    content: "#content"
  }
});

var layout = new AppLayout();
layout.render();

Once you've rendered the layout, you now have direct access to all of the specified regions as region managers.

layout.menu.show(new MenuView());

layout.content.show(new MainContentView());

Since the Layout extends directly from ItemView, it has all of the core functionality of an item view. This includes the methods necessary to be shown within an existing region manager.

MyApp = new Backbone.Marionette.Application();
MyApp.addRegions({
  mainRegion: "#main"
});

var layout = new AppLayout();
MyApp.mainRegion.show(layout);

layout.show(new MenuView());

You can nest layouts into region managers as deeply as you want. This provides for a well organized, nested view structure.

When you are finished with a layout, you can call the close method on it. This will ensure that all of the region managers within the layout are closed correctly, which in turn ensures all of the views shown within the regions are closed correctly.

If you are showing a layout within a parent region manager, replacing the layout with another view or another layout will close the current one, the same it will close a view.

All of this ensures that layouts and the views that they contain are cleaned up correctly.

It's common to use a Layout to represent a sub-application in a larger overall application. Often the components of the sub-application need to communicate with each other without allowing the other parts of the larger application in on the communication. To facilitate this, the layout manager includes an event aggregator, vent.

var layout = new MyAppLayout();

layout.vent.trigger("stuff:was:done");

An ItemView is a view that represents a single item. That item may be a Backbone.Model or may be a Backbone.Collection. Whichever it is, though, it will be treated as a single item.

An item view has a render method built in to it, and uses the Renderer object to do the actual rendering.

The render function of the item view will return a jQuery promise object.

You should provide a template attribute on the item view, which will be either a jQuery selector:

MyView = Backbone.Marionette.ItemView.extend({
  template: "#some-template"
});

new MyView().render().done(function(){
  // the view is done rendering. do stuff here
});

or a function that returns a jQuery selector:

MyView = Backbone.Marionette.ItemView.extend({
  template: function(){
    if (this.model.get("foo")){
      return "#some-template";
    } else {
      return "#a-different-template";
    }
  }
});

new MyView().render().done(function(){
  // the view is done rendering. do stuff here
});

There are several callback methods that are called for an ItemView. These methods are intended to be handled within the view definition, directly.

Before an ItemView is rendered a beforeRender method will be called on the view.

Backbone.Marionette.ItemView.extend({
  beforeRender: function(){
    // set up final bits just before rendering the view's `el`
  }
});

After the view has been rendered, a onRender method will be called. You can implement this in your view to provide custom code for dealing with the view's el after it has been rendered:

Backbone.Marionette.ItemView.extend({
  onRender: function(){
    // manipulate the `el` here. it's already
    // been rendered, and is full of the view's
    // HTML, ready to go.
  }
});

A beforeClose method will be called on the view, just prior to closing it:

Backbone.Marionette.ItemView.extend({
  beforeClose: function(){
    // manipulate the `el` here. it's already
    // been rendered, and is full of the view's
    // HTML, ready to go.
  }
});

An onClose method will be called on the view, after closing it.

Backbone.Marionette.ItemView.extend({
  onClose: function(){
    // custom closing and cleanup goes here
  }
});

There are several events that are triggers by an ItemView, which allow code outside of a view to respond to what's happening with the view.

An "item:before:render" event will be triggered just before the view is rendered

MyView = Backbone.Marionette.ItemView.extend({...});

var myView = new MyView();

myView.on("item:before:render", function(){
  alert("the view is about to be rendered");
});

An "item:rendered" event will be triggered just after the view has been rendered.

MyView = Backbone.Marionette.ItemView.extend({...});

var myView = new MyView();

myView.on("item:rendered", function(){
  alert("the view was rendered!");
});

myView.on("render", function(){
  alert("the view was rendered!");
});

An "item:before:close" event will be triggered just prior to the view closing itself. This event fires when the close method of the view is called.

MyView = Backbone.Marionette.ItemView.extend({...});

var myView = new MyView();

myView.on("item:before:close", function(){
  alert("the view is about to be closed");
});

myView.close();

An "item:closed" event will be triggered just after the view closes. This event fires when the close method of the view is called.

MyView = Backbone.Marionette.ItemView.extend({...});

var myView = new MyView();

myView.on("item:closed", function(){
  alert("the view is closed");
});

myView.close();

Item views will serialize a model or collection, by default, by calling .toJSON on either the model or collection. If both a model and collection are attached to an item view, the model will be used as the data source. The results of the data serialization will be passed to the template that is rendered.

If the serialization is a model, the results are passed in directly:

var myModel = new MyModel({foo: "bar"});

new MyItemView({
  template: "#myItemTemplate",
  model: myModel
});

MyItemView.render();

<script id="myItemTemplate" type="template">
  Foo is: <%= foo %>
</script>

If the serialization is a collection, the results are passed in as an items array:

var myCollection = new MyCollection([{foo: "bar"}, {foo: "baz"}]);

new MyItemView({
  template: "#myCollectionTemplate",
  collection: myCollection
});

MyItemView.render();

<script id="myCollectionTemplate" type="template">
  <% _.each(items, function(item){ %>
    Foo is: <%= foo %>
  <% }); %>
</script>

If you need custom serialization for your data, you can provide a serializeData method on your view. It must return a valid JSON object, as if you had called .toJSON on a model or collection.

Backbone.Marionette.ItemView.extend({
  serializeData: function(){
    return {
      "some attribute": "some value"
    }
  }
});

ItemView extends Marionette.BindTo. It is recommended that you use the bindTo method to bind model and collection events.

MyView = Backbone.Marionette.ItemView.extend({
  initialize: function(){
    this.bindTo(this.model, "change:foo", this.modelChanged);
    this.bindTo(this.collection, "add", this.modelAdded);
  },

  modelChanged: function(model, value){
  },

  modelAdded: function(model){
  }
});

The context (this) will automatically be set to the view. You can optionally set the context by passing in the context object as the 4th parameter of bindTo.

ItemView implements a close method, which is called by the region managers automatically. As part of the implementation, the following are performed:

• unbind all bindTo events
• unbind all custom view events
• unbind all DOM events
• remove this.el from the DOM
• call an onClose event on the view, if one is provided

By providing an onClose event in your view definition, you can run custom code for your view that is fired after your view has been closed and cleaned up. This lets you handle any additional clean up code without having to override the close method.

Backbone.Marionette.ItemView.extend({
  onClose: function(){
    // custom cleanup or closing code, here
  }
});

The CollectionView will loop through all of the models in the specified collection, render each of them using a specified itemView, then append the results of the item view's el to the collection view's el.

There are several callback methods that can be provided on a CollectionView. If they are found, they will be called by the view's base methods. These callback methods are intended to be handled within the view definition directly.

A beforeRender callback will be called just prior to rendering the collection view.

Backbone.Marionette.CollectionView.extend({
  beforeRender: function(){
    // do stuff here
  }
});

After the view has been rendered, a onRender method will be called. You can implement this in your view to provide custom code for dealing with the view's el after it has been rendered:

Backbone.Marionette.CollectionView.extend({
  onRender: function(){
    // do stuff here
  }
});

This method is called just before closing the view.

Backbone.Marionette.CollectionView.extend({
  beforeClose: function(){
    // do stuff here
  }
});

This method is called just after closing the view.

Backbone.Marionette.CollectionView.extend({
  onClose: function(){
    // do stuff here
  }
});

There are several events that will be triggered during the life of a collection view. These are intended to be handled from code external to the view.

Triggers just prior to the view being rendered

MyView = Backbone.Marionette.CollectionView.extend({...});

var myView = new MyView();

myView.on("collection:before:render", function(){
  alert("the collection view is about to be rendered");
});

myView.render();

A "collection:rendered" event will also be fired. This allows you to add more than one callback to execute after the view is rendered, and allows parent views and other parts of the application to know that the view was rendered.

MyView = Backbone.Marionette.CollectionView.extend({...});

var myView = new MyView();

myView.on("render", function(){
  alert("the collection view was rendered!");
});

myView.on("collection:rendered", function(){
  alert("the collection view was rendered!");
});

myView.render();

Triggered just before closing the view.

MyView = Backbone.Marionette.CollectionView.extend({...});

var myView = new MyView();

myView.on("collection:before:close", function(){
  alert("the collection view is about to be closed");
});

myView.close();

Triggered just after closing the view.

MyView = Backbone.Marionette.CollectionView.extend({...});

var myView = new MyView();

myView.on("collection:closed", function(){
  alert("the collection view is now closed");
});

myView.close();

When an item view within a collection view triggers an event, that event will bubble up through the parent collection view, with "itemview:" prepended to the event name.

That is, if a child view triggers "do:something", the parent collection view will then trigger "itemview:do:something".

// set up basic collection
var myModel = new MyModel();
var myColelction = new MyCollection();
myCollection.add(myModel);

// get the collection view in place
colView = new CollectionView({
  collection: myCollection
});
colView.render();

// bind to the collection view's events that were bubbled
// from the child view
colView.on("itemview:do:something", function(childView, msg){
  alert("I said, '" + msg + "'");
});

// hack, to get the child view and trigger from it
var childView = colView.children[myModel.cid];
childView.trigger("do:something", "do something!");

The result of this will be an alert box that says "I said, 'do something!'".

Also note that you would not normally grab a reference to the child view the way this is showing. I'm merely using that hack as a way to demonstrate the event bubbling. Normally, you would have your item view listening to DOM events or model change events, and then triggering an event of it's own based on that.

The render method of the collection view is responsible for rendering the entire collection. It loops through each of the items in the collection and renders them individually as an itemView.

The render method returns a jQuery deferred object, allowing you to know when the rendering completes. This deferred object is resolved after all of the child views have been rendered.

MyCollectionView = Backbone.Marionette.CollectionView.extend({...});

new MyCollectionView().render().done(function(){
  // all of the children are now rendered. do stuff here.
});

Specify an itemView in your collection view definition. This must be a Backbone view object definition (not instance). It can be any Backbone.View or be derived from Marionette.ItemView.

MyItemView = Backbone.Marionette.ItemView.extend({});

Backbone.Marionette.CollectionView.extend({
  itemView: MyItemView
});

Alternatively, you can specify an itemView in the options for the constructor:

MyCollectionView = Backbone.Marionette.CollectionView.extend({...});

new MyCollectionView({
  itemView: MyItemView
});

If you do not specify an itemView, an exception will be thrown stating that you must specify an itemView.

The collection view binds to the "add", "remove" and "reset" events of the collection that is specified.

When the collection for the view is "reset", the view will call render on itself and re-render the entire collection.

When a model is added to the collection, the collection view will render that one model in to the collection of item views.

When a model is removed from a collection (or destroyed / deleted), the collection view will close and remove that model's item view.

If you need to re-render the entire collection, you can call the view.render method. This method takes care of closing all of the child views that may have previously been opened.

By default the collection view will call jQuery's .append to move the HTML contents from the item view instance in to the collection view's el.

You can override this by specifying an appendHtml method in your view definition. This method takes two parameters and has no return value.

Backbone.Marionette.CollectionView.extend({
  appendHtml: function(collectionView, itemView){
    collectionView.$el.prepend(itemView.el);
  }
});

The first parameter is the instance of the collection view that will receive the HTML from the second parameter, the current item view instance.

CollectionView implements a close method, which is called by the region managers automatically. As part of the implementation, the following are performed:

• unbind all bindTo events
• unbind all custom view events
• unbind all DOM events
• unbind all item views that were rendered
• remove this.el from the DOM
• call an onClose event on the view, if one is provided

By providing an onClose event in your view definition, you can run custom code for your view that is fired after your view has been closed and cleaned up. This lets you handle any additional clean up code without having to override the close method.

Backbone.Marionette.CollectionView.extend({
  onClose: function(){
    // custom cleanup or closing code, here
  }
});

A CompositeView extends from CollectionView to be used as a composite view for scenarios where it should represent both a branch and leaf in a tree structure, or for scenarios where a collection needs to be rendered within a wrapper template.

For example, if you're rendering a treeview control, you may want to render a collection view with a model and template so that it will show a parent item with children in the tree.

You can specify a modelView to use for the model. If you don't specify one, it will default to the Marionette.ItemView.

CompositeView = Backbone.Marionette.CompositeView.extend({
  template: "#leaf-branch-template"
});

new CompositeView({
  model: someModel,
  collection: someCollection
});

For more examples, see my blog post on using the composite view

A composite view returns a jQuery deferred object from the render method. This allows you to know when the rendering for the entire composite structure has been completed.

MyComp = Backbone.Marionette.CompositeView.extend({...});

myComp = new MyComp().render().done(function(){
  // the entire composite is now rendered. do stuff here
});

When a CompositeView is rendered, the model will be rendered with the template that the view is configured with. You can override the template by passing it in as a constructor option:

new MyComp({
  template: "#some-template"
});

The default rendering mode for a CompositeView assumes a hierarchical, recursive structure. If you configure a composite view without specifying an itemView, you'll get the same composite view type rendered for each item in the collection. If you need to override this, you can specify a itemView in the composite view's definition:

var ItemView = Backbone.Marionette.ItemView.extend({});

var CompView = Backbone.Marionette.CompositeView.extend({
  itemView: ItemView
});

The model and collection for the composite view will re-render themselves under the following conditions:

• When the collection's "reset" event is fired, it will re-render the entire list
• When the collection has a model added to it (the "add" event is fired), it will render that one item to the rendered list
• When the collection has a model removed (the "remove" event is fired), it will remove that one item from the rendered list

You can also manually re-render either or both of them:

• If you want to re-render everything, call the .render() method
• If you want to re-render the model's view, you can call .renderModel()
• If you want to re-render the collection's views, you can call .renderCollection()

During the course of rendering a composite, several events will be triggered:

• "composite:item:rendered" - after the modelView has been rendered
• "composite:collection:rendered" - after the collection of models has been rendered
• "render" / "composite:rendered" - after everything has been rendered

Additionally, after the composite view has been rendered, an onRender method will be called. You can implement this in your view to provide custom code for dealing with the view's el after it has been rendered:

Backbone.Marionette.CompositeView.extend({
  onRender: function(){
    // do stuff here
  }
});

An event aggregator is an application level pub/sub mechanism that allows various pieces of an otherwise segmented and disconnected system to communicate with each other.

Backbone.Marionette provides an event aggregator with each application instance: MyApp.vent. You can also instantiate your own event aggregator:

myVent = new Backbone.Marionette.EventAggregator();

Passing an object literal of options to the constructor function will extend the event aggregator with those options:

myVent = new Backbone.Marionette.EventAggregator({foo: "bar"});
myVent.foo // => "bar"

The EventAggregator extends from the BindTo object (see below) to easily track and unbind all event callbacks, including inline callback functions.

The bindTo method, though, has been proxied to only take 3 arguments. It assumes that the object being bound to is the event aggregator directly, and does not allow the bound object to be specified:

vent = new Backbone.Marionette.EventAggregator();

vent.bindTo("foo", function(){
  alert("bar");
});

vent.unbindAll();

vent.trigger("foo"); // => nothing. all events have been unbound.

You can use an event aggregator to communicate between various modules of your application, ensuring correct decoupling while also facilitating functionality that needs more than one of your application's modules.

var vent = new Backbone.Marionette.EventAggregator();

vent.bind("some:event", function(){
  alert("Some event was fired!!!!");
});
  
vent.trigger("some:event");

For a more detailed discussion and example of using an event aggregator with Backbone applications, see the blog post: References, Routing, and The Event Aggregator: Coordinating Views In Backbone.js

The BindTo object provides event binding management and facilitates simple event binding and unbinding for any object that extends from Backbone.Events.

var binder = _.extend({}, Backbone.Marionette.BindTo);

var model = new MyModel();

var handler = {
  doIt: function(){}
}
binder.bindTo(model, "change:foo", handler.doIt);

You can optionally specify a 4th parameter as the context in which the callback method for the event will be executed:

binder.bindTo(model, "change:foo", someCallback, someContext);

When you call bindTo, it returns a "binding" object that can be used to unbind from a single event with the unbindFrom method:

var binding = binder.bindTo(model, "change:foo", someCallback, someContext);

// later in the code
binder.unbindFrom(binding);

This will unbind the event that was configured with the binding object, and remove it from the BindTo bindings.

You can call unbindAll to unbind all events that were bound with the bindTo method:

binder.unbindAll();

This even works with in-line callback functions.

The Renderer object was extracted from the ItemView rendering process, in order to create a consistent and re-usable method of rendering a template with or without data.

The basic usage of the Renderer is to call the render method. This method returns a jQuery promise object, which will provide the HTML that was rendered when it resolves.

var template = "#some-template";
var data = {foo: "bar"};
var render = Backbone.Marionette.Renderer.render(template, data);

render.done(function(html){
  // do something with the HTML here
});

By default, the renderer will take a jQuery selector object as the first parameter, and a JSON data object as the optional second parameter. It then uses the TemplateCache to load the template by the specified selector, and renders the template with the data provided (if any) using Underscore.js templates.

If you wish to override the way the template is loaded, see the TemplateCache object.

If you wish to override the template engine used, change the renderTemplate method to work however you want:

Backbone.Marionette.Renderer.renderTemplate = function(template, data){
  return $(template).tmpl(data);
});

This implementation will replace the default Underscore.js rendering with jQuery templates rendering.

It's common for developers to use RequireJS to pre-load the entire template that is needed. When the template is pre-loaded like this, the use of a jQuery selector is not needed.

If you wish to replace the entire rendering process, based on the assumption that you will always pre-load the entire template for your view, it can be done by replacing the render method on the Renderer object:

Backbone.Marionette.Renderer.render = function(template, data){
  return _.template(template, data);
};

This will skip the TemplateCache usage entirely, and all of the other code that is usually run, and return the compiled template immediately.

There's a performance hit in replacing the entire render function of the Renderer object as shown in the example above. You should only have to compile a template once, and you should be caching the pre-compiled template after that.

To accomplish this, avoid replacing the render function. Instead use a combination of the renderTemplate function and the TemplateCache.loadTemplate function.

Backbone.Marionette.TemplateCache.loadTemplate = function(template, callback){
  // pre-compile the template and store that in the cache.
  var compiledTemplate = _.template(template);
  callback.call(this, compiledTemplate);
};

Backbone.Marionette.Renderer.renderTemplate = function(template, data){
  // because `template` is the pre-compiled template object,
  // we only need to execute the template with the data
  return template(data);
}

For more information about the TemplateCache.loadTemplate function, see the next section of the documentation.

The TemplateCache provides a cache for retrieving templates from script blocks in your HTML. This will improve the speed of subsequent calls to get a template.

To use the TemplateCache, call it directly. It is not instantiated like other Marionette objects.

Templates are retrieved using a jQuery selector by default, and are handed back to you via a callback method. The template is returned as a plain string.

Backbone.Marionette.TemplateCache.get("#my-template", function(template){
 // use the template here
});

Making multiple calls to get the same template will retrieve the template from the cache on subsequence calls:

var a, b, c;
Backbone.Marionette.TemplateCache.get("#my-template", function(tmpl){a = tmpl});
Backbone.Marionette.TemplateCache.get("#my-template", function(tmpl){b = tmpl});
Backbone.Marionette.TemplateCache.get("#my-template", function(tmpl){c = tmpl});
a === b === c; // => true

The default template retrieval is to select the template contents from the DOM using jQuery. If you wish to change the way this works, you can override the loadTemplate method on the TemplateCache object.

Backbone.Marionette.TemplateCache.loadTemplate = function(templateId, callback){
  // load your template here, returning it or a deferred
  // object that resolves with the template as the only param
}

For example, if you want to load templates asychronously from the server, instead of from the DOM, you could replace loadTemplate function.

If a "template.html" file exists on the server, with this in it:

<script id="my-template" type="text/template">
  <div>some template stuff</div>
</script>

Then the loadTemplate implementation may look like this:

Backbone.Marionette.TemplateCache.loadTemplate = function(templateId, callback){
  var that = this;
  var url = templateId + ".html";

  $.get(url, function(templateHtml){
    var template = $(tmplateHtml).find(templateId);
    callback(template);
  });
}

This will use jQuery to asynchronously retrieve the template from the server. When the get completes, the callback function will select the template from the resulting HTML and then call the callback function to send it in to the template cache and allow it to be used for rendering.

You can clear one or more, or all items from the cache using the clear method. Clearing a template from the cache will force it to re-load from the DOM (or from the overriden loadTemplate function) the next time it is retrieved.

If you do not specify any parameters, all items will be cleared from the cache:

Backbone.Marionette.TemplateCache.get("#my-template");
Backbone.Marionette.TemplateCache.get("#this-template");
Backbone.Marionette.TemplateCache.get("#that-template");

// clear all templates from the cache
Backbone.Marionette.TemplateCache.clear()

If you specify one or more parameters, these parameters are assumed to be the templateId used for loading / caching:

Backbone.Marionette.TemplateCache.get("#my-template");
Backbone.Marionette.TemplateCache.get("#this-template");
Backbone.Marionette.TemplateCache.get("#that-template");

// clear 2 of 3 templates from the cache
Backbone.Marionette.TemplateCache.clear("#my-template", "#this-template")

If you're using Marionette.ItemView, you don't need to manually call the TemplateCache. Just specify the template attribute of your view as a jQuery selector, and the ItemView will use the template manager by default.

The Callbacks object assists in managing a collection of callback methods, and executing them, in an async-safe manner.

There are only three methods:

• add
• run
• setOptions

The add method adds a new callback to be executed later.

The run method executes all current callbacks in, using the specified context for each of the callbacks, and supplying the provided options to the callbacks.

var callbacks = new Backbone.Marionette.Callbacks();

callbacks.add(function(options){
  alert("I'm a callback with " + options.value + "!");
});

callbacks.run(someContext, {value: "options"});

This example will display an alert box that says "I'm a callback with options!". The executing context for each of the callback methods has been set to the someContext object, which can be any valid JavaScript object.

The Callbacks executes each callback in an async-friendly manner, and can be used to facilitate async callbacks. The Marionette.Application object uses Callbacks to manage initializers (see above).

It can also be used to guarantee callback execution in an event driven scenario, much like the application initializers.

Theses libraries are require for the use, development, testing and documentation of Backbone.Marionette.

Backbone.Marionette currently works with the following versions of these libraries:

• Backbone v0.9.2
• Underscore v1.3.1
• jQuery v1.7.2

While support for Zepto and Enderjs has been added, it is not officially tested against these libraries at this time.

Marionette has not been tested against any other versions of these libraries. You may or may not have success if you use a version other than what it listed here.

I use a number of tools to build, test and maintain Marionette, including but not limited to:

The Anvil.js project is used to generate the builds for Backbone.Marionette. You can run the build.sh file from a terminal, or the build.cmd file for Windows users, after installing anvil.

Backbone.Marionette is also tested with the Jasmine JavaScript test utility, using the Jasmine Ruby gem.

To get the test suite up and running, you need a Ruby installation with the latest RubyGems. Install the 'bundler' gem and then run 'bundle install' from the project's root folder. Then run rake jasmine to run the test suite, and load up http://localhost:8888 to see the test suite in action.

I'm using DocumentUp.com to format the readme / documentation at http://derickbailey.github.com/backbone.marionette

I'm using Docco to generate the annotated source code.

• Views now have the ability to define triggers which will convert a DOM event in to a view.trigger event

• Module definition functions will only be applied to the last module in the . chain

• Added modules and sub-modules through the Application object

• An itemView instance as part of a Collection View or Composite View, will have it's events bubbled up through the parent view, prepended with "itemview:" as the event name

• The onBefore method of ItemView can now return a deferred object
• Code cleanup for rendering methods

• Fixed issue with unbindAll in BindTo, that was skipping some items

• The bindTo method on the EventAggregator now returns a binding configuration object
• Automatic mixing in of templateMethods as template / view helper methods, in views that use the serializeData function
• A friendlier error message will be thrown from an appRouter if a route is configured with a method that does not exist on the controller

• Extracted compileTemplate method in TemplateCache for clarity and easier modification
• ItemView will wait until onRender has completed before triggering other rendered events
• Region now supports an onShow method, when defining a custom region
• Moved the default serializeData method to the base Marionette.View
• CompositeView now calls the serializeData method to get the model's data for the view
• BindTo changes:
  • The bindTo method returns a "binding" object so that it can be unbound easily
  • Now has an unbindFrom method that will unbind a binding object

• ItemView now has a renderHtml method that can be overriden to render the item view's data
• Region now supports an initialize function when extending a region to your own object type
• CollectionView correctly defers until all children are rendered
• Underscore templates are cached as pre-compiled templates, instead of re-compiling them on every render
• Updating AMD support to also work with CommonJS / NodeJS
• Correctiong build to include header / license info for all output files
• Pass JSLint with no warnings (run w/ Anvil.js build process)
• Removed GZip release files, as they were broken anyways

• BREAKING: The renderTemplate method has moved from the ItemView prototype on to the Renderer object
• BREAKING: The appendHtml method of the CollectionView now takes collectionView, itemView as the arguments, instead of el, html
• Added Marionette.View object, to contain a few basic parts of every Marionette view
• Added Marionette.Renderer object, to handle template rendering
• Views correctly trigger the "close" events before unbinding event subscribers
• Additional CollectionView changes:
  • Extracted getItemView method to retrieve the itemView type, either from this.itemView or this.options.itemView
  • Extracted buildItemView method to build each item's view
  • Renamed removeChildView to removeItemView to make the language consistent
  • Triggers "item:added" event after each item has been added
  • Triggers "item:removed" event after an item has been removed
• CompositeView changes:
  • No longer takes a modelView. Now directly renders the template specified
  • Defaults to a recurive structure, where itemView is the current composite view type
• A Region will trigger a show event from any view that it shows
• Added common "render" event to all the view types
• Updated to Backbone v0.9.2
• Updated to jQuery v1.7.2
• AMD / RequireJS compliant version is provided
• Now using Anvil.js for builds

For older change logs and release notes, see the changelog file.

Copyright (c) 2012 Derick Bailey, Muted Solutions, LLC

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.