One of the biggest issues with the Backbone documentation is not to have a clue about how to use its objects. You, as developers, should figure out the responsibilities for each object across the application; if you have some experience working with Backbone, then you would know how difficult it would be to build a Backbone application.

In this section, I will describe the best uses of the Backbone objects. Starting at this point, you will have a clearer idea about the scope of responsibilities in Backbone and this will lead the design of our application architecture. Keep in mind that Backbone is a library with only the foundation objects; therefore, you will need to bring your own objects and structure to make scalable, testable, and robust Backbone applications.

It is possible to develop Backbone applications only using the Backbone objects that are described in the previous section; however, for a medium-to-large application, it's not sufficient. We need to introduce a new kind of object with delimited responsibilities that use and coordinate Backbone foundation objects.

In this book, we will develop a simple contacts application in order to demonstrate how to develop Backbone applications following the principles explained throughout this book. The application should be able to list all the available contacts in RESTful API and provide the mechanisms to show and edit them.

The application starts when the Application infrastructure is loaded in the browser and the start() method on it is called. It will bootstrap all the common components and then instantiate all the available routers in the subapplications:

Figure 1.4. Application instantiates all the routers available in the subapplications

// app.js
var App = {
  Models: {},
  Collections: {},
  Routers: {},
  start() {
    // Initialize all available routes
    _.each(_.values(this.Routers), function(Router) {
      new Router();
    });

    // Create a global router to enable sub-applications to
    // redirect to other urls
    App.router = new DefaultRouter();
    Backbone.history.start();
  }
}

The entry point of subapplication is given by its routes, which ideally share the same namespace. For instance, in the contacts subapplication, all the routes start with the contacts/ prefix:

Subapplications should register its routers in the App.Routers global object in order to be initialized. For the Contacts subapplication, the ContactsRouter does the job:

// apps/contacts/router.js
'use strict';

App.Routers = App.Routers || {};

class ContactsRouter extends Backbone.Router {
  constructor(options) {
    super(options);
    this.routes = {
      'contacts': 'showContactList',
      'contacts/page/:page': 'showContactList',
      'contacts/new': 'createContact',
      'contacts/view/:id': 'showContact',
      'contacts/edit/:id': 'editContact'
    };
    this._bindRoutes();
  }

  showContactList(page) {
    // Page should be a postive number grater than 0
    page = page || 1;
    page = page > 0 ? page : 1;

    var app = this.startApp();
    app.showContactList(page);
  }

  createContact() {
    var app = this.startApp();
    app.showNewContactForm();
  }

  showContact(contactId) {
    var app = this.startApp();
    app.showContactById(contactId);
  }

  editContact(contactId) {
    var app = this.startApp();
    app.showContactEditorById(contactId);
  }

  startApp() {
    return App.startSubApplication(ContactsApp);
  }
}

// Register the router to be initialized by the infrastructure
// Application
App.Routers.ContactsRouter = ContactsRouter;

When the user points its browser to one of these routes, a route handler is triggered. The handler function parses the URL and delegates the request to the subapplication façade:

Figure 1.5. Route delegation to Subapplication Façade

The startSubApplication() method in the App object starts a new subapplication and closes any other subapplication that is running at a given time, this is useful to free resources in the user's browser:

var App = {
  // ...
  // Only a subapplication can be running at once, destroy any
  // current running subapplication and start the asked one
  startSubApplication(SubApplication) {
    // Do not run the same subapplication twice
    if (this.currentSubapp &&
        this.currentSubapp instanceof SubApplication) {
      return this.currentSubapp;
    }

    // Destroy any previous subapplication if we can
    if (this.currentSubapp && this.currentSubapp.destroy) {
      this.currentSubapp.destroy();
    }

    // Run subapplication
    this.currentSubapp = new SubApplication({
      region: App.mainRegion
    });
    return this.currentSubapp;
  },
}

The App.mainRegion attribute is an instance of a Region object that points to a DOM element in the page; regions are useful to render views in a contained region of the DOM. We will learn more about this object in Chapter 2, Managing views.

When the subapplication is started, a façade method is called on it to handle the user request. The responsibility of the façade is to fetch the necessary data from the RESTful API and pass the data to a controller:

Figure 1.6. Façade responsibility

// apps/contacts/app.js
'use strict';

class ContactsApp {
  constructor(options) {
    this.region = options.region;
  }

  showContactList() {
    App.trigger('loading:start');
    App.trigger('app:contacts:started');

    new ContactCollection().fetch({
      success: (collection) => {
        // Show the contact list subapplication if
        // the list can be fetched
        this.showList(collection);
        App.trigger('loading:stop');
      },
      fail: (collection, response) => {
        // Show error message if something goes wrong
        App.trigger('loading:stop');
        App.trigger('server:error', response);
      }
    });
  }

  showNewContactForm() {
    App.trigger('app:contacts:new:started');
    this.showEditor(new Contact());
  }

  showContactEditorById(contactId) {
    App.trigger('loading:start');
    App.trigger('app:contacts:started');

    new Contact({id: contactId}).fetch({
      success: (model) => {
        this.showEditor(model);
        App.trigger('loading:stop');
      },
      fail: (collection, response) => {
        App.trigger('loading:stop');
        App.trigger('server:error', response);
      }
    });
  }

  showContactById(contactId) {
    App.trigger('loading:start');
    App.trigger('app:contacts:started');

    new Contact({id: contactId}).fetch({
      success: (model) => {
        this.showViewer(model);
        App.trigger('loading:stop');
      },
      fail: (collection, response) => {
        App.trigger('loading:stop');
        App.trigger('server:error', response);
      }
    });
  }

  showList(contacts) {
    var contactList = this.startController(ContactList);
    contactList.showList(contacts);
  }

  showEditor(contact) {
    var contactEditor = this.startController(ContactEditor);
    contactEditor.showEditor(contact);
  }

  showViewer(contact) {
    var contactViewer = this.startController(ContactViewer);
    contactViewer.showContact(contact);
  }

  startController(Controller) {
    if (this.currentController &&
        this.currentController instanceof Controller) {
      return this.currentController;
    }

    if (this.currentController &&
        this.currentController.destroy) {
      this.currentController.destroy();
    }

    this.currentController = new Controller({
      region: this.region
    });
    return this.currentController;
  }
}

The façade object receives a region object as argument in order to indicate to the subapplication where it should be rendered. The Region objects will be explained in detail in Chapter 2, Managing views.

When the façade is fetching data from the RESTful server, a loading:start event is emitted on the App object in order to allow us to show the loading in progress view for the user. When the loading finishes, it creates and uses a controller that knows how to deal with the model or fetched collection.

The business logic starts when the controller is invoked, it will render all the necessary views for the request and show them to the user, then it will listen for user interactions in the views:

Figure 1.7. Controller creates the necessary views

For the ContactList controller, here is a very simple code:

// apps/contacts/contactLst.js
class ContactList {
  constructor(options) {
    // Region where the application will be placed
    this.region = options.region;

    // Allow subapplication to listen and trigger events,
    // useful for subapplication wide events
    _.extend(this, Backbone.Events);
  }

  showList(contacts) {
    // Create the views
    var layout = new ContactListLayout();
    var actionBar = new ContactListActionBar();
    var contactList = new ContactListView({collection: contacts});

    // Show the views
    this.region.show(layout);
    layout.getRegion('actions').show(actionBar);
    layout.getRegion('list').show(contactList);

    this.listenTo(contactList, 'item:contact:delete',
      this.deleteContact);
  }

  createContact() {
    App.router.navigate('contacts/new', true);
  }

  deleteContact(view, contact) {
    let message = 'The contact will be deleted';
    App.askConfirmation(message, (isConfirm) => {
      if (isConfirm) {
        contact.destroy({
          success() {
            App.notifySuccess('Contact was deleted');
          },
          error() {
            App.notifyError('Ooops... Something went wrong');
          }
        });
      }
    });
  }

  // Close any active view and remove event listeners
  // to prevent zombie functions
  destroy() {
    this.region.remove();
    this.stopListening();
  }
}

The function that handles the request is very simple and follows the same pattern for all other controllers, as follows:

If you don't entirely understand what region and layout means, don't worry, I will cover the implementation of these objects in detail in Chapter 2, Managing views. Here, the important thing is the algorithm described earlier:

Figure 1.8. ContactList controller result

As you can see in the above figure, the contact list shows a set of cards for each contact in the list. The user is allowed to delete a contact by clicking on the delete button. When this happens, a contact:delete event is triggered, the controller is listening for the event and uses the deleteContact() method to handle the event:

  deleteContact(view, contact) {
    let message = 'The contact will be deleted';
    App.askConfirmation(message, (isConfirm) => {
      if (isConfirm) {
        contact.destroy({
          success() {
            App.notifySuccess('Contact was deleted');
          },
          error() {
            App.notifyError('Ooops... Something went wrong');
          }
        });
      }
    });
  }

The handler is pretty easy to understand, it uses the askConfirmation() method in the infrastructure app to ask for the user confirmation. If the user confirms the deletion, the contact is destroyed. The infrastructure App provides two methods to show notifications to the user: notifySuccess() and notifyError().

The cool thing about these App methods is that the controllers do not need to know the details about the confirmation and notification mechanisms. From the controller point of view, it just works.

The method that asks for the confirmation can be a simple confirm() call, as follows:

// app.js
var App = {
  // ...
  askConfirmation(message, callback) {
    var isConfirm = confirm(message);
    callback(isConfirm);
  }
};

However, in the modern web applications, using the plain confirm() function is not the best way to ask for confirmation. Instead, we can show a Bootstrap dialog box or use an available library for that. For simplicity, we will use the nice JavaScript SweetAlert library; however, you can use whatever you want:

// app.js
var App = {
  // ...

  askConfirmation(message, callback) {
    var options = {
      title: 'Are you sure?',
      type: 'warning',
      text: message,
      showCancelButton: true,
      confirmButtonText: 'Yes, do it!',
      confirmButtonColor: '#5cb85c',
      cancelButtonText: 'No'
    };

    // Show the message
    swal(options, function(isConfirm) {
      callback(isConfirm);
    });
  }
};

Figure 1.9. Using SweetAlert for confirmation messages

We can implement the notification methods in a similar way. We will use the JavaScript noty library; however, you can use whatever you want:

// app.js
var App = {
  // ...

    notifySuccess(message) {
    new noty({
      text: message,
      layout: 'topRight',
      theme: 'relax',
      type: 'success',
      timeout: 3000 // close automatically
    });
  },

  notifyError(message) {
    new noty({
      text: message,
      layout: 'topRight',
      theme: 'relax',
      type: 'error',
      timeout: 3000 // close automatically
    });
  }
};	

Figure 1.10. Using noty to show notification messages

This is how you can implement a robust and maintainable Backbone application; please go to the GitHub repo for this book in order to see the complete code for the application. The views are not covered in the chapter as we will see them in detail in Chapter 2, Managing views.

File organization

When you work with MVC frameworks, file organization is trivial. However, Backbone is not an MVC framework, therefore, bringing your own file structure is the rule. You can organize the code on these paths:

• apps/: This directory is where modules or subapplications live. All subapplications should be on this path
• Components/: These are the common components that multiple subapplications require or use on the common layout as a breadcrumbs component
• core/: Under this path, we can put all the core functions such as utilities, helpers, adapters, and so on
• vendor/: On vendor, you can put all third-party libraries; here you can put Backbone and its dependencies.
• app.js: This is the entry point of the application that is loaded from index.html
• Subapplications can have a file structure as they are a small Backbone Application.
• models/: This defines the models and collections
• app.js: This is the application façade that is called from the router
• router.js: This is the router of the application that is instantiated by the root application at bootstrap process
• contactList.js, contactEditor.js, contactViewer.js: These are the controllers for the application

For a contacts application, the code organization can be as shown in the following:

Figure 1.11. File structure

We started by describing, in a general way, how a Backbone application works. It describes two main parts: a root application and subapplications. The root application provides a common infrastructure to other smaller and focused applications that we call subapplications.

Subapplications should be loose coupled with other subapplications and should own its resources such as views, controllers, routers, and so on. A subapplication manages a small part of the system with a well-focused business value and the communication between subapplications and infrastructure application is made through an events-driven bus with Backbone.Events.

The user interacts with the application using views that a subapplication renders. A subapplication controller orchestrates interaction between views, models, and collections and owns the business logic for the use case.

Finally, a file system organization explains the right sites to put your files and keep your project clean and organized. This organization does not follow an MVC pattern; however, it is powerful and simple. It encapsulates all the necessary code for a module in a single path (subapplication paths) instead of putting all the code across multiple paths.

In this way the structure of Backbone applications has been proven to be robust, a proof for this is that several open source applications such as TodoMVC follow (more or less) the principles exposed here. It facilitates the testability of the code due to separation of responsibilities so that each object can be tested separately.

Large Backbone applications are often built on top of Backbone Marionette as it reduces the boilerplate code; however, Marionette uses its own conventions to work. If you are fine with it using its own conventions, you will be happy to use Marionette on top of Backbone.

However, if you love the freedom of doing things your way, you may prefer plain Backbone and create your own utilities and classes.

In the next chapter, I will show you how to manage and organize views and simplify the complex layouts, identifying the common uses of the views. You will build general purpose views that will be useful for all your projects and forget about the implementation of the render() method.