Rough - addyosmani/backbone-fundamentals GitHub Wiki

I'm collecting initial information about various patterns here in the hopes of building up a solid, agreed upon definition of each. Multiple sources are used including Trygve Reenskaug and http://www.codeproject.com/KB/architecture/MVC_MVP_MVVM_design.aspx, http://heim.ifi.uio.no/~trygver/themes/mvc/mvc-index.html, http://www.bogotobogo.com/DesignPatterns/mvc_model_view_controller_pattern.php

(this is not the final version that will be included in the book. just rough work while I review what other communities have interpreted these patterns to be.)

MVC (traditional)

Model-View-Controller (MVC) is a software architecture architectural pattern thought to be designed by Trygve Reenskaug, a Norwegian computer engineer, while working on Smalltalk-80 in 1979 . It was subsequently described in depth in the highly influential “Design Patterns: Elements of Reusable Object-Oriented Software” , a.k.a. the “Gang of Four” book, in 1994. Two years later, Mike Potel from Taligent (IBM) published his paper, “Model-View-Presenter (MVP) - The Taligent Programming Model for C++ and Java” [3], where he aimed to address the shortfalls of the MVC.

In MVC, the model is the data, the view is the window on the screen, and the controller is the glue between the two taking the data and presenting that to the view .In other words, the MVC architectural pattern requires the isolation of business logic (Model) from the user interface (View), with the Controller receiving user input and coordinating the other two. This separation supports the modularization of an application's functionality and provides the following benefits:

Separation of Model from View components makes it possible to implement several user interfaces that reuse the common core business logic.

  • Duplication of low-level Model code is eliminated across multiple UI implementations.
  • Decoupling of Model and View code results in an improved ability to write unit tests for the core business logic code.
  • Modularity of components allows core logic developers and UI developers to work simultaneously without affecting the other.
  • Basically, it says that there are three distinct responsibilities for our application.

Models Models represent knowledge. A model could be a single object (rather uninteresting), or it could be some structure of objects.

There should be a one-to-one correspondence between the model and its parts on the one hand, and the represented world as perceived by the owner of the model on the other hand.

Views A view is a (visual) representation of its model. It would ordinarily highlight certain attributes of the model and suppress others. It is thus acting as a presentation filter.

A view is attached to its model (or model part) and gets the data necessary for the presentation from the model by asking questions. It may also update the model by sending appropriate messages. All these questions and messages have to be in the terminology of the model, the view will therefore have to know the semantics of the attributes of the model it represents.

Controllers A controller is the link between a user and the system. It provides the user with input by arranging for relevant views to present themselves in appropriate places on the screen. It provides means for user output by presenting the user with menus or other means of giving commands and data. The controller receives such user output, translates it into the appropriate messages and pass these messages on to one or more of the views.

Let me quote Peter Michaux:

"In a nutshell the classic MVC architecture work like this. There is a model that is at the heart of the whole thing. If the model changes, it notifies its observers that a change occurred. The view is the stuff you can see and the view observes the model. When the view is notified that the model has changed, the view changes its appearance. The user can interact with the view (e.g. clicking stuff) but the view doesn’t know what to do. So the view tells the controller what the user did and assumes the controller knows what to do. The controller appropriately changes the model. And around and around it goes."

##Discussion

"A careful reader of the “Gang of Four” book will notice that MVC is not referred to as a design pattern but a “set of classes to build a user interface” that uses design patterns such as Observer, Strategy, and Composite. It also uses Factory Method and Decorator, but the main MVC relationship is defined by the Observer and Strategy patterns.

There are three types of objects. The Model is our application data, the View is a screen, and the Controller defines the way the View reacts to user input. The views and models use the Publish-Subscribe protocol - when Model data is changed, it will update the View. It allows us to attach multiple Views to the same Model [2]. This is achieved by using the Observer design pattern.

The goal of this pattern is to define the one-to-many relationship between the Subject and the Observers; if the Subject is changed all Observers are updated. The Subject maintains the list of the Observers and can attach and detach objects to the list. The Observer in return exposes an Update method on its interface that Subject can use to update all objects it observes.

The other component of the MVC pattern is the View-Controller relationship. The View uses the Controller to implement a specific type of response. The controller can be changed to let the View respond differently to user input. This View-Controller link is an example of the Strategy design pattern. "

###More notes Model

  • Manages the app data and state
  • Not concerned with UI or presentation
  • Often persists somewhere
  • Same model should be reusable, unchanged in different interfaces

View

  • Present the Model to the user in an appropriate interface
  • Allows user to manipulate data
  • Does not store any data except to cache state
  • Easily reusable & configurable to display different data

Controller

  • Intermediary between Model & View
  • Updates the view when the model changes
  • Updates the model when the user manipulates the view
  • Typically where the app logic lives

This pattern decouples changes to how data are manipulated from how they are displayed or stored, while unifying the code in each component. In other words, it's a way of developing apps keeping the data (model) used in the program, and the visual (view) component of the program separate from one another, each interacting only with a controller containing the logic of our program. The view and the model interact only with the controller NEVER with each other.

A triad of three modules linked by the Observer Pattern. The View drives a presentation and elements within the View observe the Model. Elements within the Controller observe the View and Model, and elements within the Model observe the Controller.

In a simple application, the Controller can affect changes to the Model based on user input and also communicate those changes to the View so that the UI can be updated. In real-world applications, however, the View will normally also need to update to reflect additional changes to the underlying Model. This is necessary because changing one aspect of the Model may cause it to update other dependent Model state. This requires Model code to inform the View layer when state changes happen. The Model code, however, cannot statically bind and call the View code. This is where observers come in.

The Observer Pattern is a specific instance of the publish/subscribe paradigm. These techniques define a one-to-many dependency between objects such that a publisher object can notify all subscribed objects of any state changes without depending on them directly.

MVVM

MVC, MVP and MVVM were back on the radar in 2004 when Martin Fowler analysed them in his papers. He unveiled his solution to this problem in his paper “Presentation Model (PM)” . With the release of the Windows Presentation Foundation (WPF), the term, Model-View-ViewModel, entered the scene. First mentioned by the WPF Architect, John Gossman, on his blog in 2005 [7], it was later described by Josh Smith in the MSDN article “WPF Apps with the Model-View-ViewModel Design Pattern” [8]. MVVM was built around the WPF architecture, and encapsulates elements of MVC, MVP, and PM design patterns.

"The term MVVM was first mentioned by the WPF Architect, John Gossman, on his blog in 2005 [7]. It was then described in depths by Josh Smith in his MSDN article “WPF Apps with the Model-View-ViewModel Design Pattern” [8].

Gossman explains that the idea of MVVM was built around modern UI architecture where the View is the responsibly of the designer rather than the developer and therefore contains no code. Just like its MVC predecessor, the View in MVVM can bind to the data and display updates, but without any coding at all, just using XAML markup extensions. This way, the View is under the designer’s control, but can update its state from the domain classes using the WPF binding mechanism. This fits the description of the Presentation Model pattern.

This is why MVVM is similar to PM where the View will reference the Presentation Model, called ViewModel, which is a better name since it is a “Model of the View”. Unlike the Presentation Model, the ViewModel in MVVM also encapsulates commands just like the Presenter in MVP. "

"Model-View-ViewModel (MVVM) pattern splits the User Interface code into 3 conceptual parts - Model, View and ViewModel out of which the concept of the ViewModel is the new and the most exciting.

Model is a set of classes representing the data coming from the services or the database. View is the code corresponding to the visual representation of the data the way it is seen and interacted with by the user. ViewModel serves as the glue between the View and the Model. It wraps the data from the Model and makes it friendly for being presented and modified by the view. ViewModel also controls the View's interactions with the rest of the application (including any other Views).

"ViewModel code can (and should) refer to the Model, but the Model classes (if separate from the ViewModel) should not be aware of the ViewModel. View should be aware of the ViewModel, but ViewModel should be not aware of the View. At the diagram above, the arrows are showing which MVVM part is aware of which.

Similar to the models in the older patterns, the Model within MVVM pattern is simply data coming from the service or the database. Quite often the Model can be built to be part of the ViewModel. Because of that, in our samples, we'll skip the model and concentrate primarily on the ViewModel, the View and interactions between them."

Advantages of the MVVM Pattern

The major advantage of the MVVM pattern is that it provides the best separation of concerns: under MVVM, View is in charge of the visual representation while the non-visual ViewModel is in charge of all of the interactions with the rest of the software including the Model, the Services (usually via the Model) and the rest of the ViewModels. All the MVVM advantages derive from this feature:

  • Flexibility and Customization - different Views can be used with the same ViewModels allowing completely different visual representations of the same functionality depending on what different customers want.
  • Re-use - because of the separation of Visual and non-Visual functionality both the Views and the ViewModels have higher potential for re-use than if the Visual and non-Visual functionality were mixed, since e.g. a non-visual functionality usually would not be able to make use of a functionality that contains a Visual part to it.
  • Separation of the UI design and development - MVVM makes it possible to separate the work of the developer and the designer as clearly as possible: at first the developer can produce an application with the crudest possible GUI. Then the designer, using designer tools will be able to modify the GUI (the Views) without touching any non-visual code.
  • Testing - Writing automated tests for a Visual application is not easy. The View - ViewModel separation allows the ViewModel to be unit tested without the view. Since the ViewModel is in charge of the interactions with the rest of the application, such unit tests would cover the most important functionality while the View tests will only have to contain testing of the visual features, e.g. colors, animations, etc.

MVP

"As we have seen, the Model-View relationship is an indirect one based on the Observer design pattern. The Model can notify the View that new data has arrived, and the View can update its data from the Model it is subscribed to. Bower and McGlashan questioned the nature of this indirect link, and suggested that the Model can gain access to the user interface directly.

The basic idea was “twisting MVC” in a way where the View absorbs the Controller functionality and the new class (the Presenter) is added. The Presenter can access the View and the Model directly, and the Model-View relationship can still exist where relevant. Overall, the View displays data and the Presenter can update the model and the view directly."

"If we add Interactor, Selection, and Command classes, we will get the full picture. There are different flavours of MVP, as we have seen the Presenter can access the View directly, or the Model-View relationship based on the Observer pattern can still exist. In this version, we leave out the Model-View link that we have seen in our MVC code example, and assume that the Presenter can update the View directly."

"The most fundamental idea of MVC is a separation of the presentation objects and the domain model objects. Fowler calls this a Separated Presentation pattern. Separated Presentation gave us the Model and the View. The particular flavour of the Observer pattern they use to communicate, he calls Observer Synchronization, which is different from Flow Synchronization [5]. Therefore, according to Fowler, the Model and the View, and their indirect link, are the core elements that define MVC. The elements that he calls - Separated Presentation and Observer Synchronization.

He also points us towards two quite different descriptions of MVP – one by Potel, and the other one by Bower and McGlashan. Potel was only concerned with removal of the Controller and delegating more work to the View. Fowler calls this approach - Supervising Controller. The other feature described by Bower and McGlashan, the ability of the Presenter to update the View directly, he calls - Passive View.

By breaking down these patterns into smaller pieces, Fowler gives developers the tools they can use when they aim to implement a specific behaviour, not the MVC or MVP patterns themselves. The developers can choose from Separated Presentation, Observer Synchronization, Supervising Controller, or Passive View.

As powerful as they are, both MVC and MVP have their problems. One of them is persistence of the View’s state. For instance, if the Model, being a domain object, does not know anything about the UI, and the View does not implement any business logic, then where would we store the state of the View’s elements such as selected items? Fowler comes up with a solution in the form of a Presentation Model pattern. He acknowledges the roots of the pattern in the Application Model – the fruit of labour of the Visual Works Smalltalk team. Application Model introduces another class between the Model and the View that can store state. For the View, the ApplicationModel class becomes its Model, and the domain specific Model interacts with the ApplicationModel which becomes its View. The ApplicationModel knows how to update the UI, but does not reference any UI elements directly.

Just like in Smalltalk’s Application Model, the Presentation Model class sits between the Model and the View. It enriches the Model’s data with information such as state that gets synchronised with the View using Publish – Subscribe or the data binding mechanism. The View raises an event that updates the state in the Presentation Model and updates its state in return.

This model allows you to implement the synchronisation code in the View or the Presentation Model. Therefore, it is a developer’s decision whether the View should reference the Presentation Model, or the Presentation Model should reference the View "

Key Benefits

Before using any pattern a developers needs to consider the pros and cons of using it. There are a number of key benefits to using either the MVC or MVP pattern (See list below). But, there also a few draw backs to consider. The biggest drawbacks are additional complexity and learning curve. While the patterns may not be appropriate for simple solutions; advance solutions can greatly benefit from using the pattern. I’m my experience a have seen a few solutions eliminate a large amount of complexity but being re-factored to use either pattern.

  • Loose coupling – The presenter/controller are an intermediary between the UI code and the model. This allows the view and the model to evolve independently of each other.
  • Clear separation of concerns/responsibility
  • UI (Form or Page) – Responsible for rending UI elements
  • Presenter/controller – Responsible for reacting to UI events and interacts with the model
  • Model – Responsible for business behaviors and state management
  • Test Driven – By isolating each major component (UI, Presenter/controller, and model) it is easier to write unit tests. This is especially true when using the MVP pattern which only interacts with the view using an interface.
  • Code Reuse – By using a separation of concerns/responsible design approach you will increase code reuse. This is especially true when using a full blown domain model and keeping all the business/state management logic where it belongs.
  • Hide Data Access – Using these patterns forces you to put the data access code where it belongs in a data access layer. There a number of other patterns that typical works with the MVP/MVC pattern for data access. Two of the most common ones are repository and unit of work. (See Martin Fowler – Patterns of Enterprise Application Architecture for more details)
  • Flexibility/Adaptable – By isolating most of your code into the presenter/controller and model components your code base is more adaptable to change. For example consider how much UI and data access technologies have changed over the years and the number of choices we have available today. A properly design solution using MVC or MVP can support multi UI and data access technologies at the same time.

Key Differences

So what really are the differences between the MVC and MVP pattern?. Actually there are not a whole lot of differences between them. Both patterns focus on separating responsibility across multi components and promote loosely coupling the UI (View) from the business layer (Model). The major differences are how the pattern is implemented and in some advanced scenarios you need both presenters and controllers.

Here are the key differences between the patterns:

MVP Pattern

  • View is more loosely coupled to the model. The presenter is responsible for binding the model to the view.
  • Easier to unit test because interaction with the view is through an interface
  • Usually view to presenter map one to one. Complex views may have multi presenters.

MVC Pattern

  • Controller are based on behaviors and can be shared across views
  • Can be responsible for determining which view to display (Front Controller Pattern)

Summarization

Summary

As we have seen, the fundamental idea of MVC is a separation of the domain logic and the GUI objects into the Model and the View. These two are linked indirectly by using the Publish-Subscribe mechanism known as the Observer pattern. Another element of this design is a Controller that implements a particular strategy for the View.

It makes sense to use this pattern when the View is very simple and contains no code, as in the case of web based systems and HTML. This way, the View can build the UI, and the Controller handles the user interactions. Note that Microsoft chose to use this pattern in their ASP.NET MVC framework that targets the web developers.

MVP delegates more work to the View and removes the Controller. It introduces the Presenter class that encapsulates the View’s state and commands. The Presenter can access the View directly. This is perfect for Windows based systems where Views are powerful classes that can encapsulate the events and store items’ state. Note that Microsoft made this a part of the Composite Application Blocks (CAB) framework that targets Windows developers.

The Model – View relationship based on the Observer pattern still exists in MVP; however, the Presenter can access the View directly. Although this is easy to use and implement, developers need to be careful not to break the logic of the system they are trying to model. For instance, the system where the driver accelerates and checks the speed indicator is unlikely to be modeled using MVP. The Driver (Presenter) can update the engine’s state (Model), but now needs to update the speed indicator (View). Surely, more logical is the system where the driver updates the engine’s state (press gas) and the speed indicator will read these changes (the Observer pattern).

The MVVM is a powerful pattern for all WPF developers. It allows to keep the View free from any code and available to XAML designers. The WPF binding serves as a link to the ViewModel that can handle state, implement the commands, and communicate with the domain specific Model.

MVVM resembles Fowler’s PM where the Presentation Model class is called ViewModel.

The Presentation Model is a more general version of MVVM where the developers need to implement the link between the View and the ApplicationModel (ViewModel) themselves. It also allows the View to reference the ApplicationModel, or the ApplicationModel to reference the view. It can be broken down into smaller patterns like Supervising Controller or Passive View where only a specific behaviour is required.