A Quick Introduction to ASP.NET MVC

ASP.NET MVC is a framework for building web applications that applies the general Model View Controller pattern to the ASP.NET framework. Let's break that down by first looking at how ASP.NET MVC and the ASP.NET framework are related.

How ASP.NET MVC Fits in with ASP.NET

When ASP.NET 1.0 was first released in 2002, it was easy to think of ASP.NET and Web Forms as one and the same thing. ASP.NET has always supported two layers of abstraction, though:

• System.Web.UI: The Web Forms layer, comprising server controls, ViewState, and so on
• System.Web: The plumbing, which supplies the basic web stack, including modules, handlers, the HTTP stack, and so on

The mainstream method of developing with ASP.NET included the whole Web Forms stack—taking advantage of drag-and-drop controls, semi-magical statefulness, and wonderful server controls while dealing with the complications behind the scenes (an often confusing page life cycle, less than optimal HTML, and so on).

However, there was always the possibility of getting below all that—responding directly to HTTP requests, building out web frameworks just the way you wanted them to work, crafting beautiful HTML—using Handlers, Modules, and other handwritten code. You could do it, but it was painful; there just wasn't a built-in pattern that supported any of those things. It wasn't for lack of patterns in the broader computer science world, though. By the time ASP.NET MVC was announced in 2007, the MVC pattern was becoming one of the most popular ways of building web frameworks.

The MVC Pattern

Model-View-Controller (MVC) has been an important architectural pattern in computer science for many years. Originally named Thing-Model-View-Editor in 1979, it was later simplified to Model-View-Controller. It is a powerful and elegant means of separating concerns within an application (for example, separating data access logic from display logic) and applies itself extremely well to web applications. Its explicit separation of concerns does add a small amount of extra complexity to an application's design, but the extraordinary benefits outweigh the extra effort. It has been used in dozens of frameworks since its introduction. You'll find MVC in Java and C++, on Mac and on Windows, and inside literally dozens of frameworks.

The MVC separates the user interface of an application into three main aspects:

• The Model: A set of classes that describes the data you're working with as well as the business rules for how the data can be changed and manipulated
• The View: Defines how the application's user interface (UI) will be displayed
• The Controller: A set of classes that handles communication from the user, overall application flow, and application-specific logic

MVC as Applied to Web Frameworks

The MVC pattern is used frequently in web programming. With ASP.NET MVC, it's translated roughly as:

• Models: These are the classes that represent the domain you are interested in. These domain objects often encapsulate data stored in a database as well as code used to manipulate the data and enforce domain-specific business logic. With ASP.NET MVC, this is most likely a Data Access Layer of some kind using a tool like Entity Framework or NHibernate combined with custom code containing domain-specific logic.
• View: This is a template to dynamically generate HTML . We cover more on that in Chapter 3 when we dig into views.
• Controller: This is a special class that manages the relationship between the View and Model. It responds to user input, talks to the Model, and it decides which view to render (if any). In ASP.NET MVC, this class is conventionally denoted by the suffix Controller.

It's important to keep in mind that MVC is a high-level architectural pattern, and its application varies depending on use. ASP.NET MVC is contextualized both to the problem domain (a stateless web environment) and the host system (ASP.NET).

Occasionally I talk to developers who have used the MVC pattern in very different environments, and they get confused, frustrated, or both (confustrated?) because they assume that ASP.NET MVC works the exact same way it worked in their mainframe account processing system fifteen years ago. It doesn't, and that's a good thing—ASP.NET MVC is focused on providing a great web development framework using the MVC pattern and running on the .NET platform, and that contextualization is part of what makes it great.

ASP.NET MVC relies on many of the same core strategies that the other MVC platforms use, plus it offers the benefits of compiled and managed code and exploits newer .NET language features such as lambdas and dynamic and anonymous types. At its heart, though, ASP.NET applies the fundamental tenets found in most MVC-based web frameworks:

• Convention over configuration
• Don't repeat yourself (aka the DRY principle)
• Pluggability wherever possible
• Try to be helpful, but if necessary, get out of the developer's way

The Road to MVC 3

Two short years have seen three major releases of ASP.NET MVC and several more interim releases. In order to understand ASP.NET MVC 3, it's important to understand how we got here. This section describes the contents and background of each of the three major ASP.NET MVC releases.

ASP.NET MVC 1 Overview

In February 2007, Scott Guthrie (“ScottGu”) of Microsoft sketched out the core of ASP.NET MVC while flying on a plane to a conference on the East Coast of the United States. It was a simple application, containing a few hundred lines of code, but the promise and potential it offered for parts of the Microsoft web developer audience was huge.

As the legend goes, at the Austin ALT.NET conference in October 2007 in Redmond, Washington, ScottGu showed a group of developers “this cool thing I wrote on a plane” and asked if they saw the need and what they thought of it. It was a hit. In fact, many people were involved with the original prototype, codenamed Scalene. Eilon Lipton e-mailed the first prototype to the team in September 2007, and he and ScottGu bounced prototypes, code, and ideas back and forth.

Even before the official release, it was clear that ASP.NET MVC wasn't your standard Microsoft product. The development cycle was highly interactive: there were nine preview releases before the official release, unit tests were made available, and the code shipped under an open source license. All of these highlighted a philosophy that placed a high value in community interaction throughout the development process. The end result was that the official MVC 1.0 release—including code and unit tests—had already been used and reviewed by the developers who would be using it. ASP.NET MVC 1.0 was released on 13 March 2009.

ASP.NET MVC 2 Overview

ASP.NET MVC 2 was released just one year later, in March 2010. Some of the main features in MVC 2 included:

• UI helpers with automatic scaffolding with customizable templates
• Attribute-based Model validation on both client and server
• Strongly-typed HTML helpers
• Improved Visual Studio tooling

There were also lots of API enhancements and “pro” features, based on feedback from developers building a variety of applications on ASP.NET MVC 1, such as:

• Support for partitioning large applications into areas
• Asynchronous Controllers support
• Support for rendering subsections of a page/site using Html.RenderAction
• Lots of new helper functions, utilities, and API enhancements

One important precedent set by the MVC 2 release was that there were very few breaking changes. I think this is a testament to the architectural design of ASP.NET MVC, which allows for a lot of extensibility without requiring core changes.

ASP.NET MVC 3 Overview

ASP.NET MVC 3 (generally abbreviated as MVC 3 from now on) shipped just 10 months after MVC 2, driven by the release date for Web Matrix. If MVC 3 came in a box, it might say something like this on the front:

• Expressive Views including the new Razor View Engine!
• .NET 4 Data Annotation Support!
• Streamlined validation with improved Model validation!
• Powerful hooks with Dependency Resolution and Global Action Filters!
• Rich JavaScript support with unobtrusive JavaScript, jQuery Validation, and JSON binding!
• Now with NuGet!!!!

For those who have used previous versions of MVC, we'll start with a quick look at some of these major features.

If you're new to ASP.NET MVC, don't be concerned if some of these features don't make a lot of sense right now; we'll be covering them in a lot more detail throughout the book.

Razor View Engine

Razor is the first major update to rendering HTML since ASP.NET 1.0 shipped almost a decade ago. The default view engine used in MVC 1 and 2 was commonly called the Web Forms View Engine, because it uses the same ASPX/ASCX/MASTER files and syntax used in Web Forms. It works, but it was designed to support editing controls in a graphical editor, and that legacy shows. An example of this syntax in a Web Forms page is shown here:

<%@ Page Language="C#" MasterPageFile="∼/Views/Shared/Site.Master"
Inherits="System.Web.Mvc.ViewPage<MvcMusicStore.ViewModels.StoreBrowseViewModel>"
 %>

<asp:Content ID="Content1" ContentPlaceHolderID="TitleContent" runat="server">
    Browse Albums
</asp:Content>

<asp:Content ID="Content2" ContentPlaceHolderID="MainContent" runat="server">

    <div class="genre">
        <h3><em><%: Model.Genre.Name %></em> Albums</h3>

        <ul id="album-list">
            <% foreach (var album in Model.Albums) { %>

            <li>
                <a href="<%: Url.Action("Details", new { id = album.AlbumId }) %>">
                    <img alt="<%: album.Title %>" src="<%: album.AlbumArtUrl %>" />
                    <span><%: album.Title %></span>
                </a>
            </li>

            <% } %>
        </ul>

    </div>

</asp:Content>

Razor was designed specifically as a view engine syntax. It has one main focus: code-focused templating for HTML generation. Here's how that same markup would be generated using Razor:

@model MvcMusicStore.Models.Genre

@{ViewBag.Title = "Browse Albums";}

<div class="genre">
    <h3><em>@Model.Name</em> Albums</h3>

    <ul id="album-list">
        @foreach (var album in Model.Albums)
        {
            <li>
                <a href="@Url.Action("Details", new { id = album.AlbumId })">
                    <img alt="@album.Title" src="@album.AlbumArtUrl" />
                    <span>@album.Title</span>
                </a>
            </li>
        }
    </ul>
</div>

The Razor syntax is easier to type, and easier to read. Razor doesn't have the XML-like heavy syntax of the Web Forms view engine.

We've talked about how working with the Razor syntax feels different. To put this in more quantifiable terms, let's look at the team's design goals in creating the Razor syntax:

• Compact, expressive, and fluid: Razor's (ahem) sharp focus on templating for HTML generation yields a very minimalist syntax. This isn't just about minimizing keystrokes—although that's an obvious result—it's about how easy it is to express your intent. A key example is the simplicity in transitions between markup and code. You can see this in action when writing out some model properties in a loop:

@foreach (var album in Model.Albums)
{
    <li>
        <a href="@Url.Action("Details", new { id = album.AlbumId })">
            <img alt="@album.Title" src="@album.AlbumArtUrl" />
            <span>@album.Title</span>
        </a>
    </li>
}

You only needed to signify the end of a code block for the loop—in the cases where model properties were being emitted, only the @ character was needed to signify the transition from markup to code, and the Razor engine automatically detected the transition back to markup.

Razor also simplifies markup with an improvement on the Master Pages concept—called Layouts—that is both more flexible and requires less code.

• Not a new language: Razor is a syntax that lets you use your existing .NET coding skills in a template in a very intuitive way. Scott Hanselman summarized this pretty well when describing his experiences learning Razor:

I kept […] going cross-eyed when I was trying to figure out what the syntax rules were for Razor until someone said stop thinking about it, just type an “at” sign and start writing code and I realize that there really is no Razor.

—HANSELMINUTES #249: ON WEBMATRIX WITH ROB CONERY http://hanselminutes.com/default.aspx?showid=268

• Easy to learn: Precisely because Razor is not a new language, it's easy to learn. You know HTML, you know .NET; just type HTML and hit the @ sign whenever you need to write some .NET code.
• Works with any text editor: Because Razor is so lightweight and HTML-focused, you're free to use the editor of your choice. Visual Studio's syntax highlighting and IntelliSense features are nice, but it's simple enough that you can edit it in any text editor.
• Great IntelliSense: Though Razor was designed so that you shouldn't need IntelliSense to work with it, IntelliSense can come in handy for things like viewing the properties your model object supports. For those cases, Razor does offer nice IntelliSense within Visual Studio, as shown in Figure 1.1.
• Unit testable: The Razor view engine's core compilation engine has no dependencies on System.Web or ASP.NET whatsoever—it can be executed from unit tests, or even from the command line. Though there isn't direct tooling support for this yet, it's possible to use systems like David Ebbo's Visual Studio Single File Generator (http://visualstudiogallery.msdn.microsoft.com/1f6ec6ff-e89b-4c47-8e79-d2d68df894ec/) to compile your views into classes that you can then load and test like any other object.

This is just a quick highlight of some of the reasons that Razor makes writing View code really easy and, dare I say, fun. We'll talk about Razor in a lot more depth in Chapter 3.

Figure 1.1

Validation Improvements

Validation is an important part of building web applications, but it's never fun. I've always wanted to spend as little time as possible writing validation code, as long as I was confident that it worked correctly.

MVC 2's attribute-driven validation system removed a lot of the pain from this process by replacing repetitive imperative code with declarative code. However, support was focused on a short list of top validation scenarios. There were plenty of cases where you'd get outside of the “happy path” and have to write a fair amount more code. MVC 3 extends the validation support to cover most scenarios you're likely to encounter. For more information on validation in ASP.NET MVC, see chapter 6.

.NET 4 Data Annotation Support

MVC 2 was compiled against .NET 3.5 and thus didn't support any of the .NET 4 Data Annotations enhancements. MVC 3 picks up some new, very useful validation features available due to .NET 4 support. Some examples include:

• MVC 2's DisplayName attribute wasn't localizable, whereas the .NET 4 standard System.ComponentModel.DataAnnotations Display attribute is.
• ValidationAttribute was enhanced in .NET 4 to better work with the validation context for the entire model, greatly simplifying cases like validators that compare or otherwise reference two model properties.

Streamlined Validation with Improved Model Validation

MVC 3's support for the .NET 4 IValidatableObject interface deserves individual recognition. You can extend your model validation in just about any conceivable way by implementing this interface on your model class and implementing the Validate method, as shown in the following code:

public class VerifiedMessage : IValidatableObject {
    public string Message { get; set; }
    public string AgentKey { get; set; }
    public string Hash { get; set; }

    public IEnumerable<ValidationResult> Validate(
        ValidationContext validationContext) {
        if (SecurityService.ComputeHash(Message, AgentKey) != Hash)
            yield return new ValidationResult("Agent compromised");
    }
}

Rich JavaScript Support

JavaScript is an important part of any modern web application. ASP.NET MVC 3 adds some significant support for client-side development, following current standards for top quality JavaScript integration. For more information on the new JavaScript related features in ASP.NET MVC 3, see Chapter 8.

Unobtrusive JavaScript

Unobtrusive JavaScript is a general term that conveys a general philosophy, similar to the term REST (for Representational State Transfer). The high-level description is that unobtrusive JavaScript doesn't affect your page markup. For example, rather than hooking in via event attributes like onclick and onsubmit, the unobtrusive JavaScript attaches to elements by their ID or class.

Unobtrusive JavaScript makes a lot of sense when you consider that your HTML document is just that—a document. It's got semantic meaning, and all of it—the tag structure, element attributes, and so on—should have a precise meaning. Strewing JavaScript gunk across the page to facilitate interaction (I'm looking at you, doPostBack!) harms the content of the document.

MVC 3 supports unobtrusive JavaScript in two ways:

• Ajax helpers (such as Ajax.ActionLink and Ajax.BeginForm) render clean markup for the FORM tag, wiring up behavior leveraging extensible attributes (data- attributes) and jQuery.
• Ajax validation no longer emits the validation rules as a (sometimes large) block of JSON data, instead writing out the validation rules using data- attributes. While technically I considered MVC 2's validation system to be rather unobtrusive, the MVC 3 system is that much more—the markup is lighter weight, and the use of data- attributes makes it easier to leverage and reuse the validation information using jQuery or other JavaScript libraries.

jQuery Validation

MVC 2 shipped with jQuery, but used Microsoft Ajax for validation. MVC 3 completed the transition to using jQuery for Ajax support by converting the validation support to run on the popular jQuery Validation plugin. The combination of Unobtrusive JavaScript support (discussed previously) and jQuery validation using the standard plugin system means that the validation is both extremely flexible and can benefit from the huge jQuery community.

Client-side validation is now turned on by default for new MVC 3 projects, and can be enabled site-wide with a web.config setting or by code in global.asax for upgraded projects.

JSON Binding

MVC 3 includes JSON (JavaScript Object Notation) binding support via the new JsonValueProviderFactory, enabling your action methods to accept and model-bind data in JSON format. This is especially useful in advanced Ajax scenarios like client templates and data binding that need to post data back to the server.

Advanced Features

So far, we've looked at how MVC 3 makes a lot of simple-but-mind-numbing tasks like view templates and validation simpler. MVC 3 has also made some big improvements in simplifying more sophisticated application-level tasks with support for dependency resolution and global action filters.

Dependency Resolution

ASP.NET MVC 3 introduces a new concept called a dependency resolver, which greatly simplifies the use of dependency injection in your applications. This makes it easier to decouple application components, which makes them more configurable and easier to test.

Support has been added for the following scenarios:

• Controllers (registering and injecting controller factories, injecting controllers)
• Views (registering and injecting view engines, injecting dependencies into view pages)
• Action filters (locating and injecting filters)
• Model binders (registering and injecting)
• Model validation providers (registering and injecting)
• Model metadata providers (registering and injecting)
• Value providers (registering and injecting)

This is a big enough topic that we've devoted an entire new chapter (Chapter 11) to it.

Global Action Filters

MVC 2 action filters gave you hooks to execute code before or after an action method ran. They were implemented as custom attributes that could be applied to controller actions or to an entire controller. MVC 2 included some filters in the box, like the Authorize attribute.

MVC 3 extends this with global action filters, which apply to all action methods in your application. This is especially useful for application infrastructure concerns like error handling and logging.

MVC 3 Feature Summary: Easier at All Levels

They're great features, but if I was designing the box, I'd just put this on it:

• If you've been putting off learning ASP.NET MVC, it's just become so easy there's no excuse to delay anymore.
• If you've been using ASP.NET MVC for a while, MVC 3 makes your most difficult code unnecessary.

This is a quick introductory summary, and we'll be covering these and other MVC 3 features throughout the book. If you'd like an online summary of what's new in MVC 3 (perhaps to convince your boss that you should move all your projects to MVC 3 as soon as possible), see the list at http://asp.net/mvc/mvc3#overview.