By Dino Esposito

For a few moments when I first approached ASP.NET MVC I really thought it was the negation of AJAX. With a programming model heavily based on REST principles, I thought all that one could do is invoke a URL. And when you get, or post to, a URL, you inevitably involve the browser and get a full-page refresh. As I ve explored it more, and as the facts have largely shown, my first impression was dead wrong. AJAX is definitely possible in ASP.NET MVC.

The role of JavaScript is clearly relevant and fundamental. The ASP.NET MVC Framework simply provides its own set of JavaScript files and utilities to make AJAX calls occur in much the same way they occur in WebForms applications. In this article I ll go through the various aspects of the AJAX API you can leverage in ASP.NET MVC applications.

AJAX in ASP.NET MVC? Why Not!

Compared to WebForms, ASP.NET MVC provides a smarter layer of code on top of the same ASP.NET runtime.

In the WebForms model, the request that comes in is dispatched to an HTTP handler that maps the URL to a server file. It reads the content of the server file, parses that to a C# class, compiles the class into an assembly, and invokes a well-known method on the class. The well-known method is ProcessRequest, one of the members of the IHttpHandler interface that represents the public contract for any ASP.NET page. The C# class that defines the expected behavior for the page derives from a system class (the System.Web.UI.Page class). This class is nothing more than a built-in HTTP handler albeit one of the most complex HTTP handlers ever written.

The ASP.NET MVC Framework abstracts some of the steps in the procedure just described. For example, the URL is not necessarily bound to a server file. The URL is simply the representation of the requested resource. This means that a module intercepts the request, parses the URL, and forwards the request to a controller component. The role that the page class plays in WebForms is split between controller and view in ASP.NET MVC. I dare say that ASP.NET MVC smartly refactors ASP.NET WebForms, but works on top of the same runtime environment and, therefore, according to the same set of global rules. If AJAX is possible in WebForms, it has to be possible in ASP.NET MVC, as well.

JavaScript Makes AJAX Run

At its core, AJAX is about making an out-of-band request to the web server via XMLHttpRequest. A piece of JavaScript code prepares the call, runs it asynchronously, then processes the response in a callback function. The callback function is responsible for updating the user interface with downloaded data using the DOM services. To make developers lives a bit easier, Microsoft provided in ASP.NET AJAX such facilities as the partial rendering API and JavaScript proxy for Web services.

To be picky, I could say that Microsoft didn t provide a very simple API to make AJAX calls that hide the nitty-gritty details of the XMLHttpRequest object. JavaScript proxies are great, but they require a Web or WCF service at the other end. The Microsoft AJAX client library does offer a WebRequest object, but it takes a few lines of code to prepare it and make it work. Other libraries, primarily the jQuery library now included in ASP.NET AJAX, make up for this with their own API.

The package that contains the ASP.NET MVC install also includes the jQuery library, so you can use this library (or any other similar libraries) to prepare your direct AJAX calls directed at URLs of choice. This is the most natural way of having AJAX in ASP.NET MVC applications. However, Microsoft provides facilities to make AJAX happen in a way that is similar to partial rendering in WebForms.

That said, I feel the need to clarify a key point once and for all. There s only one way of placing AJAX calls, and it s all about using the XMLHttpRequest object. On top of this, some frameworks and JavaScript libraries have built their own infrastructure, with the sole purpose of making programming easier and faster.

Direct Scripting

I like to use the expression direct scripting to refer to the scenario where you have some JavaScript code that places calls to an HTTP endpoint and receives a semantic response, such as JSON data or primitive data. Direct scripting is an approach to AJAX that works in any case, as it is at the lowest possible level of abstraction. All you need in the page is a button or in general some event handler, no matter how defined. Figure 1 shows some JavaScript code that uses jQuery to attach an onclick handler to a button.

<input type="button" id="Button1" value=" Get customers"></input>

<select id="ddCustomerList" name="ddCustomerList" size="8">

</select>

<script type="text/javascript" src="/Scripts/jquery-1.2.6.min.js"></script>

<script type="text/javascript">

$(document).ready(function() {

$('#Button1').click(function() {

$.getJSON("/Home/GetCustomers",

null,

function(data) { addCustomers(data); });

});

});

function addCustomers(data)

{

// Get the DOM reference to the drop-down list. Note that

// $("#id") returns a wrapped set; to get the DOM reference

// you need to specify an index.

var list = $("#ddCustomerList")[0];

// Loop over the returned collection

for (var i = 0; i < data.length; i++) {

var cust = data[i];

var option = new Option(cust.CompanyName, cust.CustomerID);

list.add(option);

};

};

</script>

Figure 1: An example of direct scripting using jQuery

Among other things, the sample page contains a button and a dropdown list. The button is given its onclick handler as the document is fully loaded. The handler uses the $.getJSON function from the jQuery library to download a response. Once downloaded, the response is processed by the callback associated with the $.getJSON call (the addCustomers function in Figure 1). This function assumes to receive a collection of data and simply loops through it. For each encountered object, it creates an optional DOM object and adds it to the dropdown list.

How is this approach specific to ASP.NET MVC? Actually, it isn t. Looking at the example, all you need is an HTTP endpoint that returns JSON data. In ASP.NET MVC, though, creating a controller action that returns JSON couldn t be easier. In Figure 1, the $.getJSON function points to the /Home/GetCustomers URL. Such a URL is mapped to the GetCustomers method on the Home controller, as shown here:

public JsonResult GetCustomers()

{

NWindDataContext context = new NWindDataContext();

context.DeferredLoadingEnabled = false;

List<Customer> list = (from c in context.Customers

select c).ToList<Customer>();

return this.Json(list);

}

A controller action that you want to invoke via AJAX doesn t need to be connected to a view object. More simply, it can return a plain JSON string. In this case, you define the method to return a JsonResult object and create it via the Json helper method on the controller base class.

The Json method takes any .NET object and serializes it to JSON using a given content type and encoding. The Json method has a few overloads that end up calling the one shown here:

protected virtual JsonResult Json(object data,

string contentType, Encoding contentEncoding)

Internally, the Json method invokes the JavaScriptSerializer class from the .NET Framework to serialize a .NET object to JSON.

It is worth noting that the object you attempt to serialize to JSON must be serializable and should not include circular references. This might be a common source of trouble if you use a LINQ to SQL object model to represent your data. In the preceding code snippet, a list of Customer objects should be returned to the client and the list of Customer objects is obtained via a LINQ to SQL query. I ve placed Customer and Order entities from the Northwind database in the sample model. As you may recall, there s a relationship set between the entities so that the Customer class has an entity set named Orders and the Order class points back to a Customer object. This sets up a circular reference and causes the Json method to fail.

There are two possible workarounds. One is using a data transfer object (DTO) instead of the real entity class. The other is disabling the lazy loading feature so that no relationships are further expanded and no circular reference is generated. The latter option is easier to code; the former is more general. Using a DTO simply means copying the content in Customer to another custom class that doesn t have dependencies on Orders. In this case, you also can optimize the amount of data being transferred. Unless you need on the client all the properties of a customer object, a DTO approach allows you to transfer only what s strictly needed.

The ActionLink AJAX Helper

If you don t like writing that much JavaScript code yourself, you can opt for the ActionLink AJAX helper method. The Ajax.ActionLink helper method generates a hyperlink in the page that points to a controller action and runs a callback function when data has been received.

No jQuery is required, as all the machinery is available in the MicrosoftMvcAjax.js file that must be included in the page along with the MicrosoftAjax.js file from the Microsoft AJAX client library:

<script src="<%= Url.Content("~/Scripts/

MicrosoftAjax.js") %>"

type="text/javascript"></script>

<script src="<%= Url.Content("~/Scripts/

MicrosoftMvcAjax.js") %>"

type="text/javascript"></script>

You use the Ajax.ActionLink method within code blocks, as shown here:

<%= Ajax.ActionLink("Get customers",

"/GetCustomers", new AjaxOptions {

OnSuccess="addCustomersEx" })%>

The code generates a hyperlink that points to the same GetCustomers method we considered earlier in the direct scripting example. The first argument you pass to the action link is the text of the hyperlink. The second argument is the controller action to invoke. Finally, the third argument is a collection of optional settings to use in the call. At the very minimum, you must specify the JavaScript callback that runs upon a successful completion of the call. The ActionLink method has several more overloads; the one shown here is the most common and simple of all.

The controller action can return anything, including JSON, JavaScript, or plain HTML markup. The callback, if specified, will handle the response and update the user interface accordingly. Here s the source code of the addCustomersEx callback function:

function addCustomersEx(context)

{

// Grab the method s response

var response = eval(context.get_data());

// Invoke the addCustomers function of Figure 1

// to populate the dropdown list

addCustomers(response);

};

The success callback, as well as any other callbacks you can specify in the AjaxOptions object, receives only one argument of type AjaxContext. Figure 2 shows the members of the object. To get the response as plain data, call the get_data method and pass it through the JavaScript s eval function to transform a JSON string into a usable JavaScript object.

Member	Description
get_data	Gets any data returned from the controller action.
get_insertionMode	Indicates how to treat the response (only if markup), whether to replace, prepend, or append it to the markup of the specified DOM element. The default is replace.
get_loadingElement	Indicates the DOM element to be displayed to indicate that an AJAX call is going on.
get_request	Gets the Sys.Net.WebRequest object that represents the current request.
get_response	Gets the Sys.Net.WebRequestExecutor object for the current request.
get_updateTarget	Indicates the DOM element to be automatically updated with the returned markup, if any.

Figure 2: The AjaxContext object

The link emitted in the page takes the following form:

<a href="/Home/GetCustomers"

onclick="Sys.Mvc.AsyncHyperlink.handleClick(

this, new Sys.UI.DomEvent(event), { insertionMode:

Sys.Mvc.InsertionMode.replace, onSuccess:

Function.createDelegate(this, addCustomersEx) }

);">

Get customers

</a>

As you can see, AJAX in ASP.NET MVC is definitely possible, but all of it happens through JavaScript.

Partial Rendering in ASP.NET MVC

The AJAX ActionLink method also can be used to implement a sort of partial rendering. If the controller action returns HTML markup, then this content can automatically be inserted in to the inner space of the specified DOM element. To get this, simply specify the element to update in the AjaxOptions settings and, optionally, the desired insertion mode:

<%= Ajax.ActionLink("Details",

"/GetCustomerDetails", new AjaxOptions {

LoadingElementId="lblWait",

UpdateTargetId="pnlDetails" })%>

The lblWait DOM element is displayed for the time it takes to download the response, then is hidden. The pnlDetails DOM element, instead, is updated with the content downloaded. Of course, this approach works well if the downloaded content is an HTML string. Here s a sample controller method:

public string GetCustomerDetails(string id)

{

// Return HTML

:

}

I recommend that any element you use as the loading element be initially hidden from view using CSS.

Conclusion

In this article I ve demonstrated the simplest way to get AJAX in an ASP.NET MVC solution. In particular, I focused on direct scripting and the AJAX ActionLink helper method. The ActionLink approach, though, also lends itself very well to implement a sort of partial rendering in ASP.NET MVC. Because of space constraints I could only scratch the surface of this topic. Next month I ll provide full coverage of partial rendering in ASP.NET MVC, including posting an entire form to a controller. Stay tuned!

Dino Esposito (dino@idesign.net) is an architect at IDesign, specializing in ASP.NET, AJAX, and RIA solutions. Dino is the author of Microsoft .NET: Architecting Applications for the Enterprise and Microsoft ASP.NET and AJAX: Architecting Web Applications.