WCF Data Services - OData - Initial errors developers encounters

When authoring WCF Data Services - OData, lot of developers tries to view it in the browser during several stages of developement. Mostly they will come across the following error at initial stages.

The server encountered an error processing the request. See  server logs for more details.

When we think of creating WCF Data services, its better to have the following minimal understanding.

WCF Data services / OData service : A .Net framework component which is used to create services that uses Open Data Protocol (OData). This service exposes data through the endpoint which allows to access the data.

OData client libraries : Though not strictly required, several client libraries available for easy and effective access of OData format. Based on the client applications we create, we can make use of the respective client libraries. For example we have separate libraries for .Net, java, javascript, Silverlight, etc.

OData data model : Makes use of the exiting .Net member called Entity Data Model(EDM).  The data to be represented need to be modeled using EDM.

OData Protocol : This protocol is REST based. Defines  how the data model will be represented over the wire - either in XML based format defined by ATOM or in JSON format.

Now, let us look at the cause for the issue.

In the InitializeService method of the svc.cs file, do ensure whether the SetEntitySetAccessRule is specified for the respective entity and most important, the name of the entity set should be pluralized (if opted while creating the entity data model).

Snippet 1

// This method is called only once to initialize service-wide policies.
public static void InitializeService(DataServiceConfiguration config)
{
    // TODO: set rules to indicate which entity sets and service operations are visible, updatable, etc.
    // Examples:
    config.SetEntitySetAccessRule("Employees", EntitySetRights.AllRead);
    config.DataServiceBehavior.MaxProtocolVersion = DataServiceProtocolVersion.V2;
}

References

http://msdn.microsoft.com/en-us/data/hh237663.aspx (By David Chappell)

http://msdn.microsoft.com/en-us/library/ff478141.aspx

OData : Cross domain OData request using jQuery, JSONP

I've been doing a POC where I had a situation to make a request to a OData Service (service which exposed the data as OData  - Open Data Protocol) using jQuery. During my initial attempt, I got an error from the web browser which is due to the same origin policy of the web browser.  Usually Client side (request from browser using javascript) cannot be allowed to make a request to a resource that exists in another domain of which HTML <script> element is an exception. In such cases JSONP (JSON with padding) and CORS (Cross Origin Resource Sharing) is made use. Among the two, JSONP requests are commonly used right now and CORS is expected to catch the mass. In this post we'll focus on the JSONP model request for OData Service.

To get a quick idea on the need for JSONP and how that stuff works, please have a quick glimpse here.  The code snippets I represent in this post will use the Nerd Dinner OData Feed for easier understanding.

After deciding to make the JSONP request, I used jQuery.ajax() method with url : http://www.nerddinner.com/Services/OData.svc/Dinners?$format=json & jsonpCallback option which resulted in error. Upon analyzing it, found out the error is "Uncaught SyntaxError: Unexpected token : " and the error is because the returned data is a simple JSON format and not wrapped with the callback method which is expected. Then I searched and found out the right format to make the OData JSONP request. The option is available as a OData query $callback.

Following are the three different alterations we can do to make a cross domain OData request using jQuery and JSONP.

Snippet 1:

$(document).ready(function () {

    $.ajax({
        url: 'http://www.nerddinner.com/Services/OData.svc/Dinners?$format=json&$callback=nerdCallback',
        dataType: "jsonp"
    });
});

function nerdCallback(result) {
    //Manipulate the resultant here....
    alert("Result count : " + $(result["d"]["results"]).length);
}

Snippet 2:

$(document).ready(function () {

    $.ajax({
        url: 'http://www.nerddinner.com/Services/OData.svc/Dinners?$format=json&$callback=?',
        dataType: "jsonp",
        jsonpCallback: "nerdCallback"
    });
});

function nerdCallback(result) {
    //Manipulate the resultant here....
    alert("Result count : " + $(result["d"]["results"]).length);
}

Snippet 3:

$(document).ready(function () {

    $.ajax({
        url: 'http://www.nerddinner.com/Services/OData.svc/Dinners?$format=json&$callback=?',
        dataType: "jsonp",
        success: function (result) {
            //Manipulate the resultant here....
            alert("Result count : " + $(result["d"]["results"]).length);
        }
    });
});

References:
http://en.wikipedia.org/wiki/JSONP
http://msdn.microsoft.com/en-us/library/ee834511.aspx
http://www.odata.org/

ASP.NET MVC 3 - Effectively using Model metadata

In ASP.NET MVC 3, most developers might have observed the existence of the HTML helpers - DisplayForModel and EditorForModel. This post is just to give a hint on how effectively we can use the specified HTML helpers and hence we won't have any code Snippets here.

I frequently come across situations of creating proof of concept and several sample applications to evaluate certain implementations / metrics. In such cases, lot of time I really worried on creating a sample data entry / display page which consumes my time, because my core objective is to evaluate my target as early as possible. The EditorForModel HTML helper in ASP.NET MVC actually saves me lot of time and help me to just focus on my target while doing such POCs.

While doing so I thought some simple tweaks, like a way to represent display name, order ... for the property might help me to tune my sample page further. To achieve that, I got the help of DataAnnotations / Metadata.

Some sample requirements are

• Need to hide a column - ScaffoldColumn(false)
• Hidden value - HiddenInput
• Display a different name - Display(Name = "Employee Name")
• Display in a custom order - Display(Order=1)
• Display format with different type than default - DataType(DataType.MultilineText)

Using these Metadata, we can, hide a property from creating HTML elements for that, create a hidden value element for a property, display a different name than that of the property, display in a custom order we need, specify what type of HTML element and what format can be used, etc.

Once you start explore other parameters of attributes and options, you can achieve lot of cases with minimal effort. Really its a time saving mechanism. If none of the options fits our scenario, then we can do the regular way.   

ASP.NET MVC 3 - Using Ninject for Dependency Injection

This post is to show how to implement Dependency Injection in ASP.NET MVC 3 application with simple steps. For Details on Dependency Injection (DI) please refer to my article here.

For the sake of simplicity and to understand the idea, I have simply specified all the classes in same project(web application here). In real time scenario this might extend to multiple class libraries based on the nature of the project.

In our example, we're targeting a Mark sheet creation with minimal and predefined data. MarkSheetController expects a MarksheetModel and the MarkSheetModel expects a MarkEvaluator which we're going to achieve with Constructor injection as follows.

Implementaiton Details

First we need to install the Ninject.MVC3 Nuget package using Nuget manager. For details on Nuget manager please verify my post here. Upon installing the package, a file called NinjectWebCommon.cs will get created under the folder called App_Start.  In the  static method called RegisterServices, add the statement to register the dependencies with Ninject container. Following Snippets are self explanatory to understand the concept.

Snippet 1: (App_Start/NinjectWebCommon.cs => NinjectWebCommon.RegisterServices())

private static void RegisterServices(IKernel kernel)
{
    kernel.Bind<IMarkSheetModel>().To<MarkSheetModel>();
    kernel.Bind<IMarkEvaluator>().To<MarkEvaluator>();
}

In the above snippet I've registered with Ninject that MarkSheetModel concrete class is bound to IMarkSheetModel and MarkEvaluator concrete class is bound to IMarkEvaluator interface. Based on this Ninject resolves the dependency by creating instance for the respective concrete class and pass it in the place of the related interface specification. Following snippet shows how we're making use of constructor injection to achieve that.

Snippet 2: (MarkSheetController - note the IMarkSheetModel parameter in the constructor)

public class MarkSheetController : Controller
{
    private IMarkSheetModel model;


    public MarkSheetController(IMarkSheetModel modelParam)
    {
        model = modelParam;
    }

    //
    // GET: /MarkSheet/

    public ActionResult Index()
    {
        model.LoadDetails();
        return View(model);
    }

}

Snippet 3: (MarkSheetController - note the IMarkEvaluator parameter in the constructor)

public class MarkSheetModel : IMarkSheetModel
{
    private IMarkEvaluator evaluator;

    public string StudentId { get; set; }
    public string StudentName { get; set; }
    public IEnumerable<Subject> SubjectList { get; set; }
    public int Total { get; private set; }
    public decimal Average { get; private set; }
    public string Grade { get; private set; }
    
    public MarkSheetModel(IMarkEvaluator evaluatorParam)
    {
        evaluator = evaluatorParam;
    }

    public void LoadDetails()
    {
        this.StudentId = "S1001";
        this.StudentName = "Sample";

        this.SubjectList = new List<Subject> { new Subject{ SubjectId="Sub0001", SubjectName ="Sub 1", Score = 78},
        new Subject{ SubjectId="Sub0002", SubjectName ="Sub 2", Score = 84},
        new Subject{ SubjectId="Sub0003", SubjectName ="Sub 3", Score = 72},
        new Subject{ SubjectId="Sub0004", SubjectName ="Sub 4", Score = 69}};

        this.Total = this.evaluator.CalculateTotal(this.SubjectList);
        this.Average = this.evaluator.CalculateAverage(this.SubjectList);
        this.Grade = this.evaluator.CalculateGrade(this.SubjectList);
    }

}

public interface IMarkSheetModel
{
    void LoadDetails();
    IEnumerable<Subject> SubjectList { get; set; }
    string StudentId { get; set; }
    string StudentName { get; set; }
}

public class Subject
{
    public string SubjectId { get; set; }
    public string SubjectName { get; set; }
    public int Score { get; set; }
}

Snippet 4:

public class MarkEvaluator : IMarkEvaluator
{
    public int CalculateTotal(IEnumerable<Subject> subjectList)
    {
        return subjectList.Sum(su => su.Score);
    }

    public decimal CalculateAverage(IEnumerable<Subject> subjectList)
    {
        return Convert.ToDecimal(CalculateTotal(subjectList)) / subjectList.Count();
    }

    public String CalculateGrade(IEnumerable<Subject> subjectList)
    {
        decimal averageScore = CalculateAverage(subjectList);

        if (averageScore > 80)
        {
            return "Grade A";
        }
        else if (averageScore > 70)
        {
            return "Grade B";
        }
        else if (averageScore > 60)
        {
            return "Grade C";
        }
        else
        {
            return "Grade D";
        }
    }
}

public interface IMarkEvaluator
{
    decimal CalculateAverage(IEnumerable<Subject> subjectList);
    string CalculateGrade(IEnumerable<Subject> subjectList);
    int CalculateTotal(IEnumerable<Subject> subjectList);
}

While executing the above snippets, the constructor of the MarkSheetController and MarkSheetModel is supplied with the respective instance.

References:
https://github.com/ninject/ninject.web.mvc/wiki/MVC3

ASP.NET MVC 3 - Converting / Serializing .Net objects into JSON format in View

In ASP.NET MVC often we might need to use / manipulate Model objects or its property inside the javascript. In order to achieve that, we usually try to serialize the .Net object and store it in a javascript variable which will then be accessed further in javascript.

Using Razor view engine, we can achieve this easily. In the following, I'll explain two approaches with a simple View(.cshtml) snippet to understand the concept. 

Approach 1: (Using Json.Encode(...))

Snippet 1: (Index.cshtml)

@model Web.POC.Models.FeedModel
@{
    ViewBag.Title = "Feed viewer";
}

<script type="text/javascript">
    var entries = @Html.Raw(Json.Encode(Model.FeedEntries));    
</script>

Approach 2: (Using JavaScriptSerializer)

Snippet 2: (Index.cshtml)

@model Web.POC.Models.FeedModel
@{
    ViewBag.Title = "Feed viewer";
}

@{
   System.Web.Script.Serialization.JavaScriptSerializer serializer = new System.Web.Script.Serialization.JavaScriptSerializer();
}
<script type="text/javascript">
    var entries = @Html.Raw(serializer.Serialize(Model.FeedEntries));    
</script>

Assumptions:

As specified, the above snippets are just to understand the concept. In that, consider that we're having a model "FeedModel" which holds a property called "FeedEntries" of some custom type which we're trying to convert into JSON object.

Also note that we need to use the HTML.Raw() to indicate that the output should not be HTML encoded.

ASP.NET - Avoid multiple postbacks for control set using validators by disabling the button

In ASP.NET disabling the button when a postback got initiated through button click and avoiding multiple postbacks or submits until the specific response is processed is one of the common thing that we need to implement. This is especially  needed for some long running operations. During my initial days, I achieve this as specified here. Later, I followed as specified here in an article by Abhijit Jana. Few days back one of my friend asked me about how to disbale the button to avoid multiple submits / postbacks for control set that are using ASP.NET validation. To achieve that, I tried a sample and came up with two approaches.

Let us consider a scenario that we have a text box, a required field validator and button tied together using validation group.

Approach 1: (Using jQuery and a button proxy)

• Apply CSS style="display:none;"  to the button you want to disable.
• Include a new button for display purpose and add a jQuery that validates the validation group using the function Page_ClientValidate() passing the validationgroup as parameter. 
• Based on the return value of Page_ClientValidate(), raise the click event of the hidden button using jQuery and disable the display button. 

The following script snippet helps us to achieve that.

Snippet 1:(script)

<script type="text/javascript">
    $(document).ready(function () {
        initAction();
    });

    function initAction() {
        $('#Button2').click(function (event) {
            event.preventDefault();

            if (Page_ClientValidate('ValGroup1')) {
                this.disabled = true;
                $('#Button1').click();
            }

        });
    }
</script>

Here is the full implementation of the same.

Snippet 2:(DisableButtonOnClickApp1.aspx)

<%@ Page Language="C#" AutoEventWireup="true" CodeFile="DisableButtonOnClickApp1.aspx.cs"
    Inherits="DisableButtonOnClickApp1" %>

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head runat="server">
    <title></title>
        <script type="text/javascript" src="http://code.jquery.com/jquery-1.7.1.min.js"></script>
    <script type="text/javascript">
        $(document).ready(function () {
            initAction();
        });

        function initAction() {
            $('#Button2').click(function (event) {
                event.preventDefault();

                if (Page_ClientValidate('ValGroup1')) {
                    this.disabled = true;
                    $('#Button1').click();
                }

            });
        }
    </script>

</head>
<body>
    <form id="form1" runat="server">
    <div>
        <table border="0" cellpadding="0" cellspacing="0">
            <tr>
                <td>
                    <asp:Label ID="Label1" runat="server" Text="Label 1 : "></asp:Label>
                </td>
                <td>
                    <asp:TextBox ID="TextBox1" runat="server" ValidationGroup="ValGroup1"></asp:TextBox>
                    <asp:RequiredFieldValidator ID="RequiredFieldValidator1" runat="server" ErrorMessage="RequiredFieldValidator"
                        ControlToValidate="TextBox1" ValidationGroup="ValGroup1"></asp:RequiredFieldValidator>
                </td>
            </tr>
            <tr>
                <td colspan="2" style="text-align: center;">
                    <asp:Button ID="Button1" runat="server" Text="Hide Button" Style="display: none;"
                        ValidationGroup="ValGroup1" onclick="Button1_Click" />
                    <asp:Button ID="Button2" runat="server" Text="Display Button" />
                </td>
            </tr>
            <tr>
            <td colspan="2" style="text-align: center;">
                <asp:Label ID="StatusMessageLabel" runat="server" Text=""></asp:Label>
            </td>
            </tr>
        </table>
    </div>
    </form>
</body>
</html>

Snippet 3:(DisableButtonOnClickApp1.aspx.cs)

public partial class DisableButtonOnClickApp1 : System.Web.UI.Page
{
    protected void Page_Load(object sender, EventArgs e)
    {

    }

    protected void Button1_Click(object sender, EventArgs e)
    {
        StatusMessageLabel.Text = "Button 1 click .......";
    }
}

If you are not much interested in using one additional control to achieve this, the following approach might help you.

Approach 2: (Using GetPostBackEventReference with PostBackOptions and UseSubmitBehavior)

The idea for this approach evolved from the article I specified above. Following snippets helps you to understand the implementation easily. Here we're going to use ClientScript.GetPostBackEventReference with PostBackOptions and Button.UseSubmitBehavior to achieve the resultant.

Snippet 4: (DisableButtonOnClickApp2.aspx.cs)

public partial class DisableButtonOnClickApp2 : System.Web.UI.Page
{
    protected void Page_Load(object sender, EventArgs e)
    {
        DisableButtonOnClick(Button1, "ValGroup1");
    }

    protected void Button1_Click(object sender, EventArgs e)
    {
        StatusMessageLabel.Text = "button 1 click......";
    }

    private void DisableButtonOnClick(Button buttonControl, string validationGroup)
    {
        StringBuilder disableButtonScriptBuilder = new StringBuilder();
        PostBackOptions postBackOptions = new PostBackOptions(buttonControl);

        postBackOptions.ValidationGroup = validationGroup;
        postBackOptions.Argument = "";
        postBackOptions.PerformValidation = true;
        postBackOptions.ActionUrl = "";
        postBackOptions.TrackFocus = false;
        postBackOptions.ClientSubmit = false;
        postBackOptions.RequiresJavaScriptProtocol = true;

        disableButtonScriptBuilder.Append(ClientScript.GetPostBackEventReference(postBackOptions, false));
        disableButtonScriptBuilder.Append("; if(Page_ClientValidate('");
        disableButtonScriptBuilder.Append(validationGroup);
        disableButtonScriptBuilder.Append("')){this.value='Processing...';this.disabled = true;}");

        //buttonControl.OnClientClick = ClientScript.GetPostBackEventReference(postBackOptions, false) + "; if(Page_ClientValidate('" + validationGroup + "')){this.value='Processing...';this.disabled = true;}";
        buttonControl.OnClientClick = disableButtonScriptBuilder.ToString();
        buttonControl.UseSubmitBehavior = false;
    }
}

Snippet 5: (DisableButtonOnClickApp2.aspx)

<%@ Page Language="C#" AutoEventWireup="true" CodeFile="DisableButtonOnClickApp2.aspx.cs" Inherits="DisableButtonOnClickApp2" %>

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">

<html xmlns="http://www.w3.org/1999/xhtml">
<head runat="server">
    <title></title>
</head>
<body>
    <form id="form1" runat="server">
    <div>
    <table border="0" cellpadding="0" cellspacing="0">
            <tr>
                <td>
                    <asp:Label ID="Label1" runat="server" Text="Label 1 : "></asp:Label>
                </td>
                <td>
                    <asp:TextBox ID="TextBox1" runat="server" ValidationGroup="ValGroup1"></asp:TextBox>
                    <asp:RequiredFieldValidator ID="RequiredFieldValidator1" runat="server" ErrorMessage="RequiredFieldValidator"
                        ControlToValidate="TextBox1" ValidationGroup="ValGroup1"></asp:RequiredFieldValidator>
                </td>
            </tr>
            <tr>
                <td colspan="2" style="text-align: center;">
                    <asp:Button ID="Button1" runat="server" Text="Button" onclick="Button1_Click" />
                </td>
            </tr>
            <tr>
            <td colspan="2" style="text-align: center;">
                <asp:Label ID="StatusMessageLabel" runat="server" Text=""></asp:Label>
            </td>
            </tr>
        </table>
    </div>
    </form>
</body>
</html>

Based on your comfort, use any of the above two approaches.

REFERENCES
http://dailydotnettips.com/2011/02/07/disabling-button-when-performing-some-operation-in-asp-net/#more-482

Unauthorized access exception when value set for property using Secondary Thread - Silverlight

Thinking about threading in Silverlight 4, we choose between the two options

1. Background worker
2. Managing thread manually

Most of the time-consuming operations will be handled using the Background worker without blocking the UI. It ultimately executes in a separate thread. The main functions and events to know in Background worker are DoWork, RunWorkerCompleted and RunWorkerAsync. As the name specifies we need to add the event handler for DoWork event for doing the background operation. Add the handler for RunWorkerCompleted to receive the operation completed notification.  RunWorkerAsync method is to start the operation. The main thing to note is that we should not access directly access the UI objects from the DoWork event.

Also there are times where we need to manage the threads manually. Since Silverlight runs in a Single threaded apartment (STA) application model and the default is the UI thread, accessing the UI objects from other thread leads to cross thread exception. Knowing this fact I worked with a secondary thread and updated a property in my ViewModel. In that scenario I got an exception and digging into the inner exception found out the cross thread exception. For understanding I've created a sample scenario that raises exception.

Snippet 1

public partial class ThreadTest : UserControl, INotifyPropertyChanged
{        
    private string textValue;
    public string TextValue
    {
        get { return textValue; }
        set
        {
            textValue = value;
            NotifyPropertyChanged("TextValue");
        }
    }

    public ThreadTest()
    {
        InitializeComponent();

        InitControls();
    }

    private void InitControls()
    {
        this.DataContext = this;
    }

    private void Direct_Click(object sender, RoutedEventArgs e)
    {
        TextValue = "direct text......";
    }

    private void UsingTask_Click(object sender, RoutedEventArgs e)
    {
        Thread secondaryThread = new Thread(SecondaryThreadWorker);
        secondaryThread.Start();
        
    }

    private void SecondaryThreadWorker()
    {
        TextValue = "thread text.......";            
    }


    #region INotifyPropertyChanged Members

    protected void NotifyPropertyChanged(string propertyName)
    {
        PropertyChanged(this, new PropertyChangedEventArgs(propertyName));
    }

    public event PropertyChangedEventHandler PropertyChanged;

    #endregion
}

Then examining the cause, realized that the specific property raises an property changed notification . Immediately recalled the excellent article from Pete Brown in his 10rem.net blog.

The following implementation does the trick.

Snippet 2 (Just replace the INotifyPropertyChanged implementation)

#region INotifyPropertyChanged Members

protected void NotifyPropertyChanged(string propertyName)
{
    if (PropertyChanged != null)
    {
        if (Deployment.Current.Dispatcher.CheckAccess())
        {
            PropertyChanged(this, new PropertyChangedEventArgs(propertyName));
        }
        else
        {
            Deployment.Current.Dispatcher.BeginInvoke(() =>
            {
                PropertyChanged(this, new PropertyChangedEventArgs(propertyName));
            });
        }
    }
}

public event PropertyChangedEventHandler PropertyChanged;

#endregion

When I typed "Deployment.Current.Dispatcher.", it didn't show up the "CheckAccess()" method in the intellisense. But I verified it in the object which shows the existence of the method. So I manually typed the  "CheckAccess()" method name and got complied without error.

References
http://10rem.net/blog/2012/01/10/threading-considerations-for-binding-and-change-notification-in-silverlight-5

ASP.NET MVC 3 - Accessing Session inside Task of AsyncController

Before getting into the core of this post, let us have a quick look at Async controller. The base thing that we need to understand is that, Async controllers are implemented to achieve the efficient service of the incoming requests rather than faster processing of individual request. Scenarios like the case of having a network call which consumes some time might block the thread (that is processing the current request) from processing other requests. For understanding the need for using the AsyncController, there exists several articles over the web. One such nicely explained article is here.

In order to implement the Async controller, we need to do the following.
Derive the controller from AsyncController. 
Specify an <ActionName>Async and <ActionName>Completed methods for each Actions, where <ActionName>Async is of return type void and the relevant <ActionName>Completed returns the ActionResult.
Usually the network calls will be made in the <ActionName>Async method and the resultant is passed over to the <ActionName>Completed method as arguments.

Above specified info are fine enough to implement the Async controller. Apart from that, we also need to think of effectively handling the network calls and other time consuming items inside the <ActionName>Async method. In real case, usually we'll call the method in Model / ViewModel which then initiates the network call,  call to the Service Locator or Service Agent. For achieving this, we can alter the Model / ViewModel to expose the relevant methods async mode. Instead we can also seek help from the .Net built in library named Task Parallel Library (TPL). Instead of explaining further, I directly dive into the code sample from which you can easily understand the base.

Snippet 1:

public void IndexAsync(int id)
{
    AsyncManager.OutstandingOperations.Increment();

    Task.Factory.StartNew(() => {
        InfoViewModel moreViewModel = new InfoViewModel();
        InfoViewModel.MoreModel moreModel = moreViewModel.GetDetails(id);

        AsyncManager.Parameters["moreModel"] = moreModel;
        AsyncManager.OutstandingOperations.Decrement();
    });
    
}

public ActionResult IndexCompleted(InfoViewModel.MoreModel moreModel)
{
    return View("Info", moreModel);
}

The AsyncManager related operations specified in the above code helps us to achieve handling of  aysnc operations effectively. For example, the Increment(), Decrement() methods and Parameters property helps the controller to identify, when to make the call to relevant <ActionName>Completed method with specific parameters. As you guess, the Task.Factory.StartNew(Action) will be executed in a separate thread and in the mean time, statements following it got executed and waits until the AsyncManager.OutstandingOperations becomes zero after which <ActionName>Completed gets called.

OK, now comes the core of the post. Think of a situation where we need to use access the Session values inside the method of Model / ViewModel which is initiated from Task.Factory.StartNew(Action).  Since the Task is executed in a separate thread, that thread won't hold the HttpContext that is System.Web.HttpContext.Current will be null. To make it work, we can assign the System.Web.HttpContext.Current with the HttpContext got from ControllerContext as follows.

Snippet 2:

System.Web.HttpContext.Current = ControllerContext.HttpContext.ApplicationInstance.Context;

Please note that the above specified fix is just a workaround. For such scenarios, we need to alter the method in the Model in such a way to receive the value retrieved from session as one of the parameter. 

As we might use this repeatedly, we can create a base class and make use it in places wherever required.

Snippet 3: (BaseAsyncController) 

public class BaseAsyncController : AsyncController
{
    protected void StartContextEnabledTask(Action action)
    {
        Task.Factory.StartNew(() => {
            System.Web.HttpContext.Current = ControllerContext.HttpContext.ApplicationInstance.Context;
            action();
        });
    }
}

Snippet 4: (Full Implementation)

public class InfoController : BaseAsyncController
{
    public void IndexAsync(int id)
    {
        AsyncManager.OutstandingOperations.Increment();
    
        StartContextEnabledTask(() => {
            InfoViewModel moreViewModel = new InfoViewModel();
            InfoViewModel.MoreModel moreModel = moreViewModel.GetDetails(id);
    
            AsyncManager.Parameters["moreModel"] = moreModel;
            AsyncManager.OutstandingOperations.Decrement();
        });
        
    }
    
    public ActionResult IndexCompleted(InfoViewModel.MoreModel moreModel)
    {
        return View("Info", moreModel);
    }    
}

References:
http://www.aaronstannard.com/post/2011/01/06/asynchonrous-controllers-ASPNET-mvc.aspx
http://msdn.microsoft.com/en-us/library/dd537609.aspx
http://craigcav.wordpress.com/2010/12/23/asynchronous-mvc-using-the-task-parallel-library/