Single Sign-On In ASP.NET Core With Azure Active Directory

8. May 2017 15:21

.NET Core ASP.NET ASP.NET MVC Azure C# Cloud 

Visual Studio has made integrating of Azure Active Directory authentication in web applications very simple. If you are creating the application and want to integrate the AD authentication then it is a pretty easy task given that you have an Active Directory in Azure. But what if you have an application and it is out there running and now you want to change the authentication medium of your application so you can have better control and security of your web application. I already have an application and I want to integrate Azure AD authentication, hence I will not be able to change the authentication just like that. I have to manually make the changes to do so. So here it is what I have done to integrate Azure Active Directory in my ASP.NET Core web application.


Step 1: Create Azure Active Directory

When provisioning a new Azure Active Directory you have to go to the old portal At the time of writing this blog post, the product team is still in a process of migrating the complete AD feature sets from the old portal to the new portal. Therefore, to create a new AD in Azure, click the Active Directory in the left pane and you will be able to see all the active directories. In the below screen shot you can see that I am a part of Microsoft Active Directory as a User.

If you are planning to use any of the current Active Directory out of the one listed, you can click on it and register your application and users. Note that you have to be a Global Administrator to do that. If you want to create a new Active Directory, then click the NEW button at the bottom left corner to create a new Active Directory.

Click on CUSTOM CREATE and a new dialog box will appear in front of you to enter the name of your Active Directory.

The green check box in the DOMAIN NAME denotes that the name you have chosen for your active directory is available. When done click the tick button or icon at the bottom right corner of the dialog box to create a new active directory.

Step 2: Add a user

After the creation of active directory is successfully completed, you can see the new created active directory in the list. Notice the ROLE as Global Administrator. Click on the active directory name to view the details. First, I am going to add a new user to the directory. When inside the active directory, click the USERS which in the top of the page.

I can see a default user added to my new created active directory here which is my official Microsoft user. I cannot delete this user neither I can change much of the roles and other information of this user because it something which is being managed by Microsoft. So to add a new user, click the ADD USER button at the bottom of the page. This will open up a dialog box which allows you to add or create a new user to the active directory. In my case I am going to add a normal user not an administrator, I will keep the the default prompts. The whole process of adding a new user will take 3 steps to complete.
In the next screen enter the details about the user. I want to add this user as a normal user hence I am selecting User as a ROLE.
In the last step, click the create button to add a new user and generate a temporary password for the new user. 
Make a note of this password somewhere as you will need it to log in to the account. After the first login, user is prompt to change the password.

Step 3: Add an application to the Active Directory

To add a new application, click on the APPLICATION link. Click the ADD button at the bottom of the page to add a new application and then click Add an application my organization is developing.
In the next step enter the name of the application and select the type of application you are building. In my case, it is a web application.
In the next step, you add a sign-in URL and app ID URI. The app ID URI I have added is in this format: http://<directoryname> Replace the directoryname with your directory name. The adauth is the name of my application.
Once the adding of application is successful, click the USERS link. I can see the new user I have created in the user lists but it is not being assigned to the application yet. To do that, click the ASSIGN button at the bottom of the page and click YES. The ASSIGNED status will be changed from No to Yes.

Step 4: Setting up the web application

The easy way to set the application up for AD authentication is to create a new "demo" application and select Change Authentication from the new project dialog box. Select Work or School Accounts and enter the domain name. This will create a new application with all the things set up for your domain. The reason I said "demo" application because I want to just get the changes that Visual Studio does it for me when I select the AD authentication. This application will now act like a sample application from where I can take the implementation and add it to my existing application.
Here are the changes I have made to my existing application. As of now my existing application does not have user authentication and now as I want to implement, I will add the account controller and two views which will handle the authentication. In my case this is the account controller, you can name the controller as per your choice. Add two new views in the Views directory for your controller. Call them AccessDenied.cshtml and SignedOut.cshtml respectively.
Note that all the changes I am making to my existing application are exactly the same as the sample application.
Code for AccessDenied.cshtml view.
   ViewData["Title"] = "Access Denied";

    <h1 class="text-danger">Access Denied.</h1>
    <p class="text-danger">You do not have access to this resource.</p>
Code for SignedOut .cshtml view.
   ViewData["Title"] = "Sign Out";
<p class="text-success">You have successfully signed out.</p>
Add below NuGet packages for AD authentication support in the application.
Install-Package Microsoft.AspNetCore.Authentication
Install-Package Microsoft.AspNetCore.Authentication.Cookies
Install-Package Microsoft.AspNetCore.Authentication.OpenIdConnect
In the .NET Core project, you will also see a new file called appsettings.json. Add the below configuration to it. Your configuration will than mine.
"Authentication": {
  "AzureAd": {
    "ClientId": "0d9h56ba-ccfd-42f2-9a75-64d8e96756e6",
    "AadInstance": "",
    "CallbackPath": "/signin-oidc",
    "Domain": "",
    "TenantId": "d58e5363-f09c-4d65-9808-cd88396529f0"
In the Startup.cs file, configure the request pipeline by adding the below code in the Configure method. I have added this code just after the UseStaticFiles() functions. This enables the application to use the cookie authentication and OpenId authentication.

app.UseOpenIdConnectAuthentication(new OpenIdConnectOptions
    ClientId = Configuration["Authentication:AzureAd:ClientId"],
    Authority = Configuration["Authentication:AzureAd:AADInstance"] + Configuration["Authentication:AzureAd:TenantId"],
    CallbackPath = Configuration["Authentication:AzureAd:CallbackPath"]
In the ConfigureServices method, add the below code.
services.AddAuthentication(options => options.SignInScheme = CookieAuthenticationDefaults.AuthenticationScheme);
In the AccountController.cs, add the below code.
public IActionResult SignIn()
    return Challenge(
        new AuthenticationProperties { RedirectUri = "/" }, OpenIdConnectDefaults.AuthenticationScheme);

public IActionResult SignOut()
    var callbackUrl = Url.Action(nameof(SignedOut), "Account", values: null, protocol: Request.Scheme);
    return SignOut(new AuthenticationProperties { RedirectUri = callbackUrl },
        CookieAuthenticationDefaults.AuthenticationScheme, OpenIdConnectDefaults.AuthenticationScheme);

public IActionResult SignedOut()
    if (HttpContext.User.Identity.IsAuthenticated)
        return RedirectToAction(nameof(HomeController.Index), "Home");

    return View();

public IActionResult AccessDenied()
    return View();
Add a new partial view in the Shared folder and name it _LoginPartial.cshtml just like we have in new default scaffold project.
@using System.Security.Principal
@using System.Security.Claims

@if (User.Identity.IsAuthenticated)
    var claims = ((ClaimsIdentity)User.Identity).Claims;
    var username = claims.FirstOrDefault(c => c.Type == "name")?.Value;

    //Uncomment the below code to see the complete email address of the user.
    @*<li class="navbar-text">Hello @User.Identity.Name!</li>*@  
    <ul class="nav navbar-nav navbar-right">
        <li class="navbar-text">Hello @username!</li>
        <li><a asp-area="" asp-controller="Account" asp-action="SignOut">Sign Out</a></li>
    <ul class="nav navbar-nav navbar-right">
        <li><a asp-area="" asp-controller="Account" asp-action="Signin">Sign in</a></li>
Open _Layout.cshtml page, add the below line of code to render this partial view so users can sign in and sign out.
@await Html.PartialAsync("_LoginPartial")
And we are done! This is a minimalistic implementation that you can get start with. Azure Active Directory also provides groups and roles which you need to work with if you are working with the enterprise scale application where security is a major concern. You also have to tweak the code for that. Let's try executing the application and see what happens when I try to authenticate the user for the first time.
When you land on the home page you can see the Sign in link at the top right of the web page.
When you click on the sign in link, you will be redirected to the where you will be prompted for the user name and password. After the first successful login, you will be prompted to change the current password.
After the successful change of the password, you will be re-directed to the application with your user name as displayed at the top right corner.
 No Rating

Visualizing ADXL345 Accelerometer Output In Windows Forms From Raspberry PI 2

26. September 2016 21:21

C# Raspberry Pi Windows 

I have been playing a lot with Raspberry PI and Python these days and have also started a side project to create robot car controller with a Xbox 360 wireless controller. I am planning to controller a robot car using a custom interface from my Windows machine. Initially I tried with PyQt but I dropped the idea for now as Python is not my primary expertise and it will take some time for me to learn the basics and get into some serious coding. Controlling something remotely or visualizing the information from your device on the UI is always better. For a proof of concept I built a Windows Forms application which will request the output from Raspberry PI over TCP and then visualize it on the UI using a wonderful chart library that I found today. There is a slight change in the Python code as I have to transmit readings from Raspberry PI over TCP so my Windows Forms application can read the output and utilize it to visualize it.

Raspberry PI Python Application

Reading the ADXL345 output is fairly simple and you can do this without any problem by using Adafruit's ADXL345 Python library. This library will get you started in a jiffy and you will have your readings. I am not going into the details about the different features of the library I am using and ADXL345. The only thing you will have to do is to wire ADXL345 to your Raspberry PI and fire the example code you got from Adafruit's repository from Github. If you have set it up correctly this, you will see X, Y, Z values on your terminal. If the terminal displays 0s for all the three axis then there is something wrong with the wiring (assuming you are using the example code). This is how your writing should be like.

Now I have made slight changes in the code so that I can send the data over TCP. I have created a simple TCP server which exposes Raspberry PI through IP address and a port number. This is the complete code for reading values from ADXL345.
import Adafruit_ADXL345
import socket

HOST = ""
PORT = 50007

accel = Adafruit_ADXL345.ADXL345()

s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.bind((HOST, PORT))
conn, addr = s.accept()

while 1:
    data = conn.recv(1024)

    if data.strip() == "adxl":
        x, y, z =
        data = 'X:' + str(x) + '|Y:' + str(y) + '|Z:' + str(z)
In the code, you can see that I am not just transmitting the readings just like that. I only transmit when there is a request from another application is there to get the readings. This is a good approach as I don't the PI to transmit the readings even if nobody is requesting it. I am sending a single string as a response which I am going to read at my Windows Forms application and use it.
You need to change the IP address and port number according to your Pi's configuration.
Once this is done you can then execute the script from the terminal by firing the below command
sudo python

This is it from Raspberry PI side. Let's take a look from the Windows side now.

Windows Forms Application

This will be a simple Windows Forms application which will connect to the Raspberry PI using the Pi's IP address and the port on which the server is running. The Windows Form application is also a TCP client which sends a request to Raspberry PI for the updated readings from ADXL345. As I mentioned above in the Python code that the readings are not just being transmitted but a client has to send a request to to get the readings.

Let's get started with the visualizing part. While was searching for a perfect charting library, I came across this wonderful library called Live Charts. This library is open-source, easy to use, has lots of awesome charts and a very robust documentation and moreover it supports Windows Forms as well as on WPF applications. I highly recommend you to look at the examples to get the idea about this library.

As usual, I will be using Nuget command to install the library.

PM> Install-Package LiveCharts.WinForms
You also have to add the custom controls in the toolbox. You can follow the steps here to add the Live Charts control in the toolbox. In the design mode, choose the Cartesian Chart control from the toolbox and add it to the form. After adding the control to the form, I have dock it to the top of the form so I can utilize the lower section of the form to display the readings in numbers. Here is how my form looks like in design mode.
After the designing for the form is complete, I can initialize the chart to display the readings from Raspberry PI. The chart require few things to be setup before it can start displaying the data correctly. The accelerometer I am using is a 3-axis accelerometer and therefore it is going to send me readings for 3 axis i.e X, Y and Z. This means I will be needing 3 line series to visualize the information from accelerometer. To have 3 line series display the data I have to use CharValues class which will let me display the readings in the line series. As I have to update my chart when I request the new readings from Raspberry PI, I have to use ObservableValue type. The ObservableValue will make sure that whenever there is a change in the value, it gets reflected in the chart and therefore keeping the chart always up to date.
public ChartValues<ObservableValue> Values_X { get; set; }
public ChartValues<ObservableValue> Values_Y { get; set; }
public ChartValues<ObservableValue> Values_Z { get; set; }
Because I am going to make request to the server running on my Raspberry PI and request for the ADXL345 readings, I have to create a TCP client in order to do so.
Socket server;
IPEndPoint ipEndPoint;
byte[] data;
When the application is loaded, I want to make connection to Raspberry PI by using it's IP address and port number on which the server is running on. To do this I will add the below code to the constructor of my application.
data = new byte[1024];

ipEndPoint = new IPEndPoint(
                IPAddress.Parse(""), 50007);

server = new Socket(AddressFamily.InterNetwork,
               SocketType.Stream, ProtocolType.Tcp);

catch (SocketException)
    MessageBox.Show("Unable to connect to server.", "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);

The above code is self-explanatory as I am just connecting to Raspberry PI.

In the form load event, I have to initialize chart values and add 3 line series to my CartesianChart for X, Y and Z axis respectively. In my case the name of the Cartesian Chart is ADXLViz.

Values_X = new ChartValues<ObservableValue>();
Values_Y = new ChartValues<ObservableValue>();
Values_Z = new ChartValues<ObservableValue>();

ADXLViz.Series.Add(new LineSeries
    Values = Values_X,
    Title = "X",
    StrokeThickness = 4,
    PointGeometrySize = 0,
    DataLabels = true

ADXLViz.Series.Add(new LineSeries
    Values = Values_Y,
    Title = "Y",
    StrokeThickness = 4,
    PointGeometrySize = 0,
    DataLabels = true

ADXLViz.Series.Add(new LineSeries
    Values = Values_Z,
    Title = "Z",
    StrokeThickness = 4,
    PointGeometrySize = 0,
    DataLabels = true

Play around with different properties in the LineSeries class and set the output that suits your need or your eye. Notice the Values properties in the LineSeries class which holds the value which is received from Raspberry PI.

I am requesting the readings from Raspberry PI every 500ms and to do this I am using a Timer class. This is easy to use and configure. Just drag and drop the Timer control from the toolbox. In the properties change the Interval property to 500 and Enabled to True. After this I set the Tick event which will request the new readings every 500ms.

string stringData;
data = new byte[1024];
int recv = server.Receive(data);
stringData = Encoding.ASCII.GetString(data, 0, recv);
string[] adxl = stringData.Replace(" ", "").Split('|');

Values_X.Add(new ObservableValue(Convert.ToDouble(adxl[0].Split(':')[1])));
Values_Y.Add(new ObservableValue(Convert.ToDouble(adxl[1].Split(':')[1])));
Values_Z.Add(new ObservableValue(Convert.ToDouble(adxl[2].Split(':')[1])));

if (Values_X.Count > 10) Values_X.RemoveAt(0);
if (Values_Y.Count > 10) Values_Y.RemoveAt(0);
if (Values_Z.Count > 10) Values_Z.RemoveAt(0);

X_Label.Text = Convert.ToString(adxl[0].Split(':')[1]);
Y_Label.Text = Convert.ToString(adxl[1].Split(':')[1]);
Z_Label.Text = Convert.ToString(adxl[2].Split(':')[1]);
The stringData variable holds the response or reading which is received from Raspberry PI. If you check the Python code above you will see that the response is being sent in a single string called data. I am splitting the string here and extracting the values and setting it to ChartValues<ObservableValue> object. This will render all the 3 line series chart and update it with the readings received from PI. But the problem here will be that in few minutes the chart will be hard to see due to lot of data congestion in chart. In order to overcome this problem, I have removed the first value from all the 3 line series when the chart displays more than 10 values. You can change the value from 10 to whatever suits you.
The overall idea of building this application is to see whether it is possible or easy to communicate from Raspberry PI apart from using frameworks like WebioPI or Flask. The complete source code for Raspberry PI and Windows Forms application is available on Github along with Fritzing sketch file and additional instructions to install the dependencies.
 No Rating

.NET Projects You Should Be Follwing On Github

19. July 2015 04:51


Open-source has entirely change the programming and developers world. Today you can create any application, game, mobile app without spending a single penny. Thanks to open-source software and awesome community of developers and people behind them. As a .NET developer I have been developing enterprise applications for quite a long time and now I have shifted my focus towards developing products and understanding what it takes to make a successful product launch.

Back then, I used to spend most of my time in investigating the new technologies and what technology we should be using to get this thing done. I still do that today, not because it is the requirement of the project but because I have been asking a lot of questions. The list of projects I have compiled below are the projects that have helped me in learning lots of new things and insights of the programming and I hope this does the same for you as well.  Here is the list of awesome open-source project that you should be following on Github.


We all know about, it is an awesome tool and a complete replacement of Photoshop (at least for me). And yet there is another project which is almost the same and open-source and it works on Linux and Mac. It uses Gtk# (Gtk sharp) to run on both Window and Linux platforms. This project is a must have if you are a .NET guy and want to get yourself into some serious programming. You will learn about the insights of using gtk# in your projects. Though Microsoft already took the steps to have .NET FX on Linux but still this project is a great learning source.

Official site:




I take a lot of screen shots and record screen casts as well for my personal use. But I used to use two different tools to get the work done. This is one of the tools that will not just take screen shots or just let you record your screen casts easily, it will also allow you to upload them to the 40 different image storing cloud services. Dive into the source code and see the awesomeness under the hood. Here is the project description as seen on Github.

ShareX is an open source program that lets you take screenshots or screencasts of any selected area with a single key, save them in your clipboard, hard disk or instantly upload them to over 40 different file hosting services. In addition to taking screenshots, it can upload images, text files and all other different file types.

Official site:



StackExchange - Data Explorer

You got a programming question, you Google it and it redirects it to StackOverflow. StackOverflow needs no introduction among programmers. StackOverflow is one of the Q&A site dedicated to the developers to get the answers for their problems. But it is just one site. In the recent years StackExchange has grew up and not just providing support for programmers but also helping folks from other fields. Now the data StackExchange has is available for anyone out there for free under creative-commons. If you are interested in looking into the source code that powers the user to query that immense amount of data bank then head over to Github and fork this project. StackExchange is all about Microsoft stack and this tool is also written in ASP.NET MVC3.

Official App:



Mini Blog

This is the minimalistic blog engine written in ASP.NET web pages by the author of BlogEngine.NET, Mads Kristensen. I started my bog with BlogEngine.NET and I had an amazing experience with it. MiniBlog is totally different in terms of features that are offered by BlogEngine.NET. This project will tell you the power of web pages and how you can write your own simple site without wasting much time.

Demo: (with user name and password as demo).



Fluent Scheduler

If you want to run cron jobs or automated jobs in your application quietly, then this is the library you should be using. The documentation is pretty sleek and get you started in no time. But other than that you should take a look at the source code and see how nicely this has been done.




A Micro-ORM used by StackExchange sites. This is a perfect replacement for EF. This is just a single file that you can drop in your project and get started.

Dapper is a single file you can drop in to your project that will extend your IDbConnection interface.




A .NET library to access MediaWiki API. The library is almost 3 years old but the source code will worth the look. Excerpt from Github:

LinqToWiki is a library for accessing sites running MediaWiki (including Wikipedia) through the MediaWiki API from .Net languages like C# and VB.NET.

It can be used to do almost anything that can be done from the web interface and more, including things like editing articles, listing articles in categories, listing all kinds of links on a page and much more. Querying the various lists available can be done using LINQ queries, which then get translated into efficient API requests.


 No Rating

Getting Started With ASP.NET 5 On Ubuntu

16. June 2015 22:59

.NET Framework ASP.NET ASP.NET MVC C# Microsoft Ubuntu Visual Studio Web 

Ever since the .NET stack went open source last year, there is a huge excitement among the developers about the .NET stuff and developing apps using .NET which are no longer limited to Windows platform. I tried to install ASP.NET VNext on Ubuntu VM in which I terribly failed in the first go. Why? because the tutorial I used was quite old and I messed up the installation of pre-requisites. But I get everything working in the second try. So here are the steps and commands that will get you started with ASP.NET VNext on Ubuntu.

I am setting up a fresh VM for development on Ubuntu 14.04.2 LTS

Installing Mono

First thing is to install Mono. For folks who are new to Linux environment, Mono is a community driven project which allows developers to build and run .NET application on Linux platforms. Here is the set of commands that I have to execute to install Mono.

sudo apt-key adv --keyserver --recv-keys 3FA7E0328081BFF6A14DA29AA6A19B38D3D831EF

echo "deb wheezy main" | sudo tee /etc/apt/sources.list.d/mono-xamarin.list
sudo apt-get update

Install the latest version of Mono available.

sudo apt-get install mono-complete

To check if Mono is successfully installed or to determine the version of Mono on you machine run the below command in the terminal.

mono --version

Installing LibUV

As stated on Github:

Libuv is a multi-platform asynchronous IO library that is used by the KestrelHttpServer that we will use to host our web applications.

Running the below command will install LibUV along with the dependencies require to build it.

sudo apt-get install automake libtool

Getting the source and building and installing it.

curl -sSL | sudo tar zxfv - -C /usr/local/src
cd /usr/local/src/libuv-1.9.0
sudo sh
sudo ./configure
sudo make 
sudo make install
sudo rm -rf /usr/local/src/libuv-1.9.0 && cd ~/
sudo ldconfig

Here is a note at Githb repo that explains what the above set of commands are doing.

NOTE: make install puts in /usr/local/lib, in the above commands ldconfig is used to update so that dlopen (see man dlopen) can load it. If you are getting libuv some other way or not running make install then you need to ensure that dlopen is capable of loading

Getting .NET Version Manager (DNVM)

DNVM is a command line tool which allows you to get new build of the DNX (.NET Execution Environment) and allows you to switch between them. To get DNVM running fire the below command in the terminal.

curl -sSL | DNX_BRANCH=dev sh && source ~/.dnx/dnvm/

To check if the DNVM is successfully installed on your machine, type DNVM in the terminal. The output should be something like this:

At any point of time if you want to list out the installed DNX runtimes, run the below command

dnvm list

The next step after this, is to upgrade the DNVM so you can use the dnx and dnu commands. Run the following command in the terminal

dnvm upgrade

Once this is done, we are all set to run ASP.NET VNext application on Ubuntu box. Clone the aspnet/Home repository from Github. If you don't have Git installed then install it with this simple command.

sudo apt-get install git

For simplicity, I have created a new directory on Ubuntu desktop named vnext. You can name the directory as you wish. Navigate to this directory in the terminal and clone the aspnet/Home repository.

git clone

After cloning of repository is done, navigate to the 1.0.0-beta4 directory.

You can see three sample applications that you can test. For this tutorial I am going to checkout HelloMvc application. Get inside the HelloMvc directory and then, run the command 

dnu restore

This will take some time to execute. I didn't face this problem but there is a chance that someone will. When you run this command, the project.json.lock file gets created and the restore of the package will start. In the end when the restore is finalizing, it may say permission is denied. To resolve this error you can change the permission of the folder by running the following command.

sudo chmod -R 755 HelloMvc

You should always change permission to 755 for directories and 644 for files.

After the execution is completed, you can start the server by running the command.

dnx . kestrel

This command will work for both web and mvc application. If you plan to test out the console application then you can run the following command.

dnx . run

The server runs at port 5004. Fire up the browser and type in http://localhost:5004/

Hope this is helpful for the first time users of Linux.

Currently rated 5.0 by 1 person

How Generic The Code Can Be - Resolving Type T at Compile Time

7. June 2015 08:07

.NET Framework C# 

It is fun working with Generics, but sometimes it just won't work as I intended it to work. While working on a personal project, I was trying to write the code as generic as possible so I can re-use the functions or extension methods in my other projects as well. But it is not always the case, things will not always work as you want it to be. Here is the case of what I was trying to do.

I am working extensively on JSON in this project, so I thought of creating an extension method for the classes I want to serialize to JSON and save it to the file system. Creating the extension method was not a problem, but it turned out to my surprise that I just can’t use the extension method on type T. If I would have paid attention to the detail here before writing the function, I would have noticed the problem. Before I tell you what I did here, here is the extension method which will serialize class to JSON string.

public static string ToJSON<T>(this T Entity) where T : class
    return JsonConvert.SerializeObject(Entity);

As I want only classes to be serialize, I have a constraint in place that assure me that this extension method should be used when the T is of type Class. I agree with the fact the creating an extension method just for a simple work is an over kill but I am just lazy about writing this line again and again. The application I am working on can support multiple databases and for that I am using Dependency Injection (DI). I wrote this function keeping in mind that I just have to pass the model name i.e. class object and this function will then save the JSON string from the serialized class to the file system.

public bool Save<T>(T Entity, string Path) where T : class
    bool IsSaved = false;
        if (!Path.IsNullOrEmpty())
            File.AppendAllText(Path, Entity.ToJSON());
            IsSaved = true;
        return IsSaved;
    catch (Exception x)
        throw x;

The above code will not compile. This is because, there is no way for the compiler to know the type of Entity object we are passing. The constraints has nothing to do with it. It is just the compiler who has to know what kind of type it is dealing with. In this case we have a type T which is not resolved at compile time and hence the ToJSON() extension method I am using above, is an assumption from my part that Entity is a class, is not going to work. The important point here is that T is going to resolve at runtime and before that if we try to use the extension method (meant only for class) on type T, the intellisense will never show the ToJSON() extension method in the list.

Now to overcome this problem, we have to make some changes to the above code in order to have the information which the compiler can use and allow us to make use of the ToJSON()extension method. As there is a constraint in place for this function which will only allow us to pass class object as a first parameter, but the case is still the same. The Save function will never know what is the actual type of the Entity. Therefore we can use the Entity object and get the name of the Class like so.

string className = Entity.GetType().Name;

This gives us the name of the class and now we can use it to get the actual type of T and convert it to that type (which will be class). Once we have a class we can then use our ToJSON() extension method. Here is the updated code.

public bool Save<T>(T Entity, string Path) where T : class
    bool IsSaved = false;
        if (!Path.IsNullOrEmpty())
            if (Entity.GetType().Name == "Board")
                File.AppendAllText(Path, Convert.ChangeType(Entity, typeof(Board)).ToJSON());
            else if (Entity.GetType().Name == "Story")
                File.AppendAllText(Path, Convert.ChangeType(Entity, typeof(Story)).ToJSON());

            IsSaved = true;
        return IsSaved;
    catch (Exception x)
        throw x;

This code will compile and saves the JSON string to the specified path on execution. Notice the Convert.ChangeType function. It’s simple function which you can use to convert one type to your desired type. The only drawback here will be that you have to come here and update the function whenever you have a type of class added in the project which you want to serialize to JSON. The code is still pretty sleek according to me for the type of task I am trying to accomplish.

The type T is very useful, but sometimes it may not solve all your problem.

 No Rating