Thursday, October 25, 2012

[Tutorial] Test Driven Development with Visual Studio 2012
Part4: Full cycle of TDD using Visual Studio 2012

The last blog post in the series introduced some of the new features within Visual Studio 2012, applicable to Test Driven Development (TDD). This blog post is going to show you how to practically use all of those new features and apply them to Test Driven Development. You are going to see a full cycle of TDD using Visual Studio 2012, during the implementation of a simple calculator example!

Phase 1: Write a unit test for a new functionality  

Respecting the Test Driven Development approach, you have to create your unit tests before starting with any implementations. In our example we have to write some unit tests for the methods “Addition” and “Multiplication”. But first of all, we have to add a project of type “Unit Test Project” to our solution (if you do not already have one).


You may now add the necessary unit tests.Here is an example of the unit tests you could add based on the unit test framework NUnit:


Please note that the “Calculator” class and its methods “Addition” and “Multiplication” do not exist at this stage yet.

Visual Studio 2012 provides the possibility to generate the missing code in an automatic way. For that you just have to right-click on the “new Calculator” definition in the unit test project and choose to generate the class via the “Generate/New Type“ option in the menu.


A wizard opens and you are now able to configure multiple options such as the type (classe, struct, interface, enum), the access (public, internal), the destination project and the file name for the generation of the missing class.


The next step, after having auto-generated the “Calculator” class, consists of auto-generating the missing methods within this class. This can be achieved in almost the same way as it was done for the missing class. You do a right-click on the method calls "calculator.Addition(…)” and “calculator.Multiplication(…)” in the unit test project and you generate them via the “Generate/Method Stub” option in the menu.

Here is the auto-generated source code of those two methods:


In the last step of this phase you have to open the “Test Explorer” window where you may now execute all your unit tests. This can be done by clicking on the “RunAll” button or by using the already explained “Post Build Test Runs” option (see the previous blog post in the series).

As expected your unit tests will fail, since the corresponding source code has not been implemented yet. We will see how to do that in the next phase.


Phase 2: Implement the minimum code necessary to pass the test

Now in this phase, the only thing that needs to be done, is to implement the expected functionalities. The idea is to develop the minimum code necessary, which responds to the functional requirements. Everything that concerns optimization and amelioration must not be addressed since it will be treated later in the next phase (refactoring).


Following the implementation you may now restart your unit tests by clicking on the “RunAll” button or by using the already explained “Post Build Test Runs” option (see the previous blog post in the series). Your should see that you unit tests have been terminated successfully. If this is not the case you have to review your code and iterate until all of your unit tests pass successfully.


At this stage the source code corresponds exactly to the functional needs and it provides the expected behavior. The final project structure includes a unit test project as well as an application implementation project.


But the source code might not be optimized. Its quality might not adhere to your quality standards, so it has to be ameliorate. The refactoring can be done without any problems since the unit test assure that there are no regressions. Moreover, regressions can be detected very quickly and thus can be handled as soon as possible. This is going to be explained in the next phase.

Phase 3: Refactor and optimize the source code

The process of improving your source code after the initial implementation phase (Phase 2)  is called “Refactoring”. The source code structure is modified internally without any modifications to the external behavior (very important!!). A source code that just “works” is now transformed into a source code that works in an optimal way. Most of the time, the resulting source code is executing with better performance, using less memory and/or with a better software design.

The refactoring consists of the following steps (non-exhaustive list):

  • Detect and eliminate all code duplication
  • Limit complexity and the number of classes
  • Simplify and optimize method algorithms
  • Relocate, rename and harmonize methods
  • Improve code readability
  • Remove not used code (also called “dead code”)
  • Add comments to complex code sections

In our simple example there is nothing to be refactored, since there are neither enough methods nor enough classes. But this last step has to be done in bigger developments at the end of each cycle. Afterwards, a new development cycle starts with  new functionalities from Phase1 on.


Wednesday, October 24, 2012

[Tutorial] Test Driven Development with Visual Studio 2012
Part3: New features in Visual Studio 2012 (2/2)

The last blog post in the series introduced some of the new features within Visual Studio 2012, applicable to Test Driven Development (TDD). This blog post is going to continue to further present some more features of Visual Studio 2012 in this domain.

Management of unit tests via Test Explorer

The first impression when opening the “Test Explorer” window is a very positive one. Here are the principal changes when compared to Visual Studio 2010:

  • The “Test View” and “Test Results” windows have been deleted and consolidated into the “Test Explorer” window. This is going to streamline the interaction between development (the code) and tests.
  • The interface is simple but efficient: all information is accessible via a simple mouse click in a very intuitive way.
  • Unit tests are grouped by their status (failed, passed,…), failed tests are shown on top, a double mouse click allows for accessing the code source of a test (no need to open an external window anymore).
  • It is now much easier to execute a code coverage analysis. In previous versions this was not handled in a very intuitive way, since you had to create a configuration file, start the analysis via the Visual Studio menu and then open the adequate results window. In Visual Studio 2012 everything was consolidated and is now integrated in the “Test Explorer” interface.

Post Build Test Runs

A best practice when adhering to TDD principles is to execute unit tests as soon and as much as possible for being able to identify bugs, misbehaviors and regressions. There is a new feature, called “Post Build Test Runs”, in Visual Studio 2012, which allows for automatically unit test execution after each compilation. The feature can either be activated via the menu under “Test/Test Settings” or directly from within the “Test Explorer” window. When using this features, unit test are executed on a separated and dedicated thread, so there is no impact on developer efficiency.


Migration of unit tests from Visual Studio 2010 to Visual Studio 2012

As you might know, there a multiple difficulties and bugs, when migrating unit tests from Visual Studio 2008 to Visual Studio 2010, because the migration is not very transparent and sometimes even somewhat complex. In some cases you even have to migrate to the .NET 4.0 Framework to make everything work correctly!

I can assure you that there are no such migration problems, when trying to migrate from Visual Studio 2010 to Visual Studio 2012. This is partly due to the fact that most of the unit test components are the same between those two versions of Visual Studio. The library is still Microsoft.VisualStudio.QualityTools.unitTestFramework.dll in its version, which is still based on .NET runtime v2.0.50727.

Fakes Framework (Stubs and Shims)

The “Fakes Framework” based on the  “Moles” project, created by the Microsoft Research team, is now exclusively integrated into the “Ultimate” version of Visual Studio 2012 (and sadly only in this version!).

The goal of this framework is to aid development teams in producing unit tests rapidly and easily. For this, the “Fakes Framework” adds 2 notions:

  • Stubs: they provide automatic mock implementations of interfaces or abstract classes, that can be used in unit tests for being able to isolate parts that need to be unit tested.
  • Shims: they allow for runtime interception and redirection of method calls to specific objects. For example , they may be used to mock objects, which normally can’t be mocked due to access restrictions in the .NET framework. By using Shims it is possible to redirect calls to these objects with calls to objects that provide your own implementations.

There are however some restrictions. One restriction is that classes included in the “mscorlib” namespace cannot have any “Fake Assemblies”. Unfortunately, you cannot create any Shims for the “System.Configuration.ConfigurationManager” class for example.

The “Fakes Framework” provides some real advantages, when compared to existing mock testing frameworks such as RhinoMock, because developers do not need to modify their functional code for being able to execute unit tests.

In my next blog post I am going to show you how to practically use all of those new features and apply them to Test Driven Development. You are going to see a full cycle of TDD using Visual Studio 2012 !


Wednesday, October 17, 2012

[Tutorial] Test Driven Development with Visual Studio 2012
Part2: New features in Visual Studio 2012 (1/2)

The last blog post in the series introduced Test Driven Development (TDD). This blog post is going to talk about some of the new features in Visual Studio 2012, that allow you to successfully apply the TDD approach in your daily work.

Visual Studio 2012 external test framework support

The previous version Visual Studio 2010 already provides the possibility of using external test frameworks. But unfortunately with multiple limitations such as:

  • Dependency on third party applications for being able to execute unit tests. For example Gallio needs Icarus Runner if you want to run your unit tests.
  • Code coverage analysis does not work natively, only by using third party products (such as NCover for example).
  • Execution of unit tests is specific to certain plugins : for example there is a difference if you run your MSTest unit tests from within Visual Studio 2008 or by using Resharper, which may result in problems when trying to apply continuous integration processes.

As you can see, you have to multiply plugins and tools, that might not always be compatible with each other, when using Visual Studio 2010 with external test frameworks. Really not an optimal situation, when you want to benefit from specialized external test frameworks and their advanced functionalities !

One of the major updates of Visual Studio 2012 is the support of several external unit test frameworks for multiple languages without any of the limitations mentioned above.

Here are some of the unit test frameworks that you can now use easily:

For .NET :

For Javascript/HTML :

For C++ :

How to use NUnit as external test unit framework 

Lets see how to use NUnit as external test unit framework. First you have to install NUnit (currently as Prerelease) via NuGet.


Then you have to install the test adapter plugin for NUnit (currently as Beta) via the Extension Manager, which you can now find under “Tools/Extensions and Updates…”. Note that all test adapters are free of charge and that the functionality to search and download those adapters is fully integrated in Visual Studio 2012.



After installing the NUnit test adapter plugin, you are able to use all Visual Studio 2012 unit test functionalities together with NUnit. You may for example execute your NUnit unit tests directly from within the Visual Studio 2012 IDE, display the test results in the standard view “Test Explorer” and run a code coverage analysis on your code.


Support for C++ in native MSTest

Another big feature of Visual Studio 2012 is that it now supports native MSTest unit tests for C++ applications. Good news for C++ developers, who may now also apply Test Driven Development using native MSTest as unit test framework.


Following is an example implementation of a unit test in Visual C++ using the native MSTest unit test framework:


Support for async and await in VS 2012 and its unit test framework

Windows 8 and .NET 4.5 introduce a major feature that will have a high impact on software development as we know it: Asynchronous Programming. Visual Studio 2012 and its unit test framework support this new approach perfectly. The async and await keywords, available in .NET 4.5, may now be used for creating unit tests of asynchronous methods.

Lets say that you want to create a unit test for the following asynchronous example method:

16-Oct-12 16-49-01

The corresponding NUnit test method could be:

16-Oct-12 17-09-08

In my next blog post I am going to further show you other new features of Visual Studio 2012, which will make your life easier when trying to apply Test Driven Development.


Thursday, October 11, 2012

[Tutorial] Test Driven Development with Visual Studio 2012
Part1: Introduction to TDD

In one of my blog posts series, I showed you how to increase the quality of your code and how to make it much easier to maintain by applying the S.O.L.I.D. principles. I then focused on the Liskov Segregation Principle (LSP) by using Code Contracts in .NET 4.0. This blog series is going to talk about Test Driven Development (TDD) in an Agile environment using the latest Visual Studio 2012 features and how that is going to help you to be more efficient and productive.

The goal of this series will be to show you some of the new features within Visual Studio 2012, that are going to help you when trying to apply the TDD approach. The first parts of the series are more theoretical, while as the last part will be fully practical, since we are going to implement some example code using TDD.

Contrary to traditional software development methodologies and processes, where writing unit test is often done as last step after code implementation, TDD consists of writing unit test upfront as first step in the software development cycle. No development can be started without having implemented the corresponding unit tests. You understand that TDD has large impacts on the organization of your teams.

The following are the different steps when applying Test Driven Development :


  1. Write a unit test for a new functionality, then validate that the test fails, since its implementation has not been realized
  2. Implement the minimum code necessary to pass the test, then validate that the test passes successfully, meaning that the expected behavior is now provided by the implementation
  3. Refactor and optimize your code, the unit tests ensure the coherence of the functionalities

Using TDD helps in obtaining a code source that is very reliable, predictable, robust - and after refactoring - highly optimized. Unit tests ensure a correct behavior, independently from where it is going to be used, resulting in a code source that will work as expected under any circumstances.

For being able to create good unit tests, you have at first to think about the conception and software design of your application. You must not hurry into doing your implementations, before being clear about your objectives (which should be true for any development, but sadly often it is not).

Conceptual errors can be detected and resolved much quicker and in a much more efficient way. As explained the implementation is started only after the conceptual phase has been validated and tested via thorough unit tests. In this case, your unit tests become much than just tests. They become some sort of general specification, describing  “units of functionality” for your application.

When refactoring and optimizing your code, you may restructure your code without any risks, since all your modifications are now verifiable. Unit tests ensure that there are no technical or functional regressions and furthermore the coherence of behavior. They validate that your application will always behave in the same way, if they are executed with success.

Additionally if you combine TDD with Extreme Programming (XP) and pair programming, you obtain a code with a very high degree of quality.

The next blog post in the series is going to talk about some the new features in Visual Studio 2012, that allow you to apply TDD quickly and easily, stay tuned!


Monday, October 1, 2012

[Publication] Article in French Programmez Magazine on Window 8 Development and how to build your first application

You can find an article of 3 pages on how to start developing applications for  Windows 8 in the French Programmez magazine No.156 written by Jonathan Pamphile and me.

First Page and Second Page (low resolution)

Third Page (low resolution)

The article is written in French but as always I will write some English articles on my Blog in the next weeks. So stay tuned if you are interested in getting to know how to start programming application for Windows 8 and publish them on the Windows Store.