(How)
C^# You Are - 18 November 2007

It is important that developers stay ahead of technology in order to remain viable in the field.  This week's questions relate to the newer v3.5 technologies that will be available at the end of November (the beginning of the year for most developers).  I'm also jetlagged so I'll try to keep the questions short.

1. You have an existing v2 application using WinForms and ADO.NET.  You'd like to add some new features to your application that require v3.5 but you don't want to have to migrate your existing code.  What can you do?  Answer

   Although it is known as v3.5, it is actually an incremental extension to v2 and therefore backwards compatible with the existing code.  When MS first introduced WPF and WCF to the world it was in v3 of .NET.  This was a poor naming convention as v3 was simply a few additional assemblies that extended the existing v2 code.  The core assemblies weren't even installed as part of v3.  v3.5 is, therefore, an iterative version of v3 (which itself was an iterative version of v2).  So, in summary, you can simply recompile your code against v3.5 and take advantage of the v3.5 classes while continuing to use the existing v2 code.

2. What is the difference between WPF and WinForms?  Answer

   While they accomplish the same thing, displaying a UI, they go about it in different manners.  Which one is better is open for negotiation.  WinForms is based off the traditional Windows UI programming model where each window is responsible for displaying itself.  Each window has complete control over its own UI.  Within WinForms the UI is built completely in code.  There is no resource or layout file that controls how the UI is built. 

   WPF is a newer technology that alters some fundamental rules of Windows uI development.  Specifically WPF separates the UI layout from the behavior to allow more dynamic UIs.  For example a button can be displayed as a standard Windows button, an image, some text or even a movie.  As a result the UI of an application can be controlled in a central location and enforced across an entire application with little effort.  In WinForms you would have to hook into each control to make them look the same.  This is a blessing and a curse.  Windows became wildly popular partially because the UI was consistent across applications.  With WPF applications can have completely different UIs.  Whether this becomes an issue or not only time will tell.

   Another fundamental difference between WPF and WinForms is how windows are stored.  With WPF, XML (stored as XAML) is used to represent the hierarchy of controls and their properties.  Ultimately code is still used to build the UI but by using XAML it is easier to persist windows.

3. What is a binding in WCF?  Answer

   A binding determines how two endpoints will communicate.  The binding determines the protocol used, the security available and encoding schemes.  Each binding has its own advantages and disadvantages.  For example the BasicHttp binding is good for web service calls as it can go through the firewall but is slower than, say, the NetNamedPipe binding which works only on the local machine.

   Determining the binding(s) to use is fundamental to working with WCF.

4. What is a dependent property?  Answer

   Dependent properties were added for WPF.  A dependent property is dependent upon other values for its value.  A fundamental problem with existing .NET properties is that it is hard to know when property values change (although there is the INotifyProperty interface).  With dependent properties validation, change detection and management are handled by the system.  Dependent properties, unlike normal properties, are not backed by a field but rather are managed by the system.  This allows the system to manage the dependencies directly rather than requiring developers to write code. 

   In WPF animation, styles and data binding all use dependent properties.

5. What is a WPF template?  Answer

   A template in WPF is a set of default values for a control's properties.  Generally templates are used for ensuring all controls have the same layout such as color or font settings.  Templates can be used for anything though.  Like skins in ASP.NET templates can be overridden as needed by a UI element.

   In WPF applications, most UI style settings are handled through styles defined at the application level.  Templates are then used to control the behavior of the control.

6. What is a workflow activity?  Answer

   An activity in WWF is a unit of work in a workflow.  An example of an activity might be copying a file or reading data from a database.  An activity can be a single operation or a collection of them.  WWF ships with a common set of activities and you can create your own.  It is expected that, over time, a large set of activities will be available through the community.

7. What is an extension method?  Answer

   An extension method is a method that extends an existing type.  Extension methods are convenience routines only and are not required for programming.  In fact extension methods are not even supported by the CLR directly.  Instead extension methods are implemented through the compiler.

   In .NET a type can not be modified outside the assembly that defines it.  As a result many "helper" classes are created to extend a type (StringHelper for strings and FormHelper for forms, for example).  These helper classes are often static classes with static methods that provide new functionality to existing types.  Since this paradigm is so useful v3.5 (technically the languages updated for v3.5) added extension methods.  An extension method is nothing more than a static method of a class that contains the type that it is extending as the first parameter.  The following code might be something you'd see in v2 today.

   public static class StringHelper
   {
      public static string GetErrorMessage ( string message )
      { return message.Split(':', 2)[1]; }
   }

   Here's how we'd use it:

   string strMsg = "An error occurred: Access is denied.";
   string strErr = StringHelper.GetErrorMessage(strMsg);
   

   Not elegant but it works.  In v3.5 we can add the Extension attribute to the method and now it is an extension method.  To call it we would do the following.

   string strMsg = "An error occurred: Access is denied.";
   string strErr = strMsg.GetErrorMessage();
   

   Internally the compiler simply extends the lookup for instance methods to include searches for extension methods.  When an extension method is used it replaces the existing instance method call with a call to the static method.  You can verify this by dumping the code using Reflector.

   Extension methods are most useful with LINQs where existing collection types are extended to support LINQs style commands.  Nevertheless it is useful elsewhere as well.  Care must be taken that this feature is not misused.  Using extension methods in lieu of proper class design (when you control the source) is going to cause confusion and lead to poor type designs.