Visual Studio Magazine - A .NET 4.5 Parallel Dataflow Primer

"The Task Parallel Dataflow (TDF) library is built upon the existing Task Parallel Library (TPL) included in the .NET 4.0 Framework. Although the TPL provides a lot of functionality to help parallelize an application, it doesn't make it overly easy to tackle complex parallel use cases such as consumer/provider and agent-based models.

The TDF, on the other hand, provides higher-level abstractions in the form of  generic code blocks that generate and schedule the needed Task objects to handle simple to complex data flows.

The TDF is still in a preview state; it can be installed in Visual Studio 2012 RC through NuGet, as shown in Figure 1.

...

Once the package is installed, you should see a reference to System.Threading.Tasks.Dataflow in your project, as seen in Figure 2.

In the TDF, all blocks implement the common IDataflowBlock interface. The IDataflowBlock interface contains  Complete and Fault methods, and a Completion property getter. The Complete method is called upon successful completion of the dataflow block's task. The Fault method is invoked if an exception occurs and the Completion property returns the Task that will be executed asynchronously by the dataflow block.

All the dataflow blocks also implement either the ISourceBlock<T> or ITargetBlock<T> interface, or both. The ISourceBlock<T> interface defines a contract for offering data, whereas the ITargetBlock<T> interface defines an interface for receiving data.

The ActionBlock
I'll begin with the ActionBlock, which implements the ITargetBlock<T> interface. In turn, the ISourceBlock and ITargetBlock interfaces both inherit the base IDataflowBlock interface. Later I'll go over the TransformBlock  that implements both the ISourceBlock and ITargetBlock interfaces

..."