Joe Buschmann

let topics = [csharp; specflow; fun]

A Short and Easy Introduction to .NET's Task Class


You can use Task.Run to schedule a delegate to run on the thread pool. The method returns a new task, and if the work is complete, the result will be available via Task.Result. If not, Task.Result will block until it is complete.


You'll want to avoid accessing the Task.Result property because it will block until the result is ready. In the previous example, the UI will hang for about 2000 milliseconds before updating the label with the result. You can fix this with Task.ContinueWith.

In this case Task.Result will not block because ContinueWith runs after the first task is complete; however, this code will throw an exception.

An exception of type 'System.InvalidOperationException' occurred in System.Windows.Forms.dll but was not handled in user code

Additional information: Cross-thread operation not valid: Control 'lblDateTime' accessed from a thread other than the thread it was created on.

You can only update the UI from the thread that created the form. Remember Control.Invoke in WinForms? Fortunately, you can fix it by creating an instance of TaskScheduler from the UI thread's SynchronizationContext and using it when invoking ContinueWith.

Awaiting Task.Run

If you're using C# 6.0, you can take advantage of async/await and await the task returned by Task.Run. The compiler will ensure the continuation runs on the originating thread, in this case the UI thread, so you don't have to create a TaskScheduler.


Task.Delay returns a task that will complete after a period of time. It can be awaited just like any other task.


TaskCompletionSource serves as a wrapper around a task whose lifetime must be managed explicitly. No delegate is provided. It's useful for bridging the gap between older asynchronous patterns and newer code based on tasks.

For example, let's say Task.Delay didn't exist, and we wanted a more efficient alternative to Thread.Sleep. We can use System.Threading.Timer, but it would be nice to wrap it in a task. TaskCompletionSource can help us do just that.

A real world example can be found in the codebase for Nancy FX, a lightweight web framework for .NET. Nancy can be hosted in a number of ways including in ASP.NET where an implementation of IHttpAsyncHandler invokes the task-based Nancy engine. IHttpAsyncHandler uses an older construct to implement asynchronous operations, and Nancy uses TaskCompletionSource to bridge the gap.


Task.FromResult returns a completed task prepopulated with a result. No code is run on a background thread. In fact, no work is done besides creating the new task and immediately setting the result.

This method provides a way for asynchronous interfaces to have synchronous implementations. I've found it useful for creating mock classes for tests. The mock doesn't do any blocking work but still needs to return a task as a result. Task.FromResult is perfect for this scenario.

Task.Run and ASP.NET

Using Task.Run for CPU-bound work makes sense in a desktop application. The long-running task executes on a background thread which keeps the UI thread free to respond to user input. The UI is more interactive leading to a better user experience.

But does it make sense in ASP.NET? In short, no. Task.Run will consume another thread from the ASP.NET process's thread pool resulting in two threads to service the same request.

There are inefficiencies even if you use async/await. The calling thread is recycled, but another one is needed to handle the task. That kind of churn in the thread pool is a poor use of resources.

For a more in depth explanation, see Stephen Cleary's series on Task.Run etiquette. The entire series is worth a read, but part three, Don't Use Task.Run in the Implementation, specifically addresses tasks in ASP.NET.