A .NET Programmer’s Information to CancellationToken

Typically canceling is an efficient factor. In a lot of my .NET initiatives, I’ve had loads of motivation to cancel each inside and exterior processes. Microsoft realized that builders have been approaching this frequent use case in a wide range of complicated implementations and determined there have to be a greater approach. Thus, a typical cancellation communication sample was launched as CancellationToken, which was constructed utilizing lower-level multithreading and interprocess communication constructs. As a part of my preliminary analysis into this sample—and after having dug by means of the precise .NET supply code for Microsoft’s implementation—I discovered that CancellationToken can resolve a much wider set of issues: subscriptions on functions’ run states, timing out operations utilizing completely different triggers, and common interprocess communications through flags.

The Meant CancellationToken Use Case

CancellationToken was launched in .NET 4 as a way to reinforce and standardize the prevailing options for canceling operations. There are 4 common approaches to dealing with cancellation that fashionable programming languages are inclined to implement:

CancellationToken resides within the closing class, the place the cancellation dialog is cooperative.

After Microsoft launched CancellationToken, the event neighborhood shortly embraced it, notably as a result of many main .NET APIs have been up to date to make use of these tokens natively. For instance, starting with ASP.NET Core 2.0, actions assist an optionally available CancellationToken parameter that will sign if an HTTP request has been closed, permitting cancellation of any operation and thus avoiding unnecessary use of sources.

After a deep dive into the .NET codebase, it turned clear that CancellationToken’s utilization isn’t restricted to cancellation.

CancellationToken Below a Microscope

When trying extra intently at CancellationToken’s implementation, we see it’s only a easy flag (i.e., ManualResetEvent) and the supporting infrastructure that gives the flexibility to watch and alter that flag. CancellationToken’s most important utility is in its identify, which suggests that is the frequent method to cancel operations. These days, any .NET library, bundle, or framework with asynchronous or long-running operations permits cancellation by means of these tokens.

CancellationToken could also be triggered both by manually setting its flag to “true” or programming it to alter to “true” after a sure time span has elapsed. No matter how a CancellationToken is triggered, consumer code that’s monitoring this token could decide the token flag’s worth by means of one in every of three strategies:

• Utilizing a WaitHandle
• Polling the CancellationToken’s flag
• Informing the consumer code when the flag’s state is up to date by means of a programmatic subscription

After additional analysis within the .NET codebase, it turned evident that the .NET workforce discovered CancellationTokens helpful in different eventualities not related to cancellation. Let’s discover a few of these superior and off-brand use instances, which empower C# builders with multithreaded and interprocess coordination to simplify complicated conditions.

CancellationTokens for Superior Occasions

When writing ASP.NET Core functions, we typically have to know when our software has began, or we have to inject our code into the host shutdown course of. In these instances, we use the IHostApplicationLifetime interface (beforehand IApplicationLifetime). This interface (from .NET Core’s repository) makes use of CancellationToken to speak three main occasions: ApplicationStarted, ApplicationStopping, and ApplicationStopped:

namespace Microsoft.Extensions.Internet hosting
{
    /// <abstract>
    /// Permits shoppers to be notified of software lifetime occasions. 
    /// This interface isn't supposed to be user-replaceable.
    /// </abstract>
    public interface IHostApplicationLifetime
    {
        /// <abstract>
        /// Triggered when the appliance host has absolutely began.
        /// </abstract>
        CancellationToken ApplicationStarted { get; }

        /// <abstract>
        /// Triggered when the appliance host is beginning a sleek shutdown.
        /// Shutdown will block till all callbacks registered on 
        /// this token have accomplished.
        /// </abstract>
        CancellationToken ApplicationStopping { get; }

        /// <abstract>
        /// Triggered when the appliance host has accomplished a sleek shutdown.
        /// The appliance won't exit till all callbacks registered on 
        /// this token have accomplished.
        /// </abstract>
        CancellationToken ApplicationStopped { get; }

        /// <abstract>
        /// Requests termination of the present software.
        /// </abstract>
        void StopApplication();
    }
}

At first look, it might look like CancellationTokens don’t belong right here, particularly since they’re getting used as occasions. Nonetheless, additional examination reveals these tokens to be an ideal match:

• They’re versatile, permitting for a number of methods for the interface’s consumer to hear to those occasions.
• They’re thread-safe out of the field.
• They are often created from completely different sources by combining CancellationTokens.

Though CancellationTokens aren’t good for each occasion want, they are perfect for occasions that occur solely as soon as, like software begin or cease.

CancellationToken for Timeout

By default, ASP.NET offers us little or no time during which to close down. In these instances the place we would like slightly extra time, utilizing the built-in HostOptions class permits us to alter this timeout worth. Beneath, this timeout worth is wrapped in a CancellationToken and fed into the underlying subprocesses.

IHostedService’s StopAsync methodology is a good instance of this utilization:

namespace Microsoft.Extensions.Internet hosting
{
    /// <abstract>
    /// Defines strategies for objects which can be managed by the host.
    /// </abstract>
    public interface IHostedService
    {
        /// <abstract>
        /// Triggered when the appliance host is able to begin the service.
        /// </abstract>
        /// <param identify="cancellationToken">Signifies that the beginning
        ///     course of has been aborted.</param>
        Process StartAsync(CancellationToken cancellationToken);

        /// <abstract>
        /// Triggered when the appliance host is performing a sleek shutdown.
        /// </abstract>
        /// <param identify="cancellationToken">Signifies that the shutdown 
        ///     course of ought to now not be sleek.</param>
        Process StopAsync(CancellationToken cancellationToken);
    }
}

As evident within the IHostedService interface definition, the StopAsync methodology takes one CancellationToken parameter. The remark related to that parameter clearly communicates Microsoft’s preliminary intent for CancellationToken was as a timeout mechanism fairly than a cancellation course of.

For my part, if this interface had existed previous to CancellationToken’s existence, this might have been a TimeSpan parameter—to point how lengthy the cease operation was allowed to course of. In my expertise, timeout eventualities can virtually at all times be transformed to a CancellationToken with nice further utility.

For the second, let’s neglect that we all know how the StopAsync methodology is designed and as a substitute take into consideration how we’d design this methodology’s contract. First let’s outline the necessities:

• The StopAsync methodology should attempt to cease the service.
• The StopAsync methodology ought to have a sleek cease state.
• No matter whether or not a sleek cease state is achieved, a hosted service will need to have a most time during which to cease, as outlined by our timeout parameter.

By having a StopAsync methodology in any kind, we fulfill the primary requirement. The remaining necessities are tough. CancellationToken satisfies these necessities precisely through the use of an ordinary .NET flag-based communication software to empower the dialog.

CancellationToken As a Notification Mechanism

The most important secret behind CancellationToken is that it’s only a flag. Let’s illustrate how CancellationToken can be utilized to begin processes as a substitute of stopping them.

Think about the next:

1. Create a RandomWorker class.
2. RandomWorker ought to have a DoWorkAsync methodology that executes some random work.
3. The DoWorkAsync methodology should permit a caller to specify when the work ought to start.

public class RandomWorker
{
    public RandomWorker(int id)
    {
        Id = id;
    }

    public int Id { get; }

    public async Process DoWorkAsync()
    {
        for (int i = 1; i <= 10; i++)
        {
            Console.WriteLine($"[Worker {Id}] Iteration {i}");
            await Process.Delay(1000);
        }
    }
}

The above class satisfies the primary two necessities, leaving us with the third. There are a number of alternate interfaces we might use to set off our employee, like a time span or a easy flag:

# With a time span
Process DoWorkAsync(TimeSpan startAfter);

# Or a easy flag
bool ShouldStart { get; set; }
Process DoWorkAsync();

These two approaches are fantastic, however nothing is as elegant as utilizing a CancellationToken:

public class RandomWorker
{
    public RandomWorker(int id)
    {
        Id = id;
    }

    public int Id { get; }

    public async Process DoWorkAsync(CancellationToken startToken)
    {
        startToken.WaitHandle.WaitOne();

        for (int i = 1; i <= 10; i++)
        {
            Console.WriteLine($"[Worker {Id}] Iteration {i}");
            await Process.Delay(1000);
        }
    }
}

This pattern consumer code illustrates the ability of this design:

utilizing System;
utilizing System.Linq;
utilizing System.Threading;
utilizing System.Threading.Duties;

namespace CancelToStart
{
    public class Program
    {
        static void Major(string[] args)
        {
            CancellationTokenSource startCts = new CancellationTokenSource();

            startCts.CancelAfter(TimeSpan.FromSeconds(10));

            var duties = Enumerable.Vary(0, 10)
                .Choose(i => new RandomWorker(i))
                .Choose(employee => employee.DoWorkAsync(startCts.Token))
                .ToArray();

            Process.WaitAll(duties, CancellationToken.None);
        }
    }
}

The CancellationTokenSource will create our CancellationToken behind the scenes and coordinate the triggering of all of the related processes. On this case, the related course of is our RandomWorker, which is ready to begin. This strategy permits us to leverage the thread security baked into the default CancellationToken implementation.

These examples reveal how CancellationToken supplies a toolbox of options which can be helpful exterior of its supposed use case. The instruments can turn out to be useful in lots of eventualities that contain interprocess flag-based communication. Whether or not we’re confronted with timeouts, notifications, or one-time occasions, we will fall again on this elegant, Microsoft-tested implementation.

Additional Studying on the Toptal Engineering Weblog: