Dernière version du 25 novembre 2024 à 11:18

Links

Task

A Task represents an asynchronous operation.

// create and run a task in a new thread
var result = await Task.Run(async () => 
{
    await Task.Delay(4000);
    return 0;
});

await vs Result

await	Result
asynchronous wait	blocking wait
"yield" the current thread	block the current thread
(re-)raise the exception	wrap the exception in an AggregateException

WhenAll

var inputAndTask1 = new { Input = 1, Task = GetDataAsync(1) };
var inputAndTask2 = new { Input = 2, Task = GetDataAsync(2) };
var inputAndTask3 = new { Input = 3, Task = GetDataAsync(3) };

await Task.WhenAll(task1.Task, task2.Task, task3.Task);

// Result allow to acces to the result of the task without blocking because the WhenAll ensure all the tasks are finished
Console.WriteLine($"Input: {inputAndTask1.Input} - Output: {inputAndTask1.Task.Result}"));
Console.WriteLine($"Input: {inputAndTask2.Input} - Output: {inputAndTask2.Task.Result}"));
Console.WriteLine($"Input: {inputAndTask3.Input} - Output: {inputAndTask3.Task.Result}"));

async Task<int> GetDataAsync(int i)
{
    await Task.Delay(2000);
    Console.WriteLine($"{DateTime.Now} - GetDataAsync - {i}");
    return i * 2;
}

var inputAndTasks = Enumerable.Range(1, 10)
    .Select(i => new { Input = i, Task = GetDataAsync(i) })
    .ToList();

await Task.WhenAll(inputAndTasks.Select(x => x.Task));

// Result allow to acces to the result of the task without blocking because the WhenAll ensure all the tasks are finished
inputAndTasks.ForEach(x => Console.WriteLine($"{x.Input} - {x.Task.Result}"));

var inputs = Enumerable.Range(1, 10).ToArray();

var outputs = await Task.WhenAll(inputs.Select(GetDataAsync));

// as the order is the same, you can Zip
var inputAndOutputs = inputs
    .Zip(outputs, (input, output) => (input, output))
    .ToArray();

inputAndOutputs.ForEach(x => Console.WriteLine($"{input} - {output}"));

Cancel a task

var cancellationTokenSource = new CancellationTokenSource(5000); // cancel after 5s
CancellationToken token = cancellationTokenSource.Token;

try
{
    await Task.Run(async () =>
    {
        var i = 1;
        while (!token.IsCancellationRequested) // break the loop if cancellation is requested
        {
            Console.Write($"{i++} ");
            await Task.Delay(1000);
        }

        token.ThrowIfCancellationRequested(); // or throw an OperationCanceledException
    }, token);
}
catch (OperationCanceledException e)
{
    Console.WriteLine(e.Message);
}

cancellationTokenSource.Token.Register(() => {
    /* what to when the token is cancelled */
});

Fire and forget

Call an async method without waiting for the response. Exceptions will be lost.

// without await, the method is called and the following lines of code are executed without waiting the end of MyMethodAsync
MyMethodAsync();

// without await, the code is executed in background
// you may need to create a scope if you want to consume scoped services
Task.Run(async () =>
{
    Method1();
    await Method2Async(); // wait the end of Method2Async before calling Method3
    Method3();
});

MyMethodAsync().Forget();

async Task MyMethodAsync() {}

public static class TaskExtension
{
    public static void Forget(this Task task)
    {
        if (!task.IsCompleted || task.IsFaulted)
        {
            _ = ForgetAwaited(task);
        }

        async static Task ForgetAwaited(Task task)
        {
            await task.ConfigureAwait(ConfigureAwaitOptions.SuppressThrowing);
        }
    }

    // simple version: returns void
    public static async void Forget(this Task task)
    {
        await task.ConfigureAwait(false);  // ConfigureAwait to avoid deadlock
    }
}

Propriété WPF

Nuget:

Nito.Mvvm.Async prerelease
FontAwesome.WPF

MainWindow.xaml

MainVM.cs

Parallel

Utile si le code n'est pas séquentiel.

// for loop
Parallel.For(1, 20, i =>
{
    Console.WriteLine(i);
    Thread.Sleep(1000);
});

// for each loop
Parallel.ForEach(Enumerable.Range(1, 20), i =>
{
    Console.WriteLine(i);
    Thread.Sleep(1000);
});

// invoke actions
Parallel.Invoke(
    () => {
        Console.WriteLine(1);
        Thread.Sleep(1000);
    },
    () => {
        Console.WriteLine(2);
        Thread.Sleep(1000);
    }
);

Parallel options

// after 4s throw an OperationCanceledException
// no further operations will start but don't stop currently executing operations
var cancellationTokenSource = new CancellationTokenSource(4000);

var parallelOptions = new ParallelOptions
{
    MaxDegreeOfParallelism = 12, // by default use as much computer power as possible
    TaskScheduler = null,
    CancellationToken = cancellationTokenSource.Token
}

Parallel.ForEach(
    numbers,
    parallelOptions,
    (int i, ParallelLoopState loopState) =>
{
    if (loopState.ShouldExitCurrentIteration) // check if another iteration has requested to break
    {
        loopState.Break(); // break loop
    }

    if (!cancellationTokenSource.Token.IsCancellationRequested) { /* next operation step */ } // useful for long operation to break
});

Handling exceptions

All the exceptions are catched and when all the tasks have been executed then an AggregateException is thrown if any.

try
{
    Parallel.Invoke(
        () =>
        {
            var waitTime = DateTime.UtcNow.AddSeconds(4);
            while (DateTime.UtcNow < waitTime) { }
        },
        () =>
        {
            throw new Exception("MyException");
        }
    );
}
catch (AggregateException ex)
{
    ex.InnerExceptions; // ReadOnlyCollection<Exception>
}

Shared variable

Lock

private readonly object sumLock = new();

var sum = 0m; // shared variable, updated by threads
Parallel.For(0, 100, i =>
{
    lock(sumLock) // only 1 thread at a time can access
    {
        sum += 0.5m; // code inside the lock should take as little time as possible
    }
});

To avoid deadlocks:

use 1 lock object for each shared resource
avoid nested locks
use a new object

Interlocked

Create thread-safe atomic operations.

Faster than lock, but Interlocked only works with integers.

int sum = 0; // shared variable, updated by threads
Parallel.For(0, 100, i =>
{
    Interlocked.Increment(ref sum); // add 1 to sum and return sum + 1
    Interlocked.Add(ref sum, 2);    // add 2 to sum and return sum + 2
});

AsyncLocal

Allow to have a different variable for each async task.

private static AsyncLocal<decimal?> asyncLocal = new();
Parallel.For(0, 100, async (i) =>
{
    asyncLocal.Value = 10; // the asyncLocal is not shared among async tasks
});

Concurrent collections

BlockingCollection<T>	ajout et suppression thread-safe. Add, Take. FIFO par défaut.
ConcurrentBag<T>	sans ordre, doublons autorisés. Add, TryTake, TryPeek.
ConcurrentDictionary<TKey,T>	TryAdd, TryUpdate, AddOrUpdate, GetOrAdd.
ConcurrentQueue<T>	FIFO. Enqueue, TryDequeue.
ConcurrentStack<T>	LIFO. Push, TryPop.

BlockingCollection<string> col = new BlockingCollection<string>();
col.Add("text");
string s = col.Take();

foreach (string v in col.GetConsumingEnumerable())
    Console.WriteLine(v);

Tasks

// AsParallel PLINQ
var tasks = Enumerable.Range(1, 30).AsParallel().Select(x => MyTaskAsync(x));
await Task.WhenAll(tasks);

// ForEachAsync
await Parallel.ForEachAsync(Enumerable.Range(1, 30), async (x, token) => await MyTaskAsync(x));

// Task.Run run task in a new thread
// even without Task.Run, Task.WhenAll will run the tasks in parallel and then for all of them to be done
var tasks = Enumerable.Range(1, 30).Select(x => Task.Run(() => MyTaskAsync(x)));
await Task.WhenAll(tasks);

private async Task MyTaskAsync(int i)
{
    Console.WriteLine(i);
    await Task.Delay(4000);
}

Handle exceptions with Task.WhenAll

The problem

If at least one exception occurred among the tasks, it is not possible to get the result of the working tasks.

// create 10 jobs in parallel, the jobs 5 and 7 will raise an Exception
IEnumerable<Task<Job>> createJobTasks = Enumerable.Range(1, 10).Select(x => CreateJobAsync($"Job {x}"));
Task<Job[]> mainTask = Task.WhenAll(createJobTasks);
try
{
    var jobs = await mainTask;
}
catch (Exception e)
{
    e.Message; // Error Job 5, only the first exception is catched
    // AggregateException
    mainTask.Exception.Message; // One or more errors occurred. (Error Job 5) (Error Job 7)
    // ReadOnlyCollection<Exception>
    mainTask.InnerExceptions; // [0] Error Job 5, [1] Error Job 7
} 

async Task<Job> CreateJobAsync(string name)
{
    await Task.Delay(1);

    if (name.EndsWith("5") || name.EndsWith("7") )
        throw new Exception($"Error {name}");

    return new Job { Name = name };
}

class Job
{
    public string Name { get; set; }
}

Solution: use the task ids

List<(string, Task<Job>)> jobNameWithCreateJobTasks = Enumerable.Range(1, 10).Select(x => (jobName: $"Job {x}", task: CreateJobAsync($"Job {x}"))).ToList();
Dictionary<int, string> jobNameFromTaskId = jobNameWithCreateJobTasks.ToDictionary(x => x.task.Id, x => x.jobName);
List<Task<Job>> createJobTasks = jobNameWithCreateJobTasks.Select(x => x.task).ToList();

List<(string JobName, Job? Job, string ErrorMessage)> results;
var results = await TaskExtension.WhenAll(createJobTasks);

results.Select(x => new
    {
        JobName = x.Result?.JobName ?? jobNameFromTaskId[x.taskId],
        Job = x.Result,
        GlobalErrorMessage = x.Exception?.Message ?? string.Empty
    }));

TaskExtension.cs

Solution: use ContinueWith

var (jobs, exceptions) = await WhenAllWithExceptions(createJobTasks);

static Task<(T[] Results, Exception[] Exceptions)> WhenAllWithExceptions<T>(IReadOnlyCollection<Task<T>> tasks)
{
    ArgumentNullException.ThrowIfNull(tasks);

    return Task.WhenAll(tasks).ContinueWith(t =>
        {
            T[] results = tasks
                .Where(t => t.IsCompletedSuccessfully)
                .Select(t => t.Result)
                .ToArray();

            var aggregateExceptions = tasks
                .Where(t => t.IsFaulted)
                .Select(t => t.Exception!);

            var exceptions = new AggregateException(aggregateExceptions)
                .Flatten()
                .InnerExceptions
                .ToArray();

            // No exceptions and at least one task was canceled
            if (exceptions.Length == 0 && t.IsCanceled)
            {
                exceptions = new[] { new TaskCanceledException(t) };
            }

            return (results, exceptions);
        },
        default,
        TaskContinuationOptions.DenyChildAttach | TaskContinuationOptions.ExecuteSynchronously,
        TaskScheduler.Default);
}

Solution: wrapping the task into a TaskResult

Instead of having a unique try/catch for all the tasks, have it for each task.

// the created jobs are wrapped into TaskResult to handle Exception
IEnumerable<Task<TaskResult<Job>>> createJobTasks = Enumerable.Range(1, 10).Select(x => CreateJobAsync($"Job {x}").ToTaskResultAsync());
var jobs = await Task.WhenAll(createJobTasks);

// if the TaskResult is a succes then access to the Result otherwise access to the Exception ErrorMessage
var writeTasks = jobs.Select(x => Task.Run(() => Console.WriteLine(x.Success ? x.Result.Name : x.ErrorMessage)));
await Task.WhenAll(writeTasks);

TaskExtension.cs

PLINQ

AsParallel analyses the query to see if it is suitable for parallelization. This analysis adds overhead.
If it is unsafe or faster to run sequentially then it won't be run in parallel.

var numbers = Enumerable.Range(0, 100_000_000);

var parallelResult = numbers.AsParallel()
                            .WithDegreeOfParallelism(2)
                            .WithCancellation(token)
                            .WithExecutionMode(ParallelExecutionMode.ForceParallelism)
                            .WithMergeOptions(ParallelMergeOptions.Default)
                            .AsOrdered() // add overhead
                            .Where(i => i % 2 == 0);

// parcourt d'itération en mode parallèle, l'ordre est perdu.
// le parcourt commence même si parallelResult n'est pas au complet
parallelResult.ForAll(e => Console.WriteLine(e));