Muokkaa

Jaa


How to: Create a task that completes after a delay

This example shows how to use the concurrency::task, concurrency::cancellation_token_source, concurrency::cancellation_token, concurrency::task_completion_event, concurrency::timer, and concurrency::call classes to create a task that completes after a delay. You can use this method to build loops that occasionally poll for data. You can also introduce timeouts, delay handling of user input for a predetermined time, and so on.

Example: complete_after and cancel_after_timeout functions

The following example shows the complete_after and cancel_after_timeout functions. The complete_after function creates a task object that completes after the specified delay. It uses a timer object and a call object to set a task_completion_event object after the specified delay. By using the task_completion_event class, you can define a task that completes after a thread or another task signals that a value is available. When the event is set, listener tasks complete and their continuations are scheduled to run.

Tip

For more information about the timer and call classes, which are part of the Asynchronous Agents Library, see Asynchronous Message Blocks.

The cancel_after_timeout function builds on the complete_after function to cancel a task if that task doesn't complete before a given timeout. The cancel_after_timeout function creates two tasks. The first task indicates success and completes after the provided task completes. The second task indicates failure and completes after the specified timeout. The cancel_after_timeout function creates a continuation task that runs when the success or failure task completes. If the failure task completes first, the continuation cancels the token source to cancel the overall task.

// Creates a task that completes after the specified delay.
task<void> complete_after(unsigned int timeout)
{
    // A task completion event that is set when a timer fires.
    task_completion_event<void> tce;

    // Create a non-repeating timer.
    auto fire_once = new timer<int>(timeout, 0, nullptr, false);
    // Create a call object that sets the completion event after the timer fires.
    auto callback = new call<int>([tce](int)
    {
        tce.set();
    });

    // Connect the timer to the callback and start the timer.
    fire_once->link_target(callback);
    fire_once->start();

    // Create a task that completes after the completion event is set.
    task<void> event_set(tce);

    // Create a continuation task that cleans up resources and
    // and return that continuation task.
    return event_set.then([callback, fire_once]()
    {
        delete callback;
        delete fire_once;
    });
}

// Cancels the provided task after the specifed delay, if the task
// did not complete.
template<typename T>
task<T> cancel_after_timeout(task<T> t, cancellation_token_source cts, unsigned int timeout)
{
    // Create a task that returns true after the specified task completes.
    task<bool> success_task = t.then([](T)
    {
        return true;
    });
    // Create a task that returns false after the specified timeout.
    task<bool> failure_task = complete_after(timeout).then([]
    {
        return false;
    });

    // Create a continuation task that cancels the overall task 
    // if the timeout task finishes first.
    return (failure_task || success_task).then([t, cts](bool success)
    {
        if(!success)
        {
            // Set the cancellation token. The task that is passed as the
            // t parameter should respond to the cancellation and stop
            // as soon as it can.
            cts.cancel();
        }

        // Return the original task.
        return t;
    });
}

Example: Compute count of prime numbers

The following example computes the count of prime numbers in the range [0, 100000] multiple times. The operation fails if it doesn't complete in a given time limit. The count_primes function demonstrates how to use the cancel_after_timeout function. It counts the number of primes in the given range and fails if the task doesn't complete in the provided time. The wmain function calls the count_primes function multiple times. Each time, it halves the time limit. The program finishes after the operation doesn't complete in the current time limit.

// Determines whether the input value is prime.
bool is_prime(int n)
{
    if (n < 2)
        return false;
    for (int i = 2; i < n; ++i)
    {
        if ((n % i) == 0)
            return false;
    }
    return true;
}

// Counts the number of primes in the range [0, max_value].
// The operation fails if it exceeds the specified timeout.
bool count_primes(unsigned int max_value, unsigned int timeout)
{
    cancellation_token_source cts;

    // Create a task that computes the count of prime numbers.
    // The task is canceled after the specified timeout.
    auto t = cancel_after_timeout(task<size_t>([max_value, timeout, cts]
    {
        combinable<size_t> counts;
        parallel_for<unsigned int>(0, max_value + 1, [&counts, cts](unsigned int n) 
        {
            // Respond if the overall task is cancelled by canceling 
            // the current task.
            if (cts.get_token().is_canceled())
            {
                cancel_current_task();
            }
            // NOTE: You can replace the calls to is_canceled
            // and cancel_current_task with a call to interruption_point.
            // interruption_point();

            // Increment the local counter if the value is prime.
            if (is_prime(n))
            {
                counts.local()++;
            }
        });
        // Return the sum of counts across all threads.
        return counts.combine(plus<size_t>());
    }, cts.get_token()), cts, timeout);

    // Print the result.
    try
    {
        auto primes = t.get();
        wcout << L"Found " << primes << L" prime numbers within " 
              << timeout << L" ms." << endl;
        return true;
    }
    catch (const task_canceled&)
    {
        wcout << L"The task timed out." << endl;
        return false;
    }
}

int wmain()
{
    // Compute the count of prime numbers in the range [0, 100000] 
    // until the operation fails.
    // Each time the test succeeds, the time limit is halved.

    unsigned int max = 100000;
    unsigned int timeout = 5000;
    
    bool success = true;
    do
    {
        success = count_primes(max, timeout);
        timeout /= 2;
    } while (success);
}
/* Sample output:
    Found 9592 prime numbers within 5000 ms.
    Found 9592 prime numbers within 2500 ms.
    Found 9592 prime numbers within 1250 ms.
    Found 9592 prime numbers within 625 ms.
    The task timed out.
*/

When you use this technique to cancel tasks after a delay, any unstarted tasks won't start after the overall task is canceled. However, it's important for any long-running tasks to respond to cancellation quickly. For more information about task cancellation, see Cancellation in the PPL.

Complete code example

Here is the complete code for this example:

// task-delay.cpp
// compile with: /EHsc
#include <ppl.h>
#include <ppltasks.h>
#include <agents.h>
#include <iostream>

using namespace concurrency;
using namespace std;

// Creates a task that completes after the specified delay.
task<void> complete_after(unsigned int timeout)
{
    // A task completion event that is set when a timer fires.
    task_completion_event<void> tce;

    // Create a non-repeating timer.
    auto fire_once = new timer<int>(timeout, 0, nullptr, false);
    // Create a call object that sets the completion event after the timer fires.
    auto callback = new call<int>([tce](int)
    {
        tce.set();
    });

    // Connect the timer to the callback and start the timer.
    fire_once->link_target(callback);
    fire_once->start();

    // Create a task that completes after the completion event is set.
    task<void> event_set(tce);

    // Create a continuation task that cleans up resources and
    // and return that continuation task.
    return event_set.then([callback, fire_once]()
    {
        delete callback;
        delete fire_once;
    });
}

// Cancels the provided task after the specifed delay, if the task
// did not complete.
template<typename T>
task<T> cancel_after_timeout(task<T> t, cancellation_token_source cts, unsigned int timeout)
{
    // Create a task that returns true after the specified task completes.
    task<bool> success_task = t.then([](T)
    {
        return true;
    });
    // Create a task that returns false after the specified timeout.
    task<bool> failure_task = complete_after(timeout).then([]
    {
        return false;
    });

    // Create a continuation task that cancels the overall task 
    // if the timeout task finishes first.
    return (failure_task || success_task).then([t, cts](bool success)
    {
        if(!success)
        {
            // Set the cancellation token. The task that is passed as the
            // t parameter should respond to the cancellation and stop
            // as soon as it can.
            cts.cancel();
        }

        // Return the original task.
        return t;
    });
}

// Determines whether the input value is prime.
bool is_prime(int n)
{
    if (n < 2)
        return false;
    for (int i = 2; i < n; ++i)
    {
        if ((n % i) == 0)
            return false;
    }
    return true;
}

// Counts the number of primes in the range [0, max_value].
// The operation fails if it exceeds the specified timeout.
bool count_primes(unsigned int max_value, unsigned int timeout)
{
    cancellation_token_source cts;

    // Create a task that computes the count of prime numbers.
    // The task is canceled after the specified timeout.
    auto t = cancel_after_timeout(task<size_t>([max_value, timeout, cts]
    {
        combinable<size_t> counts;
        parallel_for<unsigned int>(0, max_value + 1, [&counts, cts](unsigned int n) 
        {
            // Respond if the overall task is cancelled by canceling 
            // the current task.
            if (cts.get_token().is_canceled())
            {
                cancel_current_task();
            }
            // NOTE: You can replace the calls to is_canceled
            // and cancel_current_task with a call to interruption_point.
            // interruption_point();

            // Increment the local counter if the value is prime.
            if (is_prime(n))
            {
                counts.local()++;
            }
        });
        // Return the sum of counts across all threads.
        return counts.combine(plus<size_t>());
    }, cts.get_token()), cts, timeout);

    // Print the result.
    try
    {
        auto primes = t.get();
        wcout << L"Found " << primes << L" prime numbers within " 
              << timeout << L" ms." << endl;
        return true;
    }
    catch (const task_canceled&)
    {
        wcout << L"The task timed out." << endl;
        return false;
    }
}

int wmain()
{
    // Compute the count of prime numbers in the range [0, 100000] 
    // until the operation fails.
    // Each time the test succeeds, the time limit is halved.

    unsigned int max = 100000;
    unsigned int timeout = 5000;
    
    bool success = true;
    do
    {
        success = count_primes(max, timeout);
        timeout /= 2;
    } while (success);
}
/* Sample output:
    Found 9592 prime numbers within 5000 ms.
    Found 9592 prime numbers within 2500 ms.
    Found 9592 prime numbers within 1250 ms.
    Found 9592 prime numbers within 625 ms.
    The task timed out.
*/

Compiling the Code

To compile the code, copy it and then paste it in a Visual Studio project, or paste it in a file that is named task-delay.cpp and then run the following command in a Visual Studio Command Prompt window.

cl.exe /EHsc task-delay.cpp

See also

Task Parallelism
task Class (Concurrency Runtime)
cancellation_token_source Class
cancellation_token Class
task_completion_event Class
timer Class
call Class
Asynchronous Message Blocks
Cancellation in the PPL