Packaged Task

std::packaged_task

  • Defined in <future>

  • Encapsulates a task

    • A callable object for the task's code
    • An std::promise for the result of the task

  • Provided a higher level of abstraction

  • Template class

    • The parameter is the callable object's signature
    // std::packaged_task object
// The callable object takes int and int arguments.
// It returns int
std::packaged_task<int(int, int)> ptask(...);

  • The constructor takes the callable object as argument

std::packaged_task Interface

  • Functor class
  • Overloaded operator()
    • Invokes the callable object
    • Stores the return value in the promise object
  • get_future()
    • Returns the std::future object associated with the promise
  • std::packaged_task is a move-only class

Using an std::packaged_task Object

  • Pass a callable object to the constructor
  • The packaged task starts when operator() is called
    • In the same thread, by calling it directly
    • In a new thread, by passing the task to std::thread's constructor
  • We call get_future()
  • We call get() on the returned future object
    • Or wait() and friends

Example of std::packaged_task in Same Thread


// Packaged task using lambda expression
std::packaged_task<int(int, int)> ptask([](int a, int b) {
  return a + b;
});

// The future associated with the packaged_task's promise
std::future<int> fut = ptask.get_future();

// invoke the packaged task in this thread
ptask(6, 7);

// Call get() to recieve the result
fut.get();

Example of std::packaged_task in New Thread

// Create a thread
// The packaged task will be its entry point
std::thread thr(std::move(ptask), 6, 7);
thr.join();

Advantages of std::packaged_task

  • Avoids boilerplate code
    • Create std::promise object
    • Pass it to task function

Applications of std::packaged_task

  • Create a container of packaged_task objects

    • The threads do not start up until we are ready for them

  • Useful for managing threads

    • Each task can be run on a specified thread
    • Thread scheduler runs threads in a certain order
    • Thread pool consists of threads waiting for work to arrive