Tasks are special functions that are called once each frame while your application executes. They are similar in concept to threads. However, in Panda3D, tasks are not generally separate threads; instead, all tasks are run cooperatively, one at a time, within the main thread. This design simplifies game programming considerably by removing the requirement to protect critical sections of code from mutual access. (See Task Chains in the next section if you really want to use threading.)

When you start Panda3D by initializing WindowFramework, a handful of tasks are created by default, but you are free to add as many additional tasks as you like.

The Task Function

A task is defined with a function or class method; this function is the main entry point for the task and will be called once per frame while the task is running. By default, the function receives one parameter, which is the task object; the task object carries information about the task itself, such as the amount of time that the task has been running.

Your task function should return when it has finished processing for the frame. Because all tasks are run in the same thread, you must not spend too much time processing any one task function; the entire application will be locked up until the function returns.

The task function may return either AsyncTask::DS_cont to indicate that the task should be called again next frame, or AsyncTask::DS_done to indicate that it should not be called again.

You can check how long your task has been running by checking task->get_elapsed_time() in your task function. You can also check how many times the task function has been run by using task->get_elapsed_frames().

#include "asyncTaskManager.h"

// This task runs for two seconds, then prints done
AsyncTask::DoneStatus example_task(GenericAsyncTask *task, void *data) {
  if (task->get_elapsed_time() < 2.0) {
    return AsyncTask::DS_cont;
  cout << "Done" << endl;
  return AsyncTask::DS_done;

Task Return Values

The value returned from a task affects how the task manager handles that task going forward.




Specifies that a task is finished and removes it from the task manager.


Perform the task again next frame.


Perform the task again, using the same delay as initially specified.

The Do-Later Task

If you have used Panda3D in Python, you might be familiar with the Python function taskMgr.doMethodLater(), which lets you schedule a task to be started after a certain delay. This isn’t needed in C++, because you can set a delay on a task directly with task->set_delay(). An example will be provided below in the task manager section.

The Task Object

The task object is passed into all task functions. There are several members accessible in the func object, among which:




A float that indicates how long this task function has been running since the first execution of the function. The timer is running even when the task function is not being executed.


An integer that counts the number of elapsed frames since this function was added. Count may start from 0 or 1.


An integer that gives the unique id assigned to this task by the Task Manager.


The task name assigned to the task function.

To remove the task and stop it from executing from outside the task function, call task.remove().

The Task Manager

All tasks are handled through the Task Manager object. Here we assume that you have obtained a reference to it and stored it in a variable called task_mgr, for example:

PT(AsyncTaskManager) task_mgr = AsyncTaskManager::get_global_ptr();

The Task Manager keeps a list of all currently-running tasks.

To add a task to the Task Manager, first create a task object by indicating your function and an arbitrary name, and then add it to the task list by calling task_mgr->add() with a pointer to your task.

PT(GenericAsyncTask) task;
task = new GenericAsyncTask("MyTaskName", &example_task, nullptr);


You can add an arbitrary argument to the call through the third parameter.

Although normally each task is given a unique name, you may also create multiple different tasks with the same name. This can be convenient for locating or removing many task functions at the same time. Each task remains independent of the others, even if they have the same name; this means that a task function returning a “done” status will not affect any other task functions.

To remove the task and stop it from executing, you can call task->remove().


A useful task method is task->set_delay(); it causes your task to be called after a certain amount of time (in seconds). You can, of course, implement this kind of functionality with an underlayed task that simply does nothing until a certain amount of time has elapsed (as in the above example), but using this method is a much more efficient way to achieve the same thing, especially if you will have many such tasks waiting around. Note that you need to set the delay before you add the task to the Task Manager, otherwise the call won’t have an effect.

task->set_delay (60);

Similarly, if you wish to change the delay time of a task, you have to remove the task and re-add it by hand. For instance:


You can also alter the delay of the task inside the task function, but you will have to return AsyncTask::DS_again afterwards so that it takes effect.

You may add a cleanup function to the task with the task->set_upon_death() function. Similar to task functions, this function receives a function pointer as a parameter. The cleanup function is called whenever the task finishes, for instance by return AsyncTask::DS_done;, or when it is explicitly removed via a task->remove() call.


To control order in which tasks are executed, you can use sort or priority argument. If you use only sort or only priority, tasks given lesser value will execute sooner.

If you use both sort and priority arguments, tasks with lower sort value will be executed first. However, if there are several tasks which have same sort value, but different priority value then that tasks are going to be executed in a way that ones with HIGHER priority value will be executed first.

To print the list of tasks currently running, simply call task_mgr->write(cout);.

Task timing



int task_accumulator = 0;

void clean_up(GenericAsyncTask *task, bool clean_exit, void *user_data) {
  cout << "Task func has accumulated " << task_accumulator << endl;
  //  Reset the accumulator
  task_accumulator = 0;

// A task that runs forever
AsyncTask::DoneStatus task_func(GenericAsyncTask *task, void *data) {
  return AsyncTask::DS_cont;

AsyncTask::DoneStatus task_stop(GenericAsyncTask *task, void *data) {
  ((GenericAsyncTask *)data)->remove();
  return AsyncTask::DS_done;

// Note that we skip the initialization and finalization of
// the application for the sake of simplifying the example.
int main(int argc, char *argv[]) {
  /* Insert here your app initialization code */
  /* ... */

  AsyncTaskManager *task_mgr = AsyncTaskManager::get_global_ptr();

  PT(GenericAsyncTask) task, stopper_task;

  // Add the task_func function with an upon_death callback
  task = new GenericAsyncTask("Accumulator", &task_func, nullptr);

  // Adds another task to stop the main task 2 seconds later
  stopper_task = new GenericAsyncTask("Task stopper", &task_stop, task);

  /* Insert here your app finalization code */
  /* ... */