AsyncTask execute() or executeOnExecutor()?

What's the difference between using execute() and executeOnExecuter()?

• How does execute() execute tasks by default? (in serial or in parallel?)

• What should be used for new SDKs >16?

• Is it a good practice to use parallel execution (THREAD_POOL_EXECUTOR) for tasks rather than serial even if it doesn't matter for the application or does that depends on the number of AsyncTasks that will be executed?

.execute() - this function schedules the task on a queue for a single background thread. Means that if you are calling two AsyncTasks and using .execute() method to call them, they will execute in a queue(first then second).

.executeOnExecutor() - If you want parallel execution of both AsyncTasks, you can use this method for execution of AsyncTask. Means both asyncTasks will execute simultaneously.

In simple words: .execute() method creates a single thread for execution of asyncTasks, and .executeOnExecuter() method creates separate thread for each ayncTask.

.execute executes tasks by default in serial order.

EDITED: If you want to use executeOnExecutor() you can use this code:

if(Build.VERSION.SDK_INT >= Build.VERSION_CODES.HONEYCOMB)
     task.executeOnExecutor(AsyncTask.THREAD_POOL_EXECUTOR);
 else
     task.execute();

Before HONEYCOMB execute() method run AsynkTask in parallel.

How does .execute execute tasks by default (in serial or in parallel).

Before API level 11: parallel.

API level 11 and up: serial.

which should be used for new SDKs >16 (executeOnExecuter ?)

Depends on your requirements. Use execute() if you're happy with the default executor. Use an explicit executor if you're not.

Is it a good practice to use parallel execution (THREAD_POOL_EXECUTOR) for tasks rather than serial even if it doesn't matter for the application or does that depends on the number of async tasks that will be executed?

Async tasks should only be used for relative short backround operations. Quoting AsyncTask documentation:

AsyncTasks should ideally be used for short operations (a few seconds at the most.) If you need to keep threads running for long periods of time, it is highly recommended you use the various APIs provided by the java.util.concurrent package such as Executor, ThreadPoolExecutor and FutureTask.

While the async task is running, the executor thread cannot execute other tasks. On a serial executor with only one executor thread it is easier to detect problems when your tasks run for too long. On a parallel executor detecting such problems takes more simultaneous long-running tasks.

Therefore, if you really need to switch to a parallel executor, you're probably better off revisiting your design.