The Runnable
interface has been around since the beginning of time for the Java platform. It allows you to define a task to be completed by a thread. As most people probably know already, it offers a single method run()
that accepts no arguments and returns no values, nor can it throw any checked exceptions. If you need to get a value back from the now-completed task, you must use a method outside the interface and wait for some kind of notification message that the task completed. For example, the following demonstrates what you might do for just such a scenario:
Runnable runnable = ...;
Thread t = new Thread(runnable);
t.start();
t.join();
String value = someMethodtoGetSavedValue()
Nothing is inherently wrong with this code, but it can be done differently now, thanks to the Callable
interface introduced in J2SE 5.0. Instead of having a run()
method, the Callable
interface offers a call()
method, which can return an Object or, more specifically, any type that is introduced in the genericized form:
public interface Callable<V> {
V call() throws Exception;
}
Because you cannot pass a Callable
into a Thread
to execute, you instead use the ExecutorService
to execute the Callable
object. The service accepts Callable
objects to run by way of the submit()
method:
<T> Future<T> submit(Callable<T> task)
As the method definition shows, submitting a Callable object to the ExecutorService
returns a Future
object. Theget()
method of Future
will then block until the task is completed. This is the equivalent of the join()
call in the first example. Actually, it is the equivalent of both the join()
call and the get value call as get()
returns the value calculated by the Callable
instance.
To demonstrate, the following example creates separate Callable
instances for each word passed in on the command line and sums up their length. Each Callable
will just calculate the sum of its individual word. The set ofFuture
objects are saved to acquire the calculated value from each. If the order of the returned values needed to be preserved, a List
could be used instead.
import java.util.\*;
import java.util.concurrent.\*;
public class CallableExample {
public static class WordLengthCallable
implements Callable {
private String word;
public WordLengthCallable(String word) {
this.word = word;
}
public Integer call() {
return Integer.valueOf(word.length());
}
}
public static void main(String args[]) throws Exception {
ExecutorService pool = Executors.newFixedThreadPool(3);
Set<Future<Integer>> set = new HashSet<Future≶Integer>>();
for (String word: args) {
Callable<Integer> callable = new WordLengthCallable(word);
Future<Integer> future = pool.submit(callable);
set.add(future);
}
int sum = 0;
for (Future<Integer> future : set) {
sum += future.get();
}
System.out.printf("The sum of lengths is %s%n", sum);
System.exit(sum);
}
}
The WordLengthCallable
saves each word and uses the word's length as the value returned by the call()
method. This value could take some time to generate but in this case is known immediately. The only requirement ofcall()
is the value is returned at the end of the call. When the get()
method of Future
is later called, the Future
will either have the value immediately if the task runs quickly, as in this case, or will wait until the value is done generating. Multiple calls to get()
will not cause the task to be rerun in the thread.
Because the goal of the program is to calculate the sum of all word lengths, it doesn't matter in which order theCallable
tasks finish. It is perfectly OK if the last task completes before the first three. The first get()
call to Future
will just wait for the first task in the Set
to complete. This does not block other tasks from running to completion separately. It is just waiting for that one thread or task to complete.