ListenableFutureExplained

ListenableFuture

Concurrency is a hard problem, but it is significantly simplified by working with powerful and simple abstractions. To simplify matters, Guava extends the Future interface of the JDK with ListenableFuture.

We strongly advise that you always use ListenableFuture instead of Future in all of your code, because:

• Most Futures methods require it.
• It's easier than changing to ListenableFuture later.
• Providers of utility methods won't need to provide Future and ListenableFuture variants of their methods.

Interface

A traditional Future represents the result of an asynchronous computation: a computation that may or may not have finished producing a result yet. A Future can be a handle to an in-progress computation, a promise from a service to supply us with a result.

A ListenableFuture allows you to register callbacks to be executed once the computation is complete, or if the computation is already complete, immediately. This simple addition makes it possible to efficiently support many operations that the basic Future interface cannot support.

The basic operation added by ListenableFuture is addListener(Runnable, Executor), which specifies that when the computation represented by this Future is done, the specified Runnable will be run on the specified Executor.

Adding Callbacks

Most users will prefer to use Futures.addCallback(ListenableFuture<V>, FutureCallback<V>, Executor), or the version which defaults to using MoreExecutors.directExecutor(), for use when the callback is fast and lightweight. A FutureCallback<V> implements two methods:

• onSuccess(V), the action to perform if the future succeeds, based on its result
• onFailure(Throwable), the action to perform if the future fails, based on the failure

Creation

Corresponding to the JDK ExecutorService.submit(Callable) approach to initiating an asynchronous computation, Guava provides the ListeningExecutorService interface, which returns a ListenableFuture wherever ExecutorService would return a normal Future. To convert an ExecutorService to a ListeningExecutorService, just use MoreExecutors.listeningDecorator(ExecutorService).

ListeningExecutorService service = MoreExecutors.listeningDecorator(Executors.newFixedThreadPool(10));
ListenableFuture<Explosion> explosion = service.submit(new Callable<Explosion>() {
  public Explosion call() {
    return pushBigRedButton();
  }
});
Futures.addCallback(explosion, new FutureCallback<Explosion>() {
  // we want this handler to run immediately after we push the big red button!
  public void onSuccess(Explosion explosion) {
    walkAwayFrom(explosion);
  }
  public void onFailure(Throwable thrown) {
    battleArchNemesis(); // escaped the explosion!
  }
});

Alternatively, if you're converting from an API based on FutureTask, Guava offers ListenableFutureTask.create(Callable<V>) and ListenableFutureTask.create(Runnable, V). Unlike the JDK, ListenableFutureTask is not meant to be extended directly.

If you prefer an abstraction in which you set the value of the future rather than implementing a method to compute the value, consider extending AbstractFuture<V> or using SettableFuture directly.

If you must convert a Future provided by another API to an ListenableFuture, you may have no choice but to use the heavyweight JdkFutureAdapters.listenInPoolThread(Future) to convert a Future to a ListenableFuture. Whenever possible, it is preferred to modify the original code to return a ListenableFuture.

Application

The most important reason to use ListenableFuture is that it becomes possible to have complex chains of asynchronous operations.

ListenableFuture<RowKey> rowKeyFuture = indexService.lookUp(query);
AsyncFunction<RowKey, QueryResult> queryFunction =
  new AsyncFunction<RowKey, QueryResult>() {
    public ListenableFuture<QueryResult> apply(RowKey rowKey) {
      return dataService.read(rowKey);
    }
  };
ListenableFuture<QueryResult> queryFuture =
    Futures.transformAsync(rowKeyFuture, queryFunction, queryExecutor);

Many other operations can be supported efficiently with a ListenableFuture that cannot be supported with a Future alone. Different operations may be executed by different executors, and a single ListenableFuture can have multiple actions waiting upon it.

When several operations should begin as soon as another operation starts -- "fan-out" -- ListenableFuture just works: it triggers all of the requested callbacks. With slightly more work, we can "fan-in," or trigger a ListenableFuture to get computed as soon as several other futures have all finished: see the implementation of Futures.allAsList for an example.

Method	Description	See also
transformAsync(ListenableFuture<A>, AsyncFunction<A, B>, Executor)*	Returns a new ListenableFuture whose result is the product of applying the given AsyncFunction to the result of the given ListenableFuture.	transformAsync(ListenableFuture<A>, AsyncFunction<A, B>)
transform(ListenableFuture<A>, Function<A, B>, Executor)	Returns a new ListenableFuture whose result is the product of applying the given Function to the result of the given ListenableFuture.	transform(ListenableFuture<A>, Function<A, B>)
allAsList(Iterable<ListenableFuture<V>>)	Returns a ListenableFuture whose value is a list containing the values of each of the input futures, in order. If any of the input futures fails or is cancelled, this future fails or is cancelled.	allAsList(ListenableFuture<V>...)
successfulAsList(Iterable<ListenableFuture<V>>)	Returns a ListenableFuture whose value is a list containing the values of each of the successful input futures, in order. The values corresponding to failed or cancelled futures are replaced with null.	successfulAsList(ListenableFuture<V>...)

* An AsyncFunction<A, B> provides one method, ListenableFuture<B> apply(A input). It can be used to asynchronously transform a value.

List<ListenableFuture<QueryResult>> queries;
// The queries go to all different data centers, but we want to wait until they're all done or failed.

ListenableFuture<List<QueryResult>> successfulQueries = Futures.successfulAsList(queries);

Futures.addCallback(successfulQueries, callbackOnSuccessfulQueries);

Avoid nested Futures

In cases where code calls a generic interface and returns a Future, it's possible to end up with nested Futures. For example:

executorService.submit(new Callable<ListenableFuture<Foo>() {
  @Override
  public ListenableFuture<Foo> call() {
    return otherExecutorService.submit(otherCallable);
  }
});

would return a ListenableFuture<ListenableFuture<Foo>>. This code is incorrect, because if a cancel on the outer future races with the completion of the outer future, that cancellation will not be propagated to the inner future. It's also a common error to check for failure of the other future using get() or a listener, but unless special care is taken an exception thrown from otherCallable would be suppressed. To avoid this, all of Guava's future-handling methods (and some from the JDK) have *Async versions that safely unwrap this nesting - transform(ListenableFuture<A>, Function<A, B>, Executor) and transformAsync(ListenableFuture<A>, AsyncFunction<A, B>, Executor), or ExecutorService.submit(Callable) and submitAsync(AsyncCallable<A>, Executor), etc.

CheckedFuture

Guava also provides a CheckedFuture<V, X extends Exception> interface. A CheckedFuture is a ListenableFuture that includes versions of the get methods that can throw a checked exception. This makes it easier to create a future that executes logic which can throw an exception. To convert a ListenableFuture to a CheckedFuture, use Futures.makeChecked(ListenableFuture<V>, Function<Exception, X>).