Future.apply
starts an asynchronous computation whereas Future.successful
creates an already completed Future with the specified result.
Now is
Future.apply(None)
creates asynchronous computation and executes it. It means that extra lambda object is created and extra task is scheduled (however trivial task).
Future.successful(None)
just produces already completed future. It is more efficient.
I don't think that Future(None)
gives a big overhead, but still in it's default implementation each call to apply
spawns a new task for a ForkJoin thread pool, whereas Future.successful(None)
completes immediately. And each call to map
or flatMap
on the future creates a new task for the poll, which also gives some overhead, so you might want to take a look at scalaz
Future/Task implementation which handles such details more carefully.