Sometimes, the functions we call take time to return results of the computation. Often, the cause is side-effects like reading from a disk or calling a slow remote API. At times, the operation itself just requires a lot of CPU time to finish. In both cases, the main flow of the program is stopped until the function returns the result. It might be acceptable in the latter case to wait for the result if it is required immediately after calculation (though even in this case it is suboptimal because it makes the system unresponsive), but it is undesirable in the former case because it means that our program consumes CPU while doing nothing (well, waiting for other subsystems of the computer to return the result, but still nothing related to the CPU). Often, such long-running operations are executed in a separate thread of execution.

As a functional programmer, we would like to express these—two aspects, that is, the duration of the call and the fact that the code is executed in a separate thread, as an effect. This is what Future does. To be more specific, it does not represent the duration of the call explicitly, but encodes it in a binary form—an operation either takes long and possibly runs in a separate thread or it doesn't.

The Future is a very interesting concept and deserves a full chapter on its own. Here, we'll just take a brief look at some of its aspects. We highly recommend referring to the official documentation for more details. Let's apply our ubiquitous three-step approach one more time, this time for Future.