Stateless Future is a set of Domain-specific language for asynchronous programming, in the pure functional flavor.
Stateless Futures provide similar API to scala.concurrent.Future and scala.async, except Stateless Futures are simpler, cleaner, and more powerful than scala.concurrent.Future and scala.async.
There was a continuation plugin for Scala. The continuation plugin also provided a DSL to define control flows like stateless-future or scala.async. I created the following table to compare the three DSL:
| stateless-future | scala.concurrent.Future and scala.async | scala.util.continuations | |
|---|---|---|---|
| Stateless | Yes | No | Yes |
| Threading-free | Yes | No | Yes |
| Exception handling in "A-Normal Form" | Yes | No | No |
| Tail call optimization in "A-Normal Form" | Yes | No | No |
| Pattern matching in "A-Normal Form" | Yes | Yes | Yes, but buggy |
| Lazy val in "A-Normal Form" | No, because of some underlying scala.reflect bugs | Only for those contain no await |
Yes, but buggy |
import com.qifun.statelessFuture.Future
val randomDoubleFuture: Future.Stateless[Double] = Future {
println("Generating a random Double...")
math.random()
}
A Stateless Future instance is lazy, only evaluated when you query it. Thus there is nothing printed when you create the Stateless Future.
println("I am going to read a random Double.")
for (randomDouble <- randomDoubleFuture) {
println(s"Recevied $randomDouble.")
}
Output:
I am going to read a random Double.
Generating a random Double...
Recevied 0.19722960355012198.
val anotherFuture = Future {
println("I am going to read the first random Double.")
val randomDouble1 = randomDoubleFuture.await
println(s"The first random Double is $randomDouble1.")
println("I am going to read the second random Double.")
val randomDouble2 = randomDoubleFuture.await
println(s"The second random Double is $randomDouble2.")
}
println("Before running the Future.")
for (unit <- anotherFuture) {
println("After running the Future.")
}
Output:
Before running the Future.
I am going to read the first random Double.
Generating a random Double...
The first random Double is 0.10768210465170625.
I am going to read the second random Double.
Generating a random Double...
The second random Double is 0.6134780449033244.
After running the Future.
Note the magic await postfix, which invokes the the former Stateless Future randomDoubleFuture. It looks like a normal Scala method calls, but does not block any thread.
import scala.concurrent.duration._
import scala.util.control.Exception.Catcher
import com.qifun.statelessFuture.Future
val executor = java.util.concurrent.Executors.newSingleThreadScheduledExecutor
// Manually implements a Stateless Future, which is the asynchronous version of `Thread.sleep()`
def asyncSleep(duration: Duration) = new Future.Stateless[Unit] {
import scala.util.control.TailCalls._
def onComplete(handler: Unit => TailRec[Unit])(implicit catcher: Catcher[TailRec[Unit]]) = {
executor.schedule(new Runnable {
def run() {
handler().result
}
}, duration.length, duration.unit)
done()
}
}
// Without the keyword `new`, you have the magic version of `Future` constructor,
// which enables the magic postfix `await`.
val sleep10seconds = Future {
var i = 0
while (i < 10) {
println(s"I have slept $i times.")
// The magic postfix `await` invokes the asynchronous method `asyncSleep`.
// It looks like normal `Thread.sleep()`, but does not block any thread.
asyncSleep(1.seconds).await
i += 1
}
i
}
// When `sleep10seconds` is running, it could report failures to this catcher
implicit def catcher: Catcher[Unit] = {
case e: Exception => {
println("An exception occured when I was sleeping: " + e.getMessage)
}
}
// A Stateless Future instance is lazy, only evaluating when you query it.
println("Before the evaluation of the Stateless Future `sleep10seconds`.")
for (total <- sleep10seconds) {
println("After the evaluation of the Stateless Future `sleep10seconds`.")
println(s"I slept $total times in total.")
executor.shutdown()
}
Run it and you will see the output:
Before evaluation of the Stateless Future `sleep10seconds`.
I have slept 0 times.
I have slept 1 times.
I have slept 2 times.
I have slept 3 times.
I have slept 4 times.
I have slept 5 times.
I have slept 6 times.
I have slept 7 times.
I have slept 8 times.
I have slept 9 times.
After evaluation of the Stateless Future `sleep10seconds`.
I slept 10 times in total.
There are two sorts of API to use a Stateless Future, the for-comprehensions style API and "A-Normal Form" style API.
The for-comprehensions style API for stateless-future is like the for-comprehensions for scala.concurrent.Future.
for (total <- sleep10seconds) {
println("After evaluation of the Stateless Future `sleep10seconds`")
println(s"I slept $total times in total.")
executor.shutdown()
}
A notable difference between the two for-comprehensions implementations is the required implicit parameter. A scala.concurrent.Future requires an ExecutionContext, while a Stateless Future requires a Catcher.
import scala.util.control.Exception.Catcher
implicit def catcher: Catcher[Unit] = {
case e: Exception => {
println("An exception occured when I was sleeping: " + e.getMessage)
}
}
"A-Normal Form" style API for Stateless Futures is like the pending proposal scala.async, except Stateless Futures require less limitations than scala.async.
val sleep10seconds = Future {
var i = 0
while (i < 10) {
println(s"I have slept $i times")
// The magic postfix `await` invokes asynchronous method like normal `Thread.sleep()`,
// and does not block any thread.
asyncSleep(1.seconds).await
i += 1
}
i
}
The Future function for Stateless Futures corresponds to async method in Async, and the await postfix to Stateless Futures corresponds to await method in Async.
Regardless of the familiar veneers between Stateless Futures and scala.concurrent.Future, I have made some different designed choices on Stateless Futures.
The Stateless Futures are pure functional, thus they will never store result values or exceptions. Instead, Stateless Futures evaluate lazily, and they do the same job every time you invoke foreach or onComplete. The behavior of Stateless Futures is more like monads in Haskell than futures in Java.
Also, there is no isComplete method in Stateless Futures. As a result, the users of Stateless Futures are forced not to share futures between threads, not to check the states in futures. They have to care about control flows instead of threads, and build the control flows by defining Stateless Futures.
There are too many threading models and implimentations in the Java/Scala world, java.util.concurrent.Executor, scala.concurrent.ExecutionContext, javax.swing.SwingUtilities.invokeLater, java.util.Timer, ... It is very hard to communicate between threading models. When a developer is working with multiple threading models, he must very carefully pass messages between threading models, or he have to maintain bulks of synchronized methods to properly deal with the shared variables between threads.
Why does he need multiple threading models? Because the libraries that he uses depend on different threading modes. For example, you must update Swing components in the Swing's UI thread, you must specify java.util.concurrent.ExecutionServices for java.nio.channels.CompletionHandler, and, you must specify scala.concurrent.ExecutionContexts for scala.concurrent.Future and scala.async.Async. Oops!
Think about somebody who uses Swing to develop a text editor software. He wants to create a state machine to update UI. He have heard the cool scala.async, then he uses the cool "A-Normal Form" expression in async to build the state machine that updates UI, and he types import scala.concurrent.ExecutionContext.Implicits._ to suppress the compiler errors. Everything looks pretty, except the software always crashes.
Fortunately, stateless-future depends on none of these threading model, and cooperates with all of these threading models. If the poor guy tries Stateless Future, replacing async { } to stateless-future's Future { }, deleting the import scala.concurrent.ExecutionContext.Implicits._, he will find that everything looks pretty like before, and does not crash any more. That's why threading-free model is important.
There were two Future implementations in Scala standard library, scala.actors.Future and scala.concurrent.Future. scala.actors.Futures are not designed to handling exceptions, since exceptions are always handled by actors. There is no way to handle a particular exception in a particular subrange of an actor.
Unlike scala.actors.Futures, scala.concurrent.Futures are designed to handle exceptions. But, unfortunately, scala.concurrent.Futures provide too many mechanisms to handle an exception. For example:
import scala.concurrent.Await
import scala.concurrent.ExecutionContext
import scala.concurrent.duration.Duration
import scala.util.control.Exception.Catcher
import scala.concurrent.forkjoin.ForkJoinPool
val threadPool = new ForkJoinPool()
val catcher1: Catcher[Unit] = { case e: Exception => println("catcher1") }
val catcher2: Catcher[Unit] = {
case e: java.io.IOException => println("catcher2")
case other: Exception => throw new RuntimeException(other)
}
val catcher3: Catcher[Unit] = {
case e: java.io.IOException => println("catcher3")
case other: Exception => throw new RuntimeException(other)
}
val catcher4: Catcher[Unit] = { case e: Exception => println("catcher4") }
val catcher5: Catcher[Unit] = { case e: Exception => println("catcher5") }
val catcher6: Catcher[Unit] = { case e: Exception => println("catcher6") }
val catcher7: Catcher[Unit] = { case e: Exception => println("catcher7") }
def future1 = scala.concurrent.future { 1 }(ExecutionContext.fromExecutor(threadPool, catcher1))
def future2 = scala.concurrent.Future.failed(new Exception)
val composedFuture = future1.flatMap { _ => future2 }(ExecutionContext.fromExecutor(threadPool, catcher2))
composedFuture.onFailure(catcher3)(ExecutionContext.fromExecutor(threadPool, catcher4))
composedFuture.onFailure(catcher5)(ExecutionContext.fromExecutor(threadPool, catcher6))
try { Await.result(composedFuture, Duration.Inf) } catch { case e if catcher7.isDefinedAt(e) => catcher7(e) }
Is any sane developer able to tell which catchers will receive the exceptions?
There are too many concepts about exceptions when you work with scala.concurrent.Future. You have to remember the different exception handling strategies between flatMap, recover, recoverWith and onFailure, and the difference between scala.concurrent.Future.failed(new Exception) and scala.concurrent.future { throw new Exception }.
scala.async does not make things better, because scala.async will produce a compiler error for every await in a try statement.
Fortunately, you can get rid of all those concepts if you switch to stateless-future. There is neither catcher implicit parameter in flatMap or map in Stateless Futures, nor onFailure nor recover method at all. You just simply try, and things get done. See the examples to learn that.
Tail call optimization is an important feature for pure functional programming. Without tail call optimization, many recursive algorithm will fail at run-time, and you will get the well-known StackOverflowError.
The Scala language provides scala.annotation.tailrec to automatically optimize simple tail recursions, and scala.util.control.TailCalls to manually optimize complex tail calls.
stateless-future project is internally based on scala.util.control.TailCalls, and automatically performs tail call optimization in the magic Future blocks, without any additional special syntax.
See this example. The example creates 500,000,000 stack levels recursively. And it just works, without any StackOverflowError or OutOfMemoryError. Note that if you port this example for scala.async it will throw an OutOfMemoryError or a TimeoutException.
Put these lines in your build.sbt if you use Sbt:
libraryDependencies += "com.qifun" %% "stateless-future" % "0.3.2"
stateless-future should work with Scala 2.10.3, 2.10.4, or 2.11.x.
lazy vals in magicFutureblock are not supported, due to scala issue 8499.unapplySeqs in magicFutureblock are not supported, due to scala issue 8825.- In some rare cases, if you create multiple
valwith same name in oneFutureblock, the lastvalmay be referred unexpectly. - Some complex existential types may cause compiler errors, , due to scala issue 8500.
Clone stateless-future-test and run the test cases to check these limitations.
Discussion around Stateless Future is currently happening in the Gitter channel as well as on Github issue and PR pages.
People are expected to follow the Typelevel Code of Conduct when discussing Cats on the Github page, Gitter channel, or other venues.
|
明日复明日， 文嘉／錢鶴灘 |
The Future flatMaps a Future. Wen Jia / Qian Hetan |