參考自http://stackoverflow.com/questions/19478244/how-does-a-case-anonymous-function-really-work-in-scala
http://www.scala-lang.org/files/archive/nightly/pdfs/ScalaReference.pdf
http://docs.scala-lang.org/overviews/core/futures.html
在第三篇文件《Futures and Promises》中,講到了Future物件有三個方法可以註冊callback
import scala.util.{Success, Failure} val f: Future[List[String]] = future { session.getRecentPosts } f onComplete { case Success(posts) => for (post <- posts) println(post) case Failure(t) => println("An error has occured: " + t.getMessage) } f onFailure { case t => println("An error has occured: " + t.getMessage) } f onSuccess { case posts => for (post <- posts) println(post) }
傳給onComplete、onFailture和onSuccess的都是
{ case p1 => b1 ... case pn => bn }
形式的語句,但是這三個方法接受的引數型別卻是不同的。
abstract def onComplete[U](f: (Try[T]) ⇒ U)(implicit executor: ExecutionContext): Unit
def onSuccess[U](pf: PartialFunction[T, U])(implicit executor: ExecutionContext): Unit
def onFailure[U](pf: PartialFunction[Throwable, U])(implicit executor: ExecutionContext): Unit
onCompelete的引數型別的是一個 (Try[T]) => U函式, 而onSuccess和onFailure的引數型別是偏函式。
那麼,問題來了……{ case p1 => b1 ... case pn => bn } 的型別到底是啥呢?
在<The Scala Language Specification>的第8.5章給出了說明:
An anonymous function can be defined by a sequence of cases
{case p1 =>b1 ...case pn =>bn }
which appear as an expression without a prior match.
The expected type of such an expression must in part be defined. It must be either scala.Functionk[S1, ..., Sk, R] for some k >0, or scala.PartialFunction[S1, R], where the argument type(s) S1, ..., Sk must be fully determined, but the result type R may be undetermined.
也就是說{ case p1 => b1 ... case pn => bn } 這種表示式的值的型別可以有兩種,要不是一個函式,要不是一個偏函式(偏函式也是一種函式)。在這個表示式的位置上需要哪種型別,編譯器就會用這個表示式生成對應的型別。但是無論是生成函式還是偏函式,它們的引數的型別都必須是確定的,對於一個特定的Future物件,onComplete接受的函式的引數型別是Try[T],而onSuccess接受的PartialFunction的引數型別是T,onFailure接受的PartialFunction的引數型別是Throwable。但是這些函式的返回型別U可以不是需要這個{ case p1 => b1 ... case pn => bn } 表示式的地方指定的。比如,這三個onXXX方法都沒有指定它所接受的函式的返回值型別。
例子:
import java.io.IOException import scala.concurrent._ import scala.concurrent.ExecutionContext.Implicits.global import scala.util.Success import scala.util.Failure object CallbacksOfFuture extends App { def getRecentPosts = { Thread.sleep(5000) "Good morning" :: "Good afternoon" :: Nil throw new TimeoutException("Goodbye") } val f: Future[List[String]] = future { val posts = getRecentPosts posts }
//onComplete A f onComplete { case Success(posts) => posts.foreach(println) case Failure(e) => println("An error has occured: " + e.getMessage) }
//onComplete B f onComplete { result => result match { case Success(posts) => posts.foreach(println) case Failure(e) => println("An error has occured: " + e.getMessage) } } //won't compile // f onComplete{ // case 1 => 2 // } f onSuccess { case posts => posts foreach println } f onFailure{ case r: IOException => println("got IOException: " + r.getMessage) case r: TimeoutException => println("got TimeoutException: " + r.getMessage) case e => println("An error has occured: " + e.getMessage) } f flatMap{ posts => future{posts}} foreach(println) f map (posts => posts) foreach println Thread.sleep(10000) }
那麼onComplete和onSuccess、onFailure還有啥不同呢?
如果我們把onComplete A實現裡的case Failure去掉,那麼在執行時就會報MatchError,因為
f onComplete { case Success(posts) => posts.foreach(println) case Failure(e) => println("An error has occured: " + e.getMessage) }
實際上會被翻譯為:
f onComplete { result => result match { case Success(posts) => posts.foreach(println) case Failure(e) => println("An error has occured: " + e.getMessage) } }
當result是一個Failure,那麼在去掉case Failure後,它會無法得到匹配,從而報出MatchError.
而在
f onFailure{ case r: IOException => println("got IOException: " + r.getMessage) case r: TimeoutException => println("got TimeoutException: " + r.getMessage) case e => println("An error has occured: " + e.getMessage) }
如果我們只留下case r: IOException,雖然執行時產生的異常是TimeoutException,但是執行時卻不會報錯。這是為啥呢?
看Future的原始碼吧
def onFailure[U](callback: PartialFunction[Throwable, U])(implicit executor: ExecutionContext): Unit = onComplete { case Failure(t) => callback.applyOrElse[Throwable, Any](t, Predef.conforms[Throwable]) // Exploiting the cached function to avoid MatchError case _ => }
原來onFailure會將callback註冊給onComplete,這使得呼叫onFailure也不會阻塞。當Future的執行結果為Failure時,Failure中包裝的異常會被apply給t, 如果apply失敗,會執行Predef.confirm[Throwable]。這個函式是這樣的:
sealed abstract class <:<[-From, +To] extends (From => To) with Serializable private[this] final val singleton_<:< = new <:<[Any,Any] { def apply(x: Any): Any = x } // not in the <:< companion object because it is also // intended to subsume identity (which is no longer implicit) implicit def conforms[A]: A <:< A = singleton_<:<.asInstanceOf[A <:< A]
在這裡,confirm的型別引數為Throwable,於是 singleton_<:<.asInstanceOf[A <:< A]被型別為換為 <:<[Throwable <:< Throwable]。
singleton_<:<本身是一個物件,它的超類的型別是 Any => Any,因此,singleton_<:<.asInstanceOf[Throwable <:< Throwable]是一個型別為(Throwable) => Throwable的函式。因此conform在onFailure中的使用是型別正確的。
那麼onform返回的這個函式幹了啥呢,它的apply方法接收x,返回x。用於onFailure的環境中時,就相當於
def onFailure[U](callback: PartialFunction[Throwable, U])(implicit executor: ExecutionContext): Unit = onComplete { case Failure(t) => callback.applyOrElse[Throwable, Any](t, (e) => e) // Exploiting the cached function to avoid MatchError case _ => }
在callback.applyOrElse方法中,我們需要一個函式,它的型別是(Throwable) => Any,又沒有副作用。那麼用Predef.conform[Throwable]得到一個實際上啥都沒幹的(Throwable) => (Throwable)是很合適的。
應該說這麼寫挺規範吧……