Scala single method interface implementation

Scala has experimental support for SAMs starting with 2.11, under the flag -Xexperimental:

Welcome to Scala version 2.11.0-RC3 (OpenJDK 64-Bit Server VM, Java 1.7.0_51).
Type in expressions to have them evaluated.
Type :help for more information.

scala> :set -Xexperimental

scala> val r: Runnable = () => println("hello world")
r: Runnable = $anonfun$1@7861ff33

scala> new Thread(r).run
hello world

Edit: Since 2.11.5, this can also be done inline:

scala> new Thread(() => println("hello world")).run
hello world

The usual limitations about the expected type also apply:

  • it must define a single abstract method,
  • its primary constructor (if any) must be public, no-args, not overloaded,
  • the abstract method must take a single argument list,
  • the abstract method must be monomorphic.

According to the original commit by Adriaan, some of those restrictions may be lifted in the future, especially the last two.


SAM types are supported using invokeDynamic since scala-2.12 similar to JDK-8, Below was tested on 2.12.3 - Release notes about SAM can be found here - http://www.scala-lang.org/news/2.12.0/

object ThreadApp extends App {
  println("Main thread - begins")
  val runnable: Runnable = () => println("hello world - from first thread")

  val thread = new Thread(runnable)
  println("Main thread - spins first thread")
  thread.start()

  val thread2 = new Thread(() => println("hello world - from second thread"))
  println("Main thread - spins second thread")
  thread2.start
  thread.join()
  thread2.join()

  println("Main thread - end")
}

While doing this in a generic way is certainly complicated, if you found that you really only needed this for a few certain Java types, then a few simple implicit conversions can do the job nicely. For instance:

val thread = new Thread(() => println("hello world"))
thread.start

implicit def function0ToRunnable(f:() => Unit):Runnable = 
  new Runnable{def run() = f()}

Sometimes trying to solve a problem in a generic and completely re-useable way is the wrong approach if your actual problem is more bounded then you think.