Question

While looking at scala.collection.mutable.SynchronizedStack I’ve noticed that synchronized used differently, some methods use synchronized[this.type] form

override def push(elem: A): this.type = synchronized[this.type] { super.push(elem) }
override def pushAll(xs: TraversableOnce[A]): this.type = synchronized[this.type] { super.pushAll(elems) }

and some uses synchronized form

override def isEmpty: Boolean = synchronized { super.isEmpty }
override def pop(): A = synchronized { super.pop }

What’s the difference?

Answer 1

The signature of synchronized (declared by AnyRef) is

final def synchronized[T0](arg0: => T0): T0

If you use it as

override def isEmpty: Boolean = synchronized { super.isEmpty }

then you leave it to the compiler to infer the return type of the function passed to synchronized (here Boolean). If you use it as

override def push(elem: A): this.type = synchronized[this.type] {
  super.push(elem)
}

then you explicitly specify the return type (here this.type). I assume that the compiler will not infer this.type – which states that you return exactly this object -, but that it would infer SynchronizedStack or one of its supertypes, which is not as precise as this.type.