I want to implement a prototype-based system in Scala. At the root of the type hierarchy is the ROOT node, which has a prototype that refers to itself.
The following
I don't quite see how you can do this - how can you have a constructed instance of the root node, if you need it to already exist when creating it? So the thing at fault here is your modelling of the problem domain, not Scala (or any other language for that matter)
Surely the root node could have a null
prototype, or (more idiomatic scala) an Option
prototype?
class Node private[mypackage](val prototype : Option[Node]) {
private def this() = this(None)
private def this(ptype : Node) = this(Some(ptype)) //Public c-tor
}
object Node extends (Node => Node) {
val Root = new Node
def apply(ptype : Node) = new Node(ptype)
}
You could also do the following. Maybe it is not the most idiomatic Scala code but it is short and I think it answers your question.
class Node(prot: Option[Node] = None) { def prototype = prot getOrElse this }
For this looks like you have two kinds of things (Root- and Simple-Nodes) What about this?
trait Node { def prototype: Node }
class RootNode extends Node { def prototype = this }
class SimpleNode(val prototype: Node) extends Node
In the REPL you can do this then:
scala> val rootNode = new RootNode
rootNode: RootNode = RootNode@191dd1d
scala> val n1 = new SimpleNode(rootNode)
n1: SimpleNode = SimpleNode@30e4a7
scala> val n2 = new SimpleNode(n1)
n2: SimpleNode = SimpleNode@3a0589
scala> n2.prototype.prototype
res0: Node = RootNode@191dd1d
I don't know, if that is what you are looking for.