Reputation: 1311
So I'm having some trouble with what I think is a pretty simple situation in trait implementation, and I'm hoping there is some simple solution that I'm missing. I'd like to have a method on a trait that accepts as a parameter (and returns as a value only the type of the concrete implementation that it is being called on. Specifically:
trait Foo {
type ConcreteFoo // what to put here?
def combine(that:ConcreteFoo):ConcreteFoo
}
class FooImpl1 extends Foo {
def combine(that:FooImpl1):FooImpl1 = {
// implementation
}
}
class FooImpl2 extends Foo {
def combine(that:FooImpl2):FooImpl2 = {
// implementation
}
}
Right now I have a type Self = FooImpl
on the implementing classes, but I'd rather have something on the trait that takes care of it if possible.
Upvotes: 1
Views: 287
Reputation: 149538
This is exactly F-Bounded Polymorphism:
trait Foo[T <: Foo[T]]
def combine(that: T): T
}
class FooImpl1 extends Foo[FooImpl1] {
def combine(that: FooImpl1): FooImpl1 = {
???
}
}
class FooImpl2 extends Foo[FooImpl2] {
def combine(that: FooImpl2): FooImpl2 = {
???
}
}
Upvotes: 3
Reputation: 18177
You can add a type parameter to your trait like this:
trait Foo[A] {
def combine(that: A): A
}
class FooImpl1 extends Foo[FooImpl1] {
override def combine(that: FooImpl1): FooImpl1 = ???
}
class FooImpl2 extends Foo[FooImpl2] {
override def combine(that: FooImpl2): FooImpl2 = ???
}
Upvotes: 0