Potential Semantic Clarity

[0xrutvij]

Hello! Firstly, I would like to thank all the contributors for the awesome tour of Scala Lang. While going through the example, the usage of Node[Bird] type for birdList threw me in a loop for a moment. The location of usage has been proposed as a change for both demonstration purposes and as a potential change.

I understand traits can be used to describe a type, but in this instance wouldn't it be clearer to say ListNode[Bird] instead? Seeing that the value name –birdList– indicates we are going for a list. From what I can surmise, the Node trait can be implemented by say, a GraphNode which uses the prepend method in some way (one might argue it to be a convoluted use case). For the sake of clarity and consistency with the semantic meaning of the example, I believe ListNode[Bird] would be a better choice.

[som-snytt]

The section is demonstrating why you need a type parameter for the method and give it a bound.

It doesn't quite demonstrate that.

This shows that you can prepend a "disparate" swallow:

  val africanSwallowList: Node[AfricanSwallow] = ListNode[AfricanSwallow](AfricanSwallow(), Nil())
  val birdList: Node[Bird] = africanSwallowList
  val s0 = birdList.prepend(EuropeanSwallow())
  val s1 = africanSwallowList.prepend(EuropeanSwallow())

(This is that thing where s0 is still Node[Bird] but s1 is Node[Product with Bird with Serializable].)

The last line of text is pointing out that the assignment to birdList works and s0 also works. But I think it is more surprising that s1 works.

Here is an example of why Node is useful, namely, you don't know you have a non-empty list until you prepend:

val swallowsFromAntarctica: Node[Bird] = Nil()

Normally you'd use pattern matching to inspect a Node. And now I see that ListNode is a case class but has extra members.

[som-snytt]

Possibly superseded by #2355 which attempts to shake it out a bit.

[som-snytt]

Thanks!

Expanding this recommendation at #2355 to avoid Node and say List.