Imported From: https://issues.scala-lang.org/browse/SI-9890?orig=1
Reporter: Jonathan Schuchart (j.schuchart)
Affected Versions: 2.11.8

@som-snytt said:
The second section of bullets says you can select r.x only if "r's type conforms to a type-instance of the class which contains the access." So the member should be of type B. Otherwise, you could access members of other implementations of A with arbitrary behaviors for the template method.

By contrast,

object X {
  trait A { protected[X] def callMe(): Unit }
  trait B extends A { protected def a: A ; final override protected[X] def callMe(): Unit = a.callMe() }
}

Jonathan Schuchart (j.schuchart) said:
I am not sure what the exact meaning of "r's type conforms to a type-instance of the class which contains the access" is then. Maybe you can point me to a site that has it explained in more detail?
As I understand it, in my example B is a subtype of A and should therefore conform to being of type A which should give B access to protected members of other instances of A.
Since both of the following access patterns work

trait A {
  protected def callMe(): Unit
  def runOther(a: A): Unit = a.callMe()
}
trait SubA extends A {
  callMe()
}

Why would the combination of the two as outlined above not work as well? Where is the difference to protected[this] then?

@som-snytt said:
You have def a: A, and you attempt a.callMe. A does not conform to B, where you attempt the access. You can try StackOverflow or the gitter channel for better clarifications. Protected access is kind of a stickler, especially because of mix-in traits. You might want to ask about your use case on gitter.

Jonathan Schuchart (j.schuchart) said:
Ok. It makes sense now. Sorry for the false report. Thanks