Your first two consequences are correct. You missed one though in the fact that when ParentWithExplicitChildren
fails itself, it will stop and then start again all of it's children because it's the explicit supervisor for those children. In the case of ParentWithActorRefs
, a failure in this actor will not stop the shamChildren
because it's not their supervisor; the root guardian is.
Also, yes, you can access the children of ParentWithExplicitChildren
refs via actor selection. They are proper actors with addressable paths and thus can be looked up and communicated with from outside of their parent/supervising actor.