* The race can happen if the MemberRemoved event is received followed by a Delta update from
a node that has not yet got the MemberRemoved. That will make the bucket for the removed
node to be added back in the registry.
* Based on Jonas' notes: https://gist.github.com/jboner/7692270
* Also removed "Future Extension" descriptions of cluster actor reference,
since that is probably not the direction we are taking
A subclass should not inherit private members of its parents.
In Scala 2.10, enforcement of this rule was conflated with the
notion of access: a private member of `class C` may be accessed
from its companion.
I brought the compiler into line with the Spec recently in
SI-7575 / https://github.com/scala/scala/pull/3440.
Two remedies exist: loosen access, or upcast to the parent
type before selecting the private member. I've chosen the former
here.
This is a touch counterintuitive ("why should upcasting ever make
new members visible?"). I double checked this change yesterday
with @odersky, and he answered:
> Yes, the fix matches the spec exactly, as far as I can tell.
> I just verified that javac behaves the same. In
>
> class C {
> private void foo() {}
> void test(D d) { d.foo(); }
> }
>
> class D extends C {}
>
> You get "cannot find symbol" on line 3.
This patch brought to you courtesy of dbuild! A backport to the
last stable release branch of Akka might be worthwhile, but I'll
leave that decision to you.
* actor name [endpointWriter] is not unique
* The problem was that the test used 100ms until ungate and
that made it possible for the endpointWriter to not be
completely terminated (and removed) before Ungate and
new Send in idle state, which created new endpointWriter
* I could reproduce it with a sleep in EndpointWriter.postStop
* The solution is to start the scheduled Ungate after Terminated
has been received
*Scala version:
Added package tag `` around scala.None and .Some[_];
Added "[_]" after scala.Some statement;
Removed japi part, because its a Scala documentation version;
*Java version
Fixed wrong path to japi.Option.Some and None
Added package tag ``
Removed scala packages (i know about implicit conversion, but i think it mght be beter)
* The documentation was good, but some parts were "hidden" by separating
it two places. I understand the original reason for the separation but
it might be easier for the user (as reported in the ticket) to have
everything in one place.
* Since there was no sender for the `Create` message in the test, the
created `ActorRef` was also sent to dead letters, creating a race
where the test could start to subscribe _too early_ and receive the
`ActorRef` instead of the expected string.
