It would probably be a nicer internal cleanup to use a shared token and its reference count to establish the existence of a high prio message. This way, the count is atomic, and automatically incr/decr. A copy of the actor's token needs to go in the high prio message.
I hope that this will be more natural vs the current hack.
It would probably be a nicer internal cleanup to use a shared token and its reference count to establish the existence of a high prio message. This way, the count is atomic, and automatically incr/decr. A copy of the actor's token needs to go in the high prio message. I hope that this will be more natural vs the current hack.