debugging data on actors

gistfile1.diff

diff --git a/xudd/hive.py b/xudd/hive.py
index e9bc0d0..3987aed 100644
--- a/xudd/hive.py
+++ b/xudd/hive.py
@@ -67,6 +67,7 @@ class HiveWorker(Thread):
             # We didn't do anything this round, oh well
             return False
 
+        print("%s got actor %s" % (self, actor))
         # Process messages from this actor
         messages_processed = 0
         while self.max_messages is None \
@@ -85,6 +86,7 @@ class HiveWorker(Thread):
             actor.handle_message(message)
             messages_processed += 1
 
+        print("Is %s in queue already?: %s" % (actor, actor in self.hive.actor_queue.queue))
         # Request checking if actor should be requeued with hive
         self.hive.request_possibly_requeue_actor(actor)
 
@@ -105,6 +107,7 @@ class Hive(Thread):
 
         # Actor queue
         self.__actor_queue = Queue()
+        self.actor_queue = self.__actor_queue
 
         self.num_workers = num_workers
         self.__workers = []
@@ -224,6 +227,7 @@ class Hive(Thread):
                     # Should we requeue?
                     if not actor.message_queue.queue.empty():
                         # Looks like so!
+                        print("requeueing %s" % actor)
                         self.queue_actor(actor)
 
             elif action_type == "queue_message":

Curiously, the above is in python 3. In python 2, the "in queue already" problem doesn't appear to be so:

Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False
Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False
Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False
<HiveWorker(Thread-6, started 139958060828416)> got actor <__main__.DepartmentChair object at 0x7f4a8d269310>
42b56fc7-106e-42d0-8026-534e13ee9328 experiment is done
<HiveWorker(Thread-2, started 139958163937024)> got actor <__main__.DepartmentChair object at 0x7f4a8d269310>
641bf397-f1cd-4f56-8b3b-2412f9f06aa8 experiment is done
Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False
Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False
<HiveWorker(Thread-5, started 139958069221120)> got actor <__main__.DepartmentChair object at 0x7f4a8d269310>
7748640a-65d9-4f19-95f3-10368468a9a7 experiment is done
<HiveWorker(Thread-2, started 139958163937024)> got actor <__main__.DepartmentChair object at 0x7f4a8d269310>
85724120-9703-4beb-806c-4d4e50b85d4d experiment is done
Last experiment message (f3d7eeff-5fb1-408b-85f9-594843d5b182:2200) received, shutting down
Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False
Is <__main__.DepartmentChair object at 0x7f4a8d269310> in queue already?: False

note that there's no block=True in this verison.
The python 2 version doesn't seem to be hitting the same bug!

However, something still seems wrong: it doesn't make sense that the DepartmentChair (of which there is only one) would be being wrapped up twice after the "shutting down" message, unless this is some threading goofiness in python and it's just printing out of order. At the time that the last experiment is received, that message should only appear one more time and then it should be over and done with.