Removing the last remnants of the async saving debacle. Async saving causes a host of extra problems within Django - order gets messed up and deleting simply doesn't work - but the scariest issue is that it seems to do nasty things with database references - deleting changes dbrefs for objects dynamically ... O_o; Plus, the overhead of creating threads for every save makes the whole thing very slow. It's NOT worth the effort.

src/utils/idmapper/base.py

@@ -1,73 +1,16 @@
 """
-Idmapper base functionality. Most of this is unchanged from the idmapper distribution. 
+This is mostly unmodified from the original idmapper. 
 
-Extended for Evennia: 
-
-- made object cache a dictionary rather than a WeakValueDictionary. The latter does not work
-  well for long-time persistence in memory and caused very hard-to-track bugs in Evennia's 
-  typeclass system (which depends on cached memory adresses not changing or going out of scope).
-- Added optional asynchronous save operation for use with Twisted. 
+Evennia changes:
+  The cache mechanism was changed from a WeakValueDictionary to a 
+  normal dictionary. The old way caused very hard-to-diagnose bugs
+  over long periods of time (which Evennia requires)
 
 """
 
-from weakref import WeakValueDictionary, ref
-
 from django.db.models.base import Model, ModelBase
-
 from manager import SharedMemoryManager
 
-#
-# Evennia extension: Asynchronous save functionality. This
-# is used by SharedMemoryBase.save(). 
-#
-
-from twisted.internet.defer import DeferredQueue, execute
-from twisted.internet.task import cooperate 
-from twisted.internet.threads import deferToThread, blockingCallFromThread
-from django.conf import settings
-from src.utils import logger
-from twisted.internet import reactor 
-
-ASYNC_DB_SAVE = settings.DATABASE_NONBLOCKING_SAVE
-
-# This special form of queue has a get() function that returns deferreds. 
-# Add save function tuples to this in order to add to save queue.
-ASYNC_QUEUE = DeferredQueue(backlog=1)
-
-def async_callback(funcdef):
-    "This callback is run with the item returned from queue - a tuple (func, args, kwargs)"        
-    #d = deferToThread(funcdef[0], *funcdef[1], **funcdef[2])
-    logger.log_infomsg("Callback")
-    def threadback(r):
-        print "Thread returned."
-    def threaderr(failure):
-        print "Thread error" + str(failure)
-    #d = reactor.callFromThread(funcdef[0], *funcdef[1], **funcdef[2]).addCallbacks(threadback, threaderr)1
-    d = execute(funcdef[0], *funcdef[1], **funcdef[2]).addCallbacks(threadback, threaderr)
-    #d = reactor.blockingCallFromThread(funcdef[0], *funcdef[1], **funcdef[2])
-    return d
-def async_errback(failure):
-    "errback"
-    logger.log_errmsg(str(failure))
-def async_worker(queue):
-    """
-    This will eternally yield items from queue as soon as they are
-    available (since DeferredQueue.get() only returns when there is
-    actually something in the queue). The queue automatically inserts 
-    the stored value into the callback async_exec. 
-    """
-    while True:
-        yield queue.get().addCallbacks(async_callback, async_errback)
-
-# the cooperate handler will iterate over async_worker every time its
-# returned deferred fires.
-cooperate(async_worker(ASYNC_QUEUE))
-
-# in models's save(*args, **kwargs):
-#   ASYNC_QUEUE.put((super(SharedMemoryModel, self).save, args, kwargs))
-
-# -------------------------------------------------------
-
 class SharedMemoryModelBase(ModelBase):
     #def __new__(cls, name, bases, attrs):
     #    super_new = super(ModelBase, cls).__new__
@@ -187,11 +130,7 @@ class SharedMemoryModel(Model):
     flush_cached_instance = classmethod(flush_cached_instance)
     
     def save(self, *args, **kwargs):
-        if ASYNC_DB_SAVE:
-            logger.log_infomsg("Adding to queue %s" % self)
-            ASYNC_QUEUE.put((super(SharedMemoryModel, self).save, args, kwargs))
-        else:
-            super(SharedMemoryModel, self).save(*args, **kwargs)
+        super(SharedMemoryModel, self).save(*args, **kwargs)
         self.__class__.cache_instance(self)
 
     # TODO: This needs moved to the prepare stage (I believe?)