twisted.spread.pb.RemoteCache(RemoteCopy, Serializable)class documentation
twisted.spread.pbView Source (View In Hierarchy)
Known subclasses: twisted.spread.publish.RemotePublished
A cache is a local representation of a remote
This represents the last known state of this object. It may also have
methods invoked on it -- in order to update caches, the cached class
to this object as it is originally sent.
Much like copy, I will be invoked with no arguments. Do not implement a constructor that requires arguments in one of my subclasses.
|Method||remoteMessageReceived||A remote message has been received. Dispatch it appropriately.|
|Method||jellyFor||serialize me (only for the broker I'm for) as the original cached reference|
|Method||unjellyFor||Perform the inverse operation of
|Method||__cmp__||Compare me [to another RemoteCache.|
|Method||__del__||Do distributed reference counting on finalize.|
|Method||_borgify||Create a new object that shares its state (i.e. its
A remote message has been received. Dispatch it appropriately.
The default implementation is to dispatch to a method called
observe_messagename' and call it on my with the same
Compare me [to another RemoteCache.
Do distributed reference counting on finalize.
Create a new object that shares its state (i.e. its
__dict__) and type with this object, but does not share its
This is an instance of the Borg
design pattern originally described by Alex Martelli, but unlike the
example given there, this is not a replacement for a Singleton. Instead,
it is for lifecycle tracking (and distributed garbage collection). The
purpose of these separate objects is to have a separate object tracking
each application-level reference to the root
object being tracked by the broker, and to have their
methods be invoked.
This may be achievable via a weak value dictionary to track the root
instances instead, but this implementation strategy predates the
availability of weak references in Python.
|Returns||The new instance. (type: |