> > There are a few ways to test if an endpoint is local: > > * A flag in the endpoint: The flag is set to true upon the publication of > the endpoint. If an endpoint is propagated from other nodes in the domain, > the flag is set to false. > * Check against the structural URI of the endpoint to the "fragement" of the > domain composite that is loaded by the node to see if it is local > * A node URI in the endpoint: The URI is set when the endpoint is published > to the registry and it is part of the endpoint description when the endpoint > is serialized for replication. >
The fist of these seems the most straightforward. Currently we could do it in the builders when the endpoint references are first constructed as we know then what endpoints are present in each node. However I'm not sure that this helps us out in future potential dynamic scenarios. Simon
