Re: Searching OO Associations with RDBMS Persistence Models
Date: Mon, 19 Jun 2006 17:08:10 +0200
> Dmitry A. Kazakov wrote:
>> On 16 Jun 2006 13:03:30 -0700, Aloha Kakuikanu wrote:
>>> Then logical connection is a
>>> path in the graph.
>> No. That's routing issues. A connection is either peer-to-peer or
>> multicast. It does not mention any nodes passed. Some protocols use dynamic
> > OK. Now please explain what RM meant by > "A logical connection between two endpoint is a list of circuits"
>>> There is nothing challenging in modelling graph in
>>> RDBMS and creating application objects that correspond to paths in a
>> Incidence matrix? Huh.
>>> Why "object id" is significant in this picture? |
>> Because an object can be mobile. Connection can obviously be described by a
>> trivial relation between nodes. But it is an implementation detail in, say,
>> a middleware, which deals with objects, not connections.
> > OK. Now please explain why this elusive "object id" can't be a key of > some tuple. I even suggest a name for this relation: ConnectionNode.
That's the same. You can call it key if the relation is global. But in real frameworks the scopes of object names / identities aren't global, for safety and reliability reasons. No doubts, relations can be used to describe a state of a network
But what about transitions between them?
Ii cannot serve as a basis for any reasonable implementation. Reasonable means: fits minimal functional and non-functional requirements.
The major problem, as I see it. is that when RM is used as you did, it is to the network as a whole, which clashes with the idea of network as a distributed computing system. So a technically correct description of, isn't very helpful.
-- Regards, Dmitry A. Kazakov http://www.dmitry-kazakov.deReceived on Mon Jun 19 2006 - 17:08:10 CEST