Registry architecture and id

[Marco Leoni]

I would add two ideas at the ongoing discussion.

- First is about the implementation of the registry: what about to use 
Jini architecture in order to create a distributed environment where 
each entity can plug-in itself?
Jini architecture will provide a network of services with a look-up 
facility, exactly what we need for the registry (more info at 
http://www.jini.org).

- Second thing is about ID.
As suggested each entity of the registry will probably have a different 
way-of-sameness, so we can think of  ID as the super class with several 
children, one for each type of sameness:

              |-----------------|
              |           ID           |
              |-----------------|
                          |
                          |
                         /\
                          |
          
 -------------------------------------------------------------------------------------...
           |                                                   |         
                                          |
|-----------------|                      |-----------------|             
         |-----------------|
| ObservationID  | --------------|       FileID         
|---------------|    ServiceID      |-------...
|-----------------|                      |-----------------|             
         |-----------------|

In this way it could be possible to build something that implements 
different types of sameness-checking related to the entity type.




Cheers,
    Marco