IDs: plan to conclude discussion

[Ray Plante]

Hi NVO-MWG-ers,

This week, I would very much like to wrap up the discussion on ID
requirments and scope, which will require a bit of focus during our
Thursday telecon.  The next step is to draw up a specification.  
Of course, comments are welcome throughout the prototypeing stage 
(including from the IVOA community).

Direct comment on the requirements have been sparse, so I'm going to
suggest my own evaluation of the list based on the discussion thus
far.  In short, I recommend restricting an ID's functions to testing
whether two identifiable objects are the same and to access unique
metadata descriptions of objects from a registry.  All other information
regarding the object (derivation, ownership, etc.) is obtained by
accessing the metadata descriptions.

In a bit more detail, the major changes I recommend to the requirements 
(refering to the latest version at
http://rai.ncsa.uiuc.edu/~rplante/VO/metadata/oidreq.txt) are:

  * ADD 1b: support a notion of local & global identifiers where
	local identifiers are unique only within a give context and
	global ones include a namespace.

  * DROP 5a in favor of 5b: 
        that is, a global ID is not required to identify the curating
        resource, but rather the resource that issued the ID.

  * ADD 5c: ID can be used to access unique descriptions of objects
        from a registry.

  * DROP 6: an ID for a data object is not required to indicate the
        collection it belongs to.

  * DROP 8: an ID for a data object is not required to indicate the
        collection it is derived from.  

There remains the issue of "sameness".  My suggestions for this will
appear in a subsequent message.

cheers,
Ray