Re: FW: Model question

Kevin Thomas wrote:
> Hello Kevin!
> 
> It turns out that my subscription mysteriously disappeared, so I don't think
> this got through before.
> 
> Thoughts?

This has been a fascinating discussion.  I still haven't drawn a 
conclusion on the whole matter, there have been interesting arguments on 
both sides.  If anything, it says there is value in both models and 
therefore both should be supported.  In Z39.50 terms, for example, 
perhaps one could determine whether the "treat result set as a database" 
option is available for a specific database via Explain, and perhaps 
that could be implemented by enabling (valid) Result Set Names to be 
used interchangably with database names during a session, or in 
subsequent sessions assuming persistent result sets.

I think some of the other problems you mention, such as database 
branding, last update, etc, are (in theory) naturally handled by 
identifying the resulting "database" as a derivative work of the 
original databases.  I would think one could always identify the 
original source(s) for a record/result set, perhaps via a stack trace 
sortof mechanism instead of simply the database name per record. 
Charging models, etc. might pose an implementation challenge, but I 
think it's possible and I don't believe that should be a limiting factor 
for the underlying data model.

Kevin

-- 
Kevin Gamiel <kgamiel@cnidr.org>
MCNC Center for Networked Information Discovery and Retrieval (CNIDR)
http://www.mcnc.org
http://www.cnidr.org

Received on Tuesday, 5 August 2003 09:23:31 UTC