[
Lists Home |
Date Index |
Thread Index
]
On Thu, Feb 27, 2003 at 02:13:40PM -0600, Bullard, Claude L (Len) wrote:
> It would seem to me that such a registry should
> be part of the metadata of a site (see Tim Bray's
> recent contribution to TAG on sites) and not a
> separate database as such. That way interoperation
> falls out of the standard web architecture and
> possibly, the semantic web.
That's not different than Google :-)
The service is less generic, it's convenience and
not normative. The first thing if you look for
a namespace name semantic is to try to open the
given URL in a browser, then looking in such a database
is a way to try to gather less normative informations.
For me it's exactly parallel to rpmfind services
someone looks for software foo version 1.2, don't find
it in it's harddrives or CDroms and will look it
up in the database to either download it or find a pointer
to the project site. If it's just metadata gathering,
and not a attempt to define those data, I hardly see
the problem.
> I contend that such
> a registry, particularly for government systems,
> will require related assertions with regards to
> the responsibilities of the registrar.
I understand that this database will be based on
the namespace name. I own veillard.com at the DNS level,
I can prove it for a given timeframe, and using the whois
database that can be checked. At least for namespace names
based on DNS informations some checking can be done.
On the other hand I would be really annoyed is I were to
define a namespace name and find that someone else did that
before. Actually the uniqueness should be guaranteed for
[namespace_name , start_date, end_date] tuples, and clashes
at the namespace name level simply indicate that this database
should be purely informative :-)
Daniel
--
Daniel Veillard | Red Hat Network https://rhn.redhat.com/
veillard@redhat.com | libxml GNOME XML XSLT toolkit http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/
|
