RE: Web sites as resources

I agree entirely with your analysis and general conclusions. But is anything
new actually essential beyond the assertion of there being an identifiable
resource called Site?

There are many alternatives for 1 and 2 below, but I think conflict between
them would be the exception rather than the rule.

> All that's really left to argue about is
>
> 1. How to express the membership relation between resources and sites

Personally I liked the http header idea, where

Site: http://example.org/site

would assert that what was being served was a representation of a resource
that was a member of the identified web site.

> 2. What agents can expect to find in representations of sites

The primary representation is already there - the site itself.

As in the case of namespaces, there may be no need for there to be any
http-retrievable representation of the "Site" URI, but a manifest listing
key parts of the site could be useful. That manifest could be expressed in
any of the description languages on offer (with suitable mime type) : RDF,
XLink, Atom etc.

> 3. Whether the work required to agree on #1 and #2 is cost-effective.

If a single form of representation is mandated, I suspect not. But for the
acceptance of the notion of a URI-identifiable notion of web site, the cost
may be significantly lower.

Cheers,
Danny.

Received on Friday, 16 January 2004 06:59:35 UTC