Re: [syndication] Is a Feed the right place for your Data?

[Ken MacLeod <ken@bitsko.slc.ut.us>]

Rick Bradley <roundeye@roundeye.net> writes:

> I'll probably keep pushing this notion wherever it comes up (so
> forgive me if you've seen this before), but these syndication
> formats make good encapsulation formats, but they start to seem
> cumbersome because we only think about content that is tied down to
> an URL.

I agree with you completely about a widely held *assumption* that a
URI physically exists at one location.

Where I disagree is that a URI has to not "look like" a retrievable
resource to indicate that it can somehow be retrieved from some other
location.

RDF handles this very well, for example.  Each subject-URI can have
properties associated with it from any physical location, not
necessarily the "location" given in that URI (assuming it's a
location, anyway).

In the general sense, a URI is simply an identifier.  In some cases, a
URI is retrievable, or "web accessible".  There is no where in web
specs that says that a resource can *only* be resolved using the "URI
scheme" given in the URI, in fact they go out of their way at times to
suggest otherwise.

The approach you're looking for is not about the identifier, the URI.
You're talking about access, or dereferencing the URI[1].

When you take away any perceived depedence on the URI scheme, ideas
and solutions for dereferencing URIs start popping out of the
woodwork.  HTTP Proxies are a great example of this, even if now they
seem to be hard-wired into a browser's access pipeline.  Many HTTP
Proxies will accept a variety of URI schemes for resolution (principle
among them FTP).  If HTTP Proxies were maybe relabled "web caches",
we'd have more use of them as such.

IIRC, the FreeNet P2P network is structured around using URIs for
identifying resources, but the P2P network for indexing and locating
them.

Many XML processing systems are beginning to include caches (hard or
soft) for schemas, DTDs, stylesheets, etc -- you ask for a URI, the
XML processing system first checks to see if it already has it before
requesting it over the web.

In the context of syndication, it would be wonderful if the
super-aggregators offered URI resolution.  Ask the super-aggregator
for a URI (particularly, the identifier of a feed or even an item),
and get back data about that resource.

  -- Ken

[1] http://www.w3.org/2001/tag/webarch/#dereference-uri