Lists Home |
Date Index |
> It's even worse than this. I don't think anyone *ever* uses RDF to "add
> meta-data to some existing data (also called a resource)". It just
> can't work this way. The data and the resource are separate, and you
> *always* talk about the resource in RDF, and *not* about the data.
> They really are completely separate.
I didn't meant "adding" in the sense of including the RDF statements into
the data but since you are it, yes they (the rdf statements) "may" be
included into the data, for example, included into an XHTML document. There
is nothing that ever said in the specs that RDF statements should be in a
> Note that you might even want to use different representation predicates
> for different user agents, and so on. Obviously in practice nobody
> shares resource representations this way, since we have HTTP and it
> works fine. But the point is that *architecturally* and *conceptually*,
> the result of HTTP GET is just as much "about" the resource as any other
> RDF statement.
I agree. I experiment this every day by producing different data
representations from the very same domain model (i.e. data set), in that
case, different data textual serialization. If I interact with an RDF server
through an HTTP GET I may also get a representation on some unknown internal
representation (a Grove? an LDAP directory? a relational DB? a gizmo?). By
the way speaking of representation I recently found that XML is a great
vehicle to serialize objects (class based or prototype based), even better
its inherent REST architecture allows us to support subject oriented
programming by showing different representations or subjective views on an
object collection. Even better, the transformation characteristics of XML
allows me to practice MDA by transforming a subjective view on a model (but
still a domain model) into an interactive model (a web apps running entirely
on the client). Thanks you REST and XML framework, I wasn't able to that as
easily with older technologies.
Didier PH Martin