ideas: the REST of HTTP & the SPARQL of GET

Larry Masinter points out that the HTTP spec is really two beasts: An
operational, testable portion dealing mostly with syntax of messages
and with caching, couched in MUST language, and a separate REST level
dealing with "resource state" that would not be testable (falsifiable)
by someone observing an exchange since the only way to know whether
it's being followed would be to look inside the server.  I'm thinking
this second level, which I've been avoiding, could be given its own
class and properties, and that the treatment of this part of the HTTP
spec might be done instead of (or coincident with) a treatment of REST
as described by Roy.

The class of REST-compliant resources is what HTTP calls a "network
data object" and it would be a proper subclass of "HttpResource".
Everything else is just a "network service".

My other recent idea:

We are suggesting that some response messages can be translated into
RDF. In the same way we should be able to say that a request message
(such as a GET) can be translated into SPARQL, or something very close
to it.

Jonathan

Received on Tuesday, 10 November 2009 07:39:10 UTC