Re: Issue 133, and permitting no body

> +1, exactly!  I think that's why GET can only "get" you so far in modeling 
> the whole world.  If you've got a human to figure out details of the 
> contract, you can (appear to) leave a lot of it unspecified.  To just tell 
> machines to GET and PUT/POST each others properties, with no tighter 
> contract than that isn't likely to work.

I'd like to focus on "with no tighter contract than that".  What do
you mean by "contract" here exactly?  From your multiplication example,
you would appear to mean that you require certain behaviour of
resources, exposed as new methods by those objects.  Is that a fair

If I did a GET on (with "Accept:
application/rdf+xml"), and it returned an RDF document which asserted
that it was of class, which I knew
about, is that enough of a contract?  Or if I did a similar GET on and found an RDF document which
described what a multiplier is (not *how* it does it) in terms that I
understood (and if I didn't understand them, I could keep GETing links
until I did, like looking up words in a dictionary to determine the
meaning of a sentence), is that enough of a contract?

GET/PUT/POST/DELETE is sufficient agreement over the wire, because it
comprises an application that permits extended agreement to be made
within itself.  This is hypertext, and the Semantic Web exists to make
the extended agreement currently available to humans (because we share a
lot of context with other humans), also available to machines.

At the very least, as I said before, the Semantic Web and RPC are both
computationally complete.  If you don't buy my arguments, you might
want to at least consider that TimBL's vision for the future of the
Web is the Semantic Web.  Web services is not something he had in mind
at all.  Hear what he has to say (or what he doesn't say) here,
including the Q&A session at the end;

Mark Baker, Chief Science Officer, Planetfred, Inc.
Ottawa, Ontario, CANADA.

Received on Wednesday, 6 February 2002 22:26:21 UTC