RE: Hypermedia vs. data (long)

> -----Original Message-----
> From: Newcomer, Eric [mailto:Eric.Newcomer@iona.com]
> Sent: Wednesday, January 01, 2003 11:49 AM
> To: Miles Sabin; www-ws-arch@w3.org
> Subject: RE: Hypermedia vs. data (long)
> 

I just made a New Year's resolution to NOT spend 2003 focused on the REST vs
SOAP dispute.  If there is another one of these threads on New Years Eve
2003, I'm not going to be accountable for my actions :-)

But seriously, Eric has a very good summary IMHO:
 
> A lot of this boils down, as we've discussed before, to 
> whether or not semantic information can be contained in the 
> documents being exchanged, or at least whether or not 
> pointers to such semantic information can be contained within 
> the documents being exchanged.  REST would say no, but 
> extensions to HTTP might well say yes, as they have.

The Web as we know it *is* primarily used in an environment where the
semantics of the content are encoded and decoded inside human heads;
traditional RPC and EAI is primarily used in an environment where the
semantics are encoded in program code and assumed to be shared by the
implementers of the objects and the programs that invoke methods on those
objects.  I'll submit that neither the REST paradigm or the RPC/distributed
objects/traditional SOAP/whatever paradigm has much of anything to say about
semantics per se.

Wearing  both my co-chair and my Software AG hat, *both* these perspectives
are perfectly reasonable: there is a lot of application code out there that
can be effectively integrated using XML, HTTP, SOAP, WSDL, etc. simply to
overcome the mechanical problems of integrating software whose semantics are
understood by both sides across platform/vendor/language boundaries; and
there is a lot of value in extending the design pattern of the Web to
machine-to-machine communication when the  semantics of the "documents"
being exchanged are understood by both sides.  I fail to see the religious
:-) signficance of deciding whether the shared semantics must be encoded as
methods on objects or content of documents!  For whatever its worth, so long
as one uniquely identifies either the objects/methods/paramaeters or
specific documents/messages with URIs, then the Semantic Web (or any other
knowledge representation and inferencing system) can "reason" about either.


Mark, to be blunt, I think the Web Services world has moved a lot farther
and faster in the last year toward acknowleding the points that REST
advocates have made and incoporating them into the various specs than the
RESTifarians have come in acknowledging that y'all don't have a stranglehold
on Truth either. One CAN model almost anything as distributed objects with
methods to be remotely invoked, and one CAN model almost anything as
hypermedia resources with representations to be moved around and with states
identified by unique URIs. I for one am much more interested in
understanding the circumstances under which one or the other is most
appropriate, and how they can be combined in fuzzy intermediate situations.
I really, really, really hope we can move forward in this permathread so
that both perspectives are reflected in the WSA document and that this leads
to an understanding of the architectural best practices here.

Received on Wednesday, 1 January 2003 14:16:47 UTC