W3C home > Mailing lists > Public > www-ws-arch@w3.org > July 2002

RE: A REST challenge

From: Darran Rolls <Darran.Rolls@waveset.com>
Date: Wed, 17 Jul 2002 16:28:32 -0500
Message-ID: <6244FCD1F88EC14BACDD2A319FD338242DC736@hawaii.waveset.com>
To: "bhaugen" <linkage@interaccess.com>, <www-ws-arch@w3.org>
 RESTafarian  - lol
 
Little thing much this such good fun :-)

	-----Original Message----- 
	From: bhaugen [mailto:linkage@interaccess.com] 
	Sent: Wed 7/17/2002 3:55 PM 
	To: www-ws-arch@w3.org 
	Cc: 
	Subject: Re: A REST challenge
	
	


	Francis and all,
	
	> I have my doubts about the REST architecture, not as a valid
	> technological solution but from a religious POV (i.e., it seems to me
	> that RESTers are a pretty religious lot)
	
	I am not a religious RESTafarian.  ebXML, in which I participated,
	uses message-oriented SOAP.
	
	However, when we got to the next level - orchestrating
	business conversations - business semantics became
	much more important.  We found that the simplest and
	most declarative method of orchestrating business conversations
	was around the states of semantic business objects,
	which could very well be implemented as REST resources.
	
	> As to my preferred approach, you need the notions of an agent, a
	> conversation, communicative acts, a contract, a social/business
	> relationship before you can properly model the normal business cycle.
	
	My point is that a lot of those concepts, as well as those in
	"the normal business cycle" are well-modeled as nouns, that is,
	REST resources.  It's more natural for contract to be a
	REST resource than it is to be a correlation-id.
	
	Those conversations are *about* something.
	Contracts, orders, deliveries, products, payments, etc.
	
	ISO Open-EDI Rule 1 says something like "you're
	not just exchanging data, but making business
	commitments".  Those business commitments,
	and their fulfillments, are the residual effect of
	the messages being sent back and forth, and
	can even be the destination of messages.
	I would think that modeling them as first-class
	persistent resources would meet your semantic
	goals better than having them be second-class objects.
	
	I'm not really trying to sell you anything here.
	Just wondering why a semantic modeler
	wouldn't love REST.
	
	-Bob Haugen
	
	
	
	
	

Received on Wednesday, 17 July 2002 17:29:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:02 GMT