RE: Spec draft - daveh comments

+1.

Also, in terms of the "description" [point b) in DaveH's message], besides
abstract and concrete, we may want to differentiate content in the
description used to advertise with discovery services and content in the
description that gives full details of how to interact with the (desrcibed)
service. In DAML-S, we call the first the "service profile" that is
advertised and used for requestors to discover (or be discovered)
potentially multiple providers with similar desired capability. After
discovering multiple providers, the requestor can select among the
discovered providers and go on to interact with the selected provider (or
even employ all the providers in parallel, if you think of an information
service, such as various weather providers). To perform this interaction,
the requestor accesses the more detailed description of the service (the
service/process model).
 I think a distinction such as the above is useful to make.
For example, a provider may be willing to publicize a service profile, but
unwilling to publicize all the details of a process model.

 This brings me finally to another point, namely, that I do not think that
contract and service agreements belong to the
public description either, since (a) different contracts may hold between a
provider and its various requestors, and (b) a provider may not want to
publicize details of contract agreements.

   Cheers, Katia

-----Original Message-----
From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org]On
Behalf Of Dave Hollander
Sent: Monday, October 07, 2002 2:41 PM
To: www-ws-arch@w3.org
Subject: RE: Spec draft - daveh comments




First, many thanks. It is good to have some wording to ground this
discussion. here
are my specific points/questions.


1) publish vs. advertise
	For many reasons[1] I prefer advertise. Can we change?

	[1]
http://lists.w3.org/Archives/Public/www-ws-arch/2002Oct/0037.html


2) Description
	a) In section 1.2 you state: "A web service is described using a
standard, formal XML notation..."
	Could it not also be described over the phone using Greek, Latin or
English?

	Later on you use "may be realized as a set of XML description
documents" This seems
	more accurate.

	b) Later in the paragraph you state: "provides all of the details
necessary to interact with the service..."
	Not really. It can not describe pre and post-conditions, the
business or legal context etc.
	It seems worth while to break description down into at least three
pieces: concrete, abstract and
	service level where WSDL can provide concrete and abstract
description data.

	c) In "Artifacts of a Web Service" the last sentenance of the web
services description bullet says
	that a "description may be published to requestor directly or a
discovery agency".

	Do we want to say that or do we want to say that at times the
requestor acts as the discovery agency
	and requests the description directly? This says that even though
the exchange is direct, there is
	still a distinction of roles.


3) Discover Services
	I like this term. In fact, I think having it plural makes the need
to incorporate advertising/publishing
	clear enough.

	But, in the section "Roles in a Web Service Architecture" you switch
to "Service Discovery Agency"
	The text works for me, the term used is singular and different than
the graphic.

Regards,
DaveH

Received on Monday, 7 October 2002 16:15:53 UTC