Re: Label for Top Node of "triangle diagram"

Hugo,

I guess I'm somewhere between Chris and you, but perhaps closer to Chris.

	Date: Mon, 30 Sep 2002 18:22:25 +0200
	From: Hugo Haas <hugo@w3.org>
	To: www-ws-arch@w3.org
	Subject: Re: Label for Top Node of "triangle diagram"

	* Christopher B Ferris <chrisfer@us.ibm.com> [2002-09-27 16:27-0400]
	> I want to emphasize that Heather is talking about ROLES here, not
	> things, not mechanisms. A role is an abstraction as are the other
	> nodes (roles) in the triangle diagram.

Chris/Heather are right on this point.  Many kinds of things can
assume roles -- entities like people, organizations or directory
services.  Many mechanisms can be employed by each, and these
mechanisms can be centralized or distributed.  Even for the other
corners of the triangle, we don't prohibit distributed implementations
of web service requestors or providers.

	The problem with using a role with the top node is that, at least to
	me, it feels like this is a third entity different from the others,
	and registry has this feel of centralized thing. If we ended up

Perhaps this feeling is due to the statistics of the predominant
role-mechanism realizations in use today.

	settling for a role, I think that the following would need to be very
	clearly underlined:
	- the publication/discovery role is an abstract role which may be
	  fulfilled by the service provider itself.

Since the service provider is presumably the thing that knows how the
service is being offered, it obviously must be complicit in providing
that description.  But it seems equally clear that in an enterprise
the size of the web, service requestors must somehow find out about
the existence of the service.  (Even distributed broadcast mechanisms
for service requests such as those used in file sharing generally must
do some caching at internet scale -- and these caches don't belong to
the service provider.)  Typically some (possibly distributed) agency
other than the service provider takes an active part in the
matchmaking enterprise, accepting publication and/or actively
conducting discovery.  Whatever entities/mechanisms are carrying out
that activity are acting in the publication/discovery role.

	- there is no claim made as to how it is performed: it could be
	  achieved using any number of solutions, be it centralized or not.

Certainly.

	Ideally, the word used would carry all that, which IMO registry does
	not. Hence my preference is to use the description of an action:
	Publish/Discover, or Publication and Discovery Mechanisms, or
	Advertizing and Discovery Mechanisms to quote DaveH.

I don't like the move to an action or a mechanism.  I think 'registry'
is bad because it tends to connote mechanism.  This is why we labeled
it 'Discovery Agency' at the f2f, because it sounded more like a role.
Or how about 'Publisher/Discoverer'?  That is similar to one of your
options, but rolifies it.

--mark

Mark A. Jones
AT&T


	Regards,

	Hugo

	-- 
	Hugo Haas - W3C
	mailto:hugo@w3.org - http://www.w3.org/People/Hugo/

Received on Monday, 30 September 2002 15:29:09 UTC