Re: mid-course correction on abstract model for module processing

Thanks for the comments, Yuhichi!

	Subject: Re: mid-course correction on abstract model for module processing
	To: Mark Jones <jones@research.att.com>
	Cc: xml-dist-app@w3.org
	From: "Yuhichi Nakamura" <NAKAMURY@jp.ibm.com>
	Date: Sat, 17 Mar 2001 10:18:50 +0900

	Hi Mark,
	Great job!  I have some questions and comments.

	Module or Handler?:
	In Stuarts Williams's abstract processing documents, module and handler
	are explicity distinguished.  A module is a collection of blocks, and
	a handler is a component which processes module(s).  I may just memorize
	the terminology wrongly.

The most recent gloassay has a "module" encompassing the definition of
the a set of related blocks and their processing rules, realised in
one or more handlers.

	Actor:
	In SOAP, an actor targets an "application" in my opinion because there is
	no
	handler concept.  Once we introduce the handler concept, should we
	introduce
	two kinds of actor attribute, i.e. applicationActor and handlerActor?
	I am not sure how handlers and applications are properly targeted with
	a single Actor attribute.

I think the recent discussion indicates that most people want to think
of targetting a block at a handler (a chunk of code) rather than a
module -- the semantically broader, more abstract concept.  I've come
to the conclusion that it probably doesn't matter much.  You could
always define a "handler" at a suitably abstract level so that it did
all of the interesting encapsulation of the mapping between blocks
and functionality.


Something seems to have happened to your text below, after
"discussion", and I can't interpret it.

	Removal of Blocks:
	We had a long discussion on this in Apache Axis project.  In the
	discussion,
	I always a digital signature usecase.  Assume that there is a message that
	includes SOAP digital signature header according to W3C SOAP Security
	spec, and it is sent from a one company to another company.  The receiver
	company wants verify the signature, and "log" the singed message only
	when valid.  There are some ways of system configuration within the
	receiver side, i.e. (1) a single SOAP application, (2) a Gateway (SOAP
	intermediary)
	and an ulitmate destination, and so on.  In this scenario, the sender does
	not
	care (even should no know) whether the signature header is logged.
	I am not sure if giving "non-targeted" is responsibility of the sender.

	Regards,

	Yuhichi Nakamura
	IBM Tokyo Research Laboratory
	Tel: +81-462-73-4668


	From: Mark Jones <jones@research.att.com>@w3.org on 2001/03/16 05:26
        ...

Received on Monday, 19 March 2001 10:07:13 UTC