RE: Discussion of 2.2.11 Intermediary

inline.

> -----Original Message-----
> From: Hugo Haas [mailto:hugo@w3.org]
> Sent: Monday, July 07, 2003 6:34 AM
> To: www-ws-arch@w3.org
> Subject: Discussion of 2.2.11 Intermediary
> 
> 
> 
> As per my action item, this email introduces discussion of the 2.2.11
> Intermediary concept.
> 
> -=- Current situation -=-
> 
> The current text is[2]:
> 
> |    2.2.11 Intermediary
> |
> |      2.2.11.1 Summary
> |
> |   An intermediary is a message processing node that does 
> not necessarily
> |   represent the message's intended recipient; but which, 
> none-the-less
> |   processes some aspect of the message.
> |
> |      2.2.11.2 Relationships to other elements
> |
> |   an intermediary is
> |
> |           a agent
> |
> |      2.2.11.3 Description
> |
> |   Intermediaries process messages that are intended for 
> other recipients.
> |   An intermediary may act as a gateway to bridge transport 
> services, or may
> |   process specific aspects of messages (such as security 
> information).
> 
> -=- Discussion -=-
> 
> The discussion about intermediary started in the "Message Recipient
> 2.2.26 & Sender 2.2.27 text" thread[1].
> 
> Rereading Shishir's email, my original proposal[3] doesn't work: an
> intermediary may or may not be one of the recipients of the message,
> and an intermediary doesn't *originate* a message.


yes, agreed.

> 
> Yin-Leng, I think that you are partially right in your email[4] to say
> that an intermediary is not the sender's intended recipient: it is not
> the intended ultimate recipient, i.e. one expects further processing.
> 

Further, the sender my not know of the intermediary as a possible recipient
of a message at the time of sending.

> I think that the confusion comes from the word "intended", and the
> concept of final recipient which isn't reflected in our concepts.
> 
> The section on SOAP intermediaries in the SOAP 1.2 specification[5]
> gives us a good direction about how to model them.
> 
> An intermediary:
> - is an agent which processes a message.
> - is not the ultimate recipient of the message, i.e. it is not the end
>   of the message path.
> - may or may not be an intended recipient for the message, i.e. a
>   message recipient as per 2.2.26; an example of an intermediary which
>   could be not an intended recipient could be a transparent SOAP proxy
>   or firewall.
> 
> Comment about 2.2.26 Message recipient:
> 
> In Shishir's proposal[3], the description 2.2.26.3 talks about "the
> message recipient is the agent that the sender intends the message to
> be consumed by", i.e. not allowing several message recipients for a
> message. It seems to me that this captures the semantics of what SOAP
> calls the ultimate recipient. Moreover, I am wondering if "the"
> instead of "a" doesn't prevent multicast.
> 

I don't know if using "a" rather than "the" is enough to bring out the
possibility of multi-casting. If more people think it's enough, we can use
"a" else we can reword that sentence to read:

"The message recipient is a agent of possibly multiple agents that a sender
intends the message to be consumed by".


> As said in my original email[3], I would remove all mentions of
> intermediary in 2.2.26 and 2.2.27.
> 

The reason I thought it was important to mention intermediaries in 2.2.26 &
2.2.27 is that I think that these "concepts" have a "relationship" that we
need to mention. In my mind, the intermediary is a combination of a message
sender and recipient. 

> The current summary talks about "message processing node". This is the
> only place where the concept of a processing node appears in the
> document, so I have replaced it by agent.
> 
> -=- The proposed changes -=-
> 
> Note: concepts are surrounded by "_".
> 
> Proposed 2.2.11.1 Summary
> 
>   An _intermediary_ is an _agent_ which processes part of a 
> _message_. An
>   _intermediary_ is not the ultimate _message recipient_ of the
>   _message_, and relays the message to the next _message recipient_
>   along the _message path_.
> 
>   Note: I have introduced a _message path_ concept.
> 
> Proposed 2.2.11.2 Relationships to other elements
> 
>   an _intermediary_ is
>     an _agent_
> 
>   an _intermediary_ may be
>     a _message recipient_
> 

I think of this in reverse: A _message recipient_ may be an _intermediary_ 


>   Note: if the above statement looks odd, we may have to take
>   "intended" out of _message recipient_.
> 
> Proposed 2.2.11.3 Description
> 
>   Intermediaries process messages along the message path. A message
>   may be intended for an intermediary, or may be transparently
>   processed by one.
> 
>   An intermediary may act as a gateway to bridge transport services,
>   or may process specific aspects of messages (such as security
>   information).
> 
> -=- Last comments -=-
> 
> Another way to approach intermediaries is to introduce 2 other
> concepts[4] as proposed by Yin-Leng:
> - message originator.
> - message receiver.
> 
> We would have:
> - a _message originator_ is an _agent_ originating a 
> _message_ (which is our
>   current definition of _message sender_).
> - a _message sender_ is an agent sending a _message_.
> - a _message receiver_ is an _agent_ receiving a _message_.
> - a _message recipient_ is an _agent_ intended to receive a _message_.
> 
> We have the following relationships:
> - a _message originator_ is a _message sender_.
> - a _message recipient_ is a _message receiver_.
> - an _intermediary_ is a _message receiver_ and a _message sender_,
>   and may be a _message recipient_.
> 
> This increases the number of concepts but may make things clearer.
> 
> Comments?

I like this to the extent that it brings out the relationship between the
sender and recipient (and is much better than the existing text), but not
sure if we need 2 additional "concepts" to be defined just to distinguish
the "intent". I would prefer to remove intended from the _message recipient_
and relate it to _intermediary_ directly. 

> 
> Regards,
> 
> Hugo
> 
>   1. http://lists.w3.org/Archives/Public/www-ws-arch/2003Jul/0005.html
>   2. 
> http://dev.w3.org/cvsweb/~checkout~/2002/ws/arch/wsa/wd-wsa-ar
> ch-review2.html?rev=1.31&content-type=text/html#intermediary
>   3. http://lists.w3.org/Archives/Public/www-ws-arch/2003Jul/0008.html
>   4. http://lists.w3.org/Archives/Public/www-ws-arch/2003Jul/0036.html
>   5. http://www.w3.org/TR/2003/REC-soap12-part1-20030624/#relaysoapmsg
> -- 
> Hugo Haas - W3C
> mailto:hugo@w3.org - http://www.w3.org/People/Hugo/
> 
>