RE: SOAP actor model

Hi Mark,

> -----Original Message-----
> From: Mark Nottingham [mailto:mnot@akamai.com]
> Sent: 20 March 2001 16:47
> To: Williams, Stuart
> Cc: 'frystyk@microsoft.com'; 'Mark Jones'; xml-dist-app@w3.org
> Subject: Re: SOAP actor model
> 
> Yes. I'm saying that routing doesn't need to be provided by the XMLP
> layer, or directly available to an XMLP Intermediary, to make them
> useful, as this service is often provided by other layers.
> 
> One of the most common attributes of intermediaries at a particular
> layer is that they coincide with intermediaries at one or more other
> layers.
> 
> Cheers,

So I think what your saying is that we don't have a solution/proposal on how
to do routing at the moment and that perhaps we shouldn't be looking to
limit ourselves to just one approach in any case. And that's fine. I can
conceive of a world in which there are multiple ways to make these choices.
In the IP world we have routing tables build from RIP, OSPF, CIDR, and/or
with administratively installed routes. 

I was trying to answer Henrik's question...

> Are there any scenarios that this model doesn't cover regarding targeting?

...mostly in from an agreeing point of view. The point being however, that I
think I at least (bear of little brain) and possibly others have difficulty
with this stuff, because faced with a message say at an XML protocol
intermediary, we have no (defined) algorithm, method whatever, to determine
who/what should get the message next. On a coarse grain I think this is a
similar problem to that being discussed in respect of selecting handlers to
process blocks.

Without even one particular model of how to choose where to send something
next it gets a bit awkward to answer a question that is begging the answer
"...there are no scenarios that this model of targetting doesn't cover."

I'd be interested to know of there has been any deployment of SOAP based
solutions that involve SOAP intermediaries and routing of SOAP messages
through intermediaries. I think it would be instructive for many of us if
someone were able to describe the practical deloyment of such a
configuration and how message routing issues were addressed in a particular
case.

Regards

Stuart

Received on Tuesday, 20 March 2001 13:02:59 UTC