Mailing-List: contact axis-dev-help@xml.apache.org; run by ezmlm
Precedence: bulk
Reply-To: axis-dev@xml.apache.org
Message-ID: <00f701c16ec5$82a2e210$9fa17cca@watson.ibm.com>
From: "Sanjiva Weerawarana" <sanjiva@watson.ibm.com>
To: <axis-dev@xml.apache.org>
References: <47FD4196C273D411950300508BCF197801656D06@S0001EXC0006>
 <032501c16e47$328cb4a0$48a17cca@watson.ibm.com>
 <001b01c16e4a$a9ff9860$b3185480@genesis.adtech.internet.ibm.com>
 <003901c16ea8$700ebe20$9fa17cca@watson.ibm.com>
 <000d01c16eb6$8757dc70$8a161818@genesis.adtech.internet.ibm.com>
Subject: Re: What do we do with operation types:  notification,
 solicit-response?
Date: Fri, 16 Nov 2001 23:38:24 +0600
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit

Hi Ben,


> Well, transport is not totally an orthogonal issue.  Suppose our transport
> is Http.  And we have a service that provides notification every time
event
> E happens on the server.  We don't want the client to have to have a web
> server to receive all the notifications by HTTP PUT.  IM provides a much
> lighter-weight method for the client to receive messages from the server.

OK so you've interpreted solicit-response as event mechanism. That,
however, is not the only interpretation - it can also be viewed as
a piece of "required" functionality of the service and that the
required functionality can be provided by a 3rd party w.r.t. the
service client and the provider. There are other possible views too.

I agree the transport is not orthogonal to the convenience of
implementing events. If the semantic is event semantics, then it
can certainly be implemented for HTTP too; yes that would require
the client to offer a service to which the event is delivered.

> I
> wrote the Jabber code for Axis and Soap 2.2.  I've been thinking of
> comitting it, but I don't have time at school to keep updating my code.
> Once a new alpha comes out, I'll update it again.

Cool. I thought your name was familiar but couldn't place the context!
Its certainly nice to have Jabber transports .. it validates that the
transport abstractions can be mapped to a relatively different transport.

Sanjiva.