Re: MEP proposal

On 21 Feb 2003 20:01:42 +0100
Jacek Kopecky <jacek@systinet.com> wrote:
> On Fri, 2003-02-21 at 19:51, Amelia A. Lewis wrote:
> > On 21 Feb 2003 19:38:17 +0100
> > Jacek Kopecky <jacek@systinet.com> wrote:
> > 
> > > Is something a service if it only provides output-first operations? 
> > 
> > Yes.  This is exactly what a publishing service is.  The clients learn
> > where to subscribe from the service description.  The service has
> > absolutely no need to ever see a description of what the clients do,
> > because it is telling them what they can do.
> > 
> > > If so, how exactly do you rule out clients from this? 
> > 
> > What?
> 
> Any time someone mentions describing the client someone else shouts that
> we agreed to describe the service, not the client.
> 
> A client is something that only provides output-first operations, isn't
> it? It is a service, then. Someone might want to describe this service,
> wouldn't they? 8-)

I think that we're simply coming from very different networking conceptions, or something.

Why is a client something that only provides output-first operation?  I don't agree at all; a subscriber is a client always does input first.

So ... the rest of the para fails to follow.

Please understand that I really *don't care* about the client/server model of computing; I work for a firm that is heavily invested in the publish/subscribe model.  And client/server has plenty of advocates already.  publish/subscribe stuff very frequently has service-initiated operations (that is, the publisher == service, not subscriber == service).

And getting WSDL able to support this as gracefully as it describes request/response client/server oriented HTTP is one of the tasks assigned my by my bosses.  *shrug*

Amy!
-- 
Amelia A. Lewis
Architect, TIBCO/Extensibility, Inc.
alewis@tibco.com

Received on Friday, 21 February 2003 14:12:58 UTC