RE: Port type extension proposal

I agree that 1a is preferable (i.e. make operations top-level syntactic 
constructs and have portTypes reference them by qname).  It has always 
seemed strange to me that operations are second class citizens in the WSDL 
model (i.e. you cannot refer to them by qname, as you can with portTypes, 
messages, XSD types and elements, etc).  I suppose that if all you are 
doing is writing tooling to consume WSDL to produce stubs and skeletans, 
then its not a big deal.  But when you want to use the WSDL definitions for 
discovery, being able to refer to operations by qname is desirable.

-Steve

At 10:24 AM 10/3/2002, Martin Gudgin wrote:
>I think that would be my preference at this point. I'd go for 1a.
>
>Gudge
>
> > -----Original Message-----
> > From: Jean-Jacques Moreau [mailto:moreau@crf.canon.fr]
> > Sent: 03 October 2002 16:18
> > To: Martin Gudgin
> > Cc: Steve Tuecke; Sanjiva Weerawarana; www-ws-desc@w3.org
> > Subject: Re: Port type extension proposal
> >
> >
> > Personnally, I see a number of advantages to making operations
> > first class objects (option 1). Comments?
> >
> > Jean-Jacques.
> >

Received on Friday, 4 October 2002 13:05:38 UTC