RE: issue: service type

 Jeffrey,
 I believe the problem is not that a service location URL cannot 
be passed around, for it can, but that a single URL cannot 
identify a service along with its type. If a WSDL document could 
contain only one <service>, it would be possible to pass the URL 
to this WSDL document and this would identify the contained 
<service>, too, and this service would contain the service 
location URLs.
 We can't do that with multiservice WSDL documents.

                   Jacek Kopecky

                   Senior Architect, Systinet Corporation
                   http://www.systinet.com/



On Thu, 6 Jun 2002, Jeffrey Schlimmer wrote:

 > I agree. It is also useful to be able to know the type too. The type
 > could be
 > (a) Known a priori
 > (b) Retrieved by sending a well-known message to the URL
 > (c) Included along with the URL
 > 
 > In Cases (a) and (b), you have just the URL. In Case (b) you have an
 > extra round trip that Case (c) eliminates.
 > 
 > There are precedents for each of these in other network systems, so I'm
 > not surprised that there are folks that see value to each.
 > 
 > --Jeff
 > 
 > -----Original Message-----
 > From: David Booth [mailto:dbooth@w3.org] 
 > Sent: Thursday, June 06, 2002 7:35 AM
 > To: Jacek Kopecky; Matt Long
 > Cc: 'WS-Desc WG (Public)'
 > Subject: RE: issue: service type
 > 
 > 
 > That sounds like a HUGE problem.  It would be horrible if one couldn't 
 > identify the service just by a URI.
 > 
 > At 04:49 PM 6/5/2002 +0200, Jacek Kopecky wrote:
 > 
 > >  Matt, one of the issues is that you cannot just pass a URL as a
 > >pointer to a service, you need the service QName, too. And the
 > >QName by itself is not sufficient either because you may not know
 > >where a WSDL definition of that QName is located.

Received on Thursday, 6 June 2002 12:05:32 UTC