W3C home > Mailing lists > Public > public-ws-desc-meps@w3.org > April 2003

RE: MEP teleconf

From: Martin Gudgin <mgudgin@microsoft.com>
Date: Mon, 14 Apr 2003 13:24:55 -0700
Message-ID: <7C083876C492EB4BAAF6B3AE0732970E0B211B5B@red-msg-08.redmond.corp.microsoft.com>
To: "David Booth" <dbooth@w3.org>, <public-ws-desc-meps@w3.org>

It seems obvious, to me at least, that there is quite a gulf between my
position and the position other people are taking. I think that
operations in WSDL should be abstract, defining ONLY the sequence,
direction and cardinality of messages. All other details that are needed
to 'make stuff actually work', like # of channels, sync vs async,
timing, are defined at the binding level ( possibly with multiple levels
of binding, if we end up supporting multi-level bindings ).

The thing I find quite bizarre with the current MEP/IOP document is that
given the WSDL below, both port types are exactly the same by the time
you get to the binding level, but they are not compatible. I think this
is very bad, because mep2b is essentially a subset of iop2, so it seems
very strange that they are modelled as being completely separate.

Cheers

Gudge


<w:definitions xmlns:w='http://www.w3.org/2003/03/wsdl' 
               xmlns:tns='http://example.org/stuff' 
               xmlns:ws='http://www.w3.org/2003/01/wsdl/soap12'
               targetNamespace='http://example.org/stuff' >

 <!-- Message definitions elided -->

 <w:portType name='Foo' >
  <w:operation name='op' pattern='mep2b' >
   <w:input message='msga' />
   <w:output message='msgb' />
  </w:operation>
 </w:portType>

 <w:portType name='Bar' >
  <w:operation name='op' pattern='iop2' >
   <w:input message='msga' />
   <w:output message='msgb' />
  </w:operation>
 </w:portType>

 <w:binding name='FooBind' type='tns:Foo' >
  <ws:binding transport='http://www.w3.org/2002/12/soap/bindings/HTTP/'
/> 
  <w:operation name='op'>
   <ws:operation style='document' />
   <w:input>
    <ws:body />
   </w:input>
   <w:output>
    <ws:body/>
   </w:output>
  </w:operation>
 </w:binding>

 <w:binding name='BarBind' type='tns:Bar' >
  <ws:binding transport='http://www.w3.org/2002/12/soap/bindings/HTTP/'
/> 
  <w:operation name='op'>
   <ws:operation style='document' />
   <w:input>
    <ws:body />
   </w:input>
   <w:output>
    <ws:body/>
   </w:output>
  </w:operation>
 </w:binding>

</w:definitions>


Gudge

> 
> -----Original Message-----
> From: public-ws-desc-meps-request@w3.org 
> [mailto:public-ws-desc-meps-request@w3.org] On Behalf Of David Booth
> Sent: 14 April 2003 20:44
> To: public-ws-desc-meps@w3.org
> 
> 
> Only Umit and I showed up at today's MEP teleconference.  Amy 
> and Philippe gave regrets; others were AWOL.
> 
> Umit and I discussed the clarified assumptions in 
> http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl12/meps-v
> s-iops/meps-vs-iops_clean.htm
> 
> At this point we really need Gudge on the call to ferret out 
> the remaining disagreements/misunderstandings, as others seem 
> to be pretty much in agreement.
> 
> 
> -- 
> David Booth
> W3C Fellow / Hewlett-Packard
> Telephone: +1.617.253.1273
> 
> 
> 
Received on Monday, 14 April 2003 16:25:31 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Friday, 25 March 2005 11:17:39 GMT