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

Re: SOAP pattenrs vs. WSDL patterns

From: FABLET Youenn <youenn.fablet@crf.canon.fr>
Date: Fri, 27 Jun 2003 17:26:36 +0200
Message-ID: <3EFC622C.2010203@crf.canon.fr>
To: Amy Lewis <alewis@tibco.com>
CC: Jean-Jacques Moreau <jean-jacques.moreau@crf.canon.fr>, dbooth@w3.org, public-ws-desc-meps@w3.org
I agree (even if the soap mep model is quite extensible and large enough 
IMO to cover the 80/20 cases :-).

What is also of interest is to allow the maximum compatibility between 
the soap mep model and the wsdl pattern model.
The soap mep framework mixes generic mep information (exactly what a 
wsdl pattern provides) and additional information (state machine, 
protocol specific data).
IMO, someone (the wsdl wg ?) should clarify how wsdl patterns and soap 
meps are working together. Guidelines should be provided about how to 
write new SOAP meps. These guidelines would promote the idea of defining 
a SOAP mep through:
    - reusing an already defined wsdl pattern/ creating a new wsdl pattern
    - in the new soap mep spec, refer to this wsdl pattern and add 
additional information/constraints to be implementable
The benefit is to clearly separate abstract information/concrete 
information in two specs, these two specs being not redundant.
It seems difficult to enforce these guidelines for the Soap/response and 
req/resp meps but I see benefits for to be created soap meps.
The reuse of wsdl patterns to define soap meps add some constrains to 
the wsdl pattern model. I only see a few constrains however and quite 
(?) common sense:
    - identify each message (or message type for the multicast case) 
through a uri
    - identify each mep through a uri
    - identify each described node (or node type for the multicast case) 
through a uri
    - (any other constraint?)
The biggest difference between soap meps and wsdl patterns as defined in 
the current document might be that wsdl patterns differentiate the 
service node from the other nodes of the network while soap does not 
make that distinction.
Depending on the wsdl pattern description model , this could also be 
solved nicely.
    Youenn


Amy Lewis wrote:

>Sorry.
>
>If WSDL supports more than SOAP (as I believe that it does), then we
>must support message exchange patterns that they define.  We cannot
>restrict the set of available MEPs at the binding level to only those
>defined by SOAP, because WSDL supports binding styles other than SOAP.
>
>Amy!
>On Fri, 27 Jun 2003 14:31:53 +0200
>"Jean-Jacques Moreau" <jean-jacques.moreau@crf.canon.fr> wrote:
>
>  
>
>>I'm lost by the double negative (can't-only). Didn't you mean
>>something different?
>>
>>Jean-Jacques.
>>
>>Amy Lewis wrote:
>>
>>    
>>
>>>We can't mandate SOAP MEPs only at the binding level, if we plan to
>>>support bindings other than SOAP.  Can we?
>>>      
>>>
>>
>>    
>>
>
>
>  
>
Received on Friday, 27 June 2003 11:27:05 GMT

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