RE: proposed breaking change to echo global element declaration

OK, I've made the exmples match the WSDL with echoIn and echoOut
as the element names.

I've also removed the 'echo' operation from the WSDL and
added some Action attributes in a meaningless wsaw: namespace.

Paul


-----Original Message-----
From: public-ws-addressing-tests-request@w3.org on behalf of Rick Rineholt
Sent: Tue 12/20/2005 5:04 PM
To: public-ws-addressing-tests@w3.org; public-ws-addressing@w3.org
Subject: Re: proposed breaking change to echo global element declaration
 
Paul,
I think this is the right decision. However, one minor clarification, the 
only reason I chose to use "echoIn" and "echoOut" for my messages is that 
I believed that is the message pattern as I interpreted the WSDL at the 
time. There was an "echo" in the schema but it did not connect to any 
operations in the port as I read the WSDL.  My personal experiences with 
interops are when there appears to be conflicting information the order 
they are resolved in are:  specifications, WSDL and then the samples and 
other interop documentation.  In addition while I can't say it's wrong or 
that it violates any best practices I haven't come across any cases ever 
working with interops,demos etc where they are the same. 

There is 
Rick Rineholt
"MAINTENANCE FREE -- It's impossible to fix."

rineholt@us.ibm.com




<paul.downey@bt.com> 
Sent by: public-ws-addressing-tests-request@w3.org
12/20/2005 07:35

To
Rick Rineholt/Raleigh/IBM@IBMUS
cc
<public-ws-addressing-tests@w3.org>, <public-ws-addressing@w3.org>
Subject
proposed breaking change to echo global element declaration








The tests currently use the same Global Element Declaration (GED) 
'echo' for both request and response which reflects my personal 
preference not to use 'wrapped' style documents. 

However this does impose an architectural style on implementations:
the current tests require an endpoint which handles both 'echo'
requests and 'echo' responses to dispatch based upon wsa:action, 
and not the GED. 

*but* I'm told GED dispatching is intrinsic to more than one
of implementations we hope to demonstrate interoperating
during CR.

I understand that this may be contentious, but the only 
way to support both world views is to make the GED of the echo 
request an 'echoIn' element and the GED of the response an 
'echoOut' element, as per IBM's public endpoint:

http://lists.w3.org/Archives/Public/public-ws-addressing-tests/2005Dec/0034.html


I'm happy to make such a change, but this will impact anyone 
who has implemented the testsuite, so please raise any objections 
you have ASAP.

Paul

Received on Wednesday, 21 December 2005 22:01:23 UTC