Self-describing Messages wrt MEPs

I was thinking about Paco's desire for ReplyTo to have a desire for a
distinguished attribute to indicate that no response is expected.
Effectively, I think that he wants messages to be self-describing wrt
MEPs.

 

A one-way MEP over HTTP would be no-response for ReplyTo and FaultTo.

A robust-in-only MEP over HTTP would be no-response for ReplyTo and
anonymous FaultTo.

A request-response MEP over HTTP would be anonymous ReplyTo and
anonymous FaultTo.

 

>From an intermediary's perspective, it could look at the ReplyTo and
FaultTo to determine the MEP.

 

It seems worth calling out, that making messages self-describing from an
MEP perspective hasn't been forcefully called out as a requirement on
WS-A to date.

 

Another way of looking at this is that it moves the WSDL 2.0 MEP
functionality into WSDL 1.1 via WS-Addressing.  If you want a robust
in-only MEP over HTTP, you use WSDL 1.1 and then WS-A with the values
listed above.   This seems like it might have a side-effect of hurting
wsdl 2.0 adoption, in the same way the WS-I BP "backporting" parts of
SOAP 1.2 into SOAP 1.1 has probably hurt SOAP 1.2 adoption.

 

Cheers,

Dave