RE: [ws-rx] Relationship of ISSUE 008 with ISSUE 021

Ashok: 

 

-----Original Message-----

From: Ashok Malhotra [mailto:] 

Sent: Thursday, September 29, 2005 8:34 AM

To: Tom Rutt2; wsrx

Subject: RE: [ws-rx] Relationship of ISSUE 008 with ISSUE 021

 

Tom:

You can attach policies at the binding level.

This allows you to do what you want.

 

<JD> mmmh... rmpolicy says at
wsdl:binding level. that seems to only allow port-level granularity. 

 

 

But there are other problems.  As I see it, the

binding (or message) definition is for the messages

in the sequence - not the framing messages such as

CreateSequence, CreateSequenceResponse, Ack, etc. 

So if we attach a QoS policy to the binding, which

messages does it apply to? 

 

<JD> clearly not to these "protocol-signaling"
messages, which are not exposed in WSDL. But I do not see a problem here -
unless you want them to be reliable too? (I think there is a consensus that
they don't need to - e.g. if a CreateSequence "fails" ,
an RMS may always decide to resend a new one. And if a stand-alone Ack is lost,
subsequent ones will make-up.)

 

All the best, Ashok

 

 

> -----Original Message-----

> From: Tom Rutt [mailto:] 

> Sent: Thursday, September 29, 2005 8:23 AM

> To: wsrx

> Subject: [ws-rx] Relationship of ISSUE 008 with ISSUE 021

> 

> The Sept 2004 version of WS-policy attachment includes the
following

> section:

> "

> 4.1.4 Message Policy Subject The following WSDL/1.1 elements 

> are used to describe messages:

> . wsdl:message

> . wsdl:portType/wsdl:operation/wsdl:input

> . wsdl:portType/wsdl:operation/wsdl:output

> . wsdl:portType/wsdl:operation/wsdl:fault

> . wsdl:binding/wsdl:operation/wsdl:input

> . wsdl:binding/wsdl:operation/wsdl:output

> . wsdl:binding/wsdl:operation/wsdl:fault

> 

> The current WS-RM policy spec attaches policy at the endpoint 

> subject level.

> 

> If we could attach the Qos policy for DA at the message 

> subject level, we would be in a better position to express 

> the desired semantics for Reliability of a WSDL 

> Request/response operation.

> 

> For example, exactly once, ordered could be attached to the 

> request message, while no reliability qos is attached to the
response.

> 

> Other examples could attache reliability DA levels to the 

> response message as well.

> 

> Thus I now see that Issue 008 on granularity of reliability 

> policy attachment, and Issue 021 on reliability for two way 

> exchanges should be discussed together.

> 

> Tom Rutt

> 

> --

> ----------------------------------------------------

> Tom Rutt  email: ; 

> Tel: +1 732 801
5744          Fax: +1 732 774 5133

> 

> 

> 

>