RE: Another proposal for async extensions

 

> -----Original Message-----
> From: public-ws-async-tf-request@w3.org 
> [mailto:public-ws-async-tf-request@w3.org] On Behalf Of Marc Hadley
> Sent: Thursday, Jun 16, 2005 10:48 AM
> To: public-ws-async-tf@w3.org
> Subject: Another proposal for async extensions
> 
> Following yesterdays telcon I thought I should take the time 
> to write  
> up my thoughts on where we should be headed. This proposal adds some  
> deltas to the SOAP 1.2 HTTP binding to allow responses to go  
> elsewhere and to allow empty HTTP response entity bodies (we might  
> still want to define a one-way MEP but I don't think this is 
> strictly  
> necessary). This proposal also extends the wsaw:UsingAddressing WSDL  
> extension to allow specification of the bindings an endpoint 
> supports  
> for [reply endpoint] and [fault endpoint] messages addressing  
> properties.
> 
> Another proposal for Asynchronous Support in SOAP and WSDL
> 
> 1. SOAP 1.2 HTTP Binding
> 
> Tweak the existing SOAP 1.2 binding[1] as follows.
> 
> 1.1 New Features
> 
> Say that it supports the SOAP 1.2 Addressing 1.0 feature[2] 
> using the  
> SOAP 1.2 Addressing 1.0 Module[3].
> 
> 1.2 Update table 17[4]
> 
> Add a new row for HTTP response code 202
> 
> Status Code: 202
> reason phrase: Accepted
> Significance/Action: The response message (if any) will be sent to  
> the node identified by the http://www.w3.org/@@@@/@@/addressing/ 
> feature/ReplyEndpoint property of the SOAP 1.2 Addressing 1.0 feature 
> [2].
> NextState: Success
> 

I am comparing your proposal with the original SOAP 1.2 fix that I sent
a while back. 

You chose to indicate the NextState as Success, which indicates terminal
states of the soap request response MEP. It seems to me adding a
separate item to the table for 202 is necessary but not sufficient,
because you chose to terminate the MEP. Termination leaves the
definition of how the response message can be generated and sent out,
which seems incomplete to me.  

I tried to take a stab at this, which I did not get responses other than
DavidO. 

Perhaps you can clarify this point as to why you think that we can
terminate the MEP without considering what happens to the response
(NextState)

Thanks. 

--umit

Received on Thursday, 16 June 2005 19:37:11 UTC