RE: [ws-rx-implement] Problem terminating response sequence

["Videlov, Vladimir" <vladimir.videlov@sap.com>]

Hi Doug,

 

Please confirm that this solution isn't the way everyone should do this in the real world, it's just the best way they could facilitate the testing. Actually on the other hand this solution does make sense to me now. We specifically decided not to develope the API between the app layer and the RM layer for this on RMD side or at least may not be ready for the interop. I'm agree with you that in this case the app level solution to this problem is the cleanest, when the spec can not offer protocol level one. Perhaps it's the easier solution too.

 

What's especially have confused me is that you decided to use a LastMessage tag in the payload instead of inventing some new term which wouldn't have the potential of being confused with the LastMessage protocol element. From the first overview of the scenario doc, it seemed to me that you want to move the old protocol LastMessage to the application level. That confuses me a lot, so that as a proposal we may change it to something more test scenario specific.  

 

Technically this solution will work also for SAP (hopefully on the interop).

 

Regards,

Vladimir

 

---

From: Doug Davis [mailto:dug@us.ibm.com]
Sent: Freitag, 17. Februar 2006 18:42
To: Videlov, Vladimir
Cc: ws-rx-implement@lists.oasis-open.org
Subject: RE: [ws-rx-implement] Problem terminating response sequence

Hi Vladimir, 
  In one of your earlier notes you talk about not 
including it in the business layer - that's really the only place it can go 
since the RM layer doesn't have the notion of LastMessage anymore.  RM does 
not talk about any kind of linking between the two (request vs offered/response) 
sequences.  In fact, the spec currently says: 
Note that offering a Sequence within the 
<wsrm:CreateSequence> element is simply a protocol optimization. There is 
no semantic difference between offering a Sequence, and choosing not to offer 
one and subsequently creating a new Sequence to carry messages from the RM 
Destination to the RM Source. 
Among 
other things, this means that just because one sequence is shut down it does not 
mean the other one needs to as well. So, that means we need a way for the server 
side (RM Destination) to know when to shut down the response sequence.  We 
could have mandated that there would be exactly 3 app-level messages but that 
seemed a bit restrictive.  The other option is to assume that the 
application on the server side will determine when it is done with the entire 
sequence of messages - this would be akin to the scenario where a single request 
could generate multiple responses - the RM layer has no way of knowing this - 
the only component that can know this is the app. 

Can you elaborate on why you think the spec isn't ready 
for this scenario? The anon req/resp scenario - perhaps - but including the 
lastMessage flag in the app message is something different and shouldn't be an 
issue - I hope anyway :-) 

If I 
remember correctly, LastMessage was removed because we decided it really wasn't 
needed - it provided no additional value/information beyond what we already had 
in the protocol. 

thanks 
-Doug

"Videlov, Vladimir" <vladimir.videlov@sap.com>

02/17/2006 03:07 AM

To	Doug Davis/Raleigh/IBM@IBMUS
cc	<ws-rx-implement@lists.oasis-open.org>
Subject	RE: [ws-rx-implement] Problem terminating response sequence

Hi Doug, 
  
The app layer 
does not know nothing about RM on the SOAP level and it should not deal with 
triggering protocol messages. Even if I make this hack somehow only for those 
req-resp scenarios, it doesn’t make sense to me to test that since the spec is 
obviously not ready for this scenario. Perhaps we will support this, but we have 
to discuss this internally if we have to put efforts in this. In the worst case 
we will not send TS from RMD to RMS, which should not mean that we are not 
interoperable. What do you thing? 
  
PS: I don’t 
follow the main TC, but why LastMessage flag was removed from Sequence element, 
when CloseSequence have a different meaning then and in this case there is a gap 
in the spec? 
  
  
Kind regards, 
Vladimir Videlov

---

From: Doug Davis 
[mailto:dug@us.ibm.com] 
Sent: Friday, February 17, 2006 12:48 
AM
To: Daniel Millwood
Cc: Videlov, Vladimir; 
ws-rx-implement@lists.oasis-open.org
Subject: Re: [ws-rx-implement] 
Problem terminating response sequence 
  

I don't think we should 
remove the terminate from the response side of things - that seems like too much 
of mainline scenario.  Vladimir - is it really a problem for your 
application on the server-side to be in control over terminating the response 
sequence?  When the two sequences are not linked I would think that would 
be the normal way the sequence would be shutdown. 
thanks, 
-Doug

Daniel Millwood <MILLWOOD@uk.ibm.com>

02/16/2006 05:38 PM

To	"Videlov, Vladimir" <vladimir.videlov@sap.com>
cc	<ws-rx-implement@lists.oasis-open.org>
Subject	[ws-rx-implement] Problem terminating response sequence

 

Hi Vladimir,

We wanted to test the server closing the response sequence back to the
client.  In the general case (except for the two-way through a firewall
scenario proposed this week), the life cycle of the request sequence and
the response sequence are not linked, so we needed a prompt for the RMD to
know when to terminate the response sequence.  This was what we came up
with on the last interop call.

In a real implementation, it may be that the server would not choose to
terminate the response sequence until the sequence expires.  We talked
about this as an option on last weeks interop call (basically, not
expecting the TerminateSequence to be sent for the response sequence), but
decided on the application level prompt instead so that we could test the
specification.

At the time, the proposal was not identified as a problem for anyone.  If
it is for your implementation, perhaps the TerminateSequence for the
response sequence should be removed from the 2-way scenarios (except
two-way through a firewall).

Thoughts anyone?

Thanks,  Dan

WS-Reliable Messaging Architect and Team Lead
IBM WebSphere Messaging Design and Development
MP 211
Hursley
Tel. Internal 248617
Tel. External +44 1962 818617
Email. millwood@uk.ibm.com



                                                                         
           "Videlov,                                                    
           Vladimir"                                                    
           <vladimir.videlov                                          To
           @sap.com>                 Daniel Millwood/UK/IBM@IBMGB        
                                                                      cc
           16/02/2006 17:44          <ws-rx-implement@lists.oasis-open.o
                                     rg>                                
                                                                 Subject
                                     RE: [ws-rx-implement] Groups -      
                                     InteropScenarios.doc uploaded      
                                                                         
                                                                         
                                                                         
                                                                         
                                                                         
                                                                         




Hi Dan,

Sorry for the late joining in on this issue ("last message" notification
to RMD), but why you decided to use an flag within the business
operation payload instead of CloseSequence for example? This flag will
never go to the RM layer at least in our impl.

I'm not following the WS-RX TC closely, so that the answer could be
obvious, but this solution bothers me a lot.


Kind regards,
Vladimir Videlov

-----Original Message-----
From: millwood@uk.ibm.com [mailto:millwood@uk.ibm.com]
Sent: Friday, February 10, 2006 1:31 PM
To: ws-rx-implement@lists.oasis-open.org
Subject: [ws-rx-implement] Groups - InteropScenarios.doc uploaded

As requested on the call yesterday, I have updated the interop doc with
the
following changes:


For scenarios 2.1, 2.2 and 2.3, indicate in the last application request
message that it is the last one and how many were sent, so that the RMD
knows when it can close the sequence for replies.

I have added a missing Clients TerminateSequence Message example for
scenario 2.1
I have added a note about how the offered sequence should be used
pending
resolution to issue i090 for scenario 2.1

Please review and let me know of any problems.

Thanks, Dan


-- Mr. Daniel Millwood

The document revision named InteropScenarios.doc has been submitted by
Mr.
Daniel Millwood to the WS-RX Implementation SC document repository.
This
document is revision #2 of InteropScenarios.doc.

Document Description:
Initial interop scenario doc

View Document Details:
http://www.oasis-open.org/apps/org/workgroup/ws-rx-implement/document.ph
p?document_id=16617

Download Document:
http://www.oasis-open.org/apps/org/workgroup/ws-rx-implement/download.ph
p/16617/InteropScenarios.doc

Revision:
This document is revision #2 of InteropScenarios.doc.  The document
details
page referenced above will show the complete revision history.


PLEASE NOTE:  If the above links do not work for you, your email
application
may be breaking the link into two pieces.  You may be able to copy and
paste
the entire link address into the address field of your web browser.

-OASIS Open Administration