Re: [ws-rx] Issue i078: Lost TerminateSequence -- proposal

[Christopher B Ferris <chrisfer@us.ibm.com>]

+1

Christopher Ferris
STSM, Emerging e-business Industry Architecture
email: chrisfer@us.ibm.com
blog: http://www.ibm.com/developerworks/blogs/dw_blog.jspa?blog=440
phone: +1 508 377 9295

Doug Davis/Raleigh/IBM@IBMUS wrote on 01/26/2006 03:37:22
PM:

> 
> ok - that clears it up.  I think CS/CSR is clearly a req/res
and as 
> long as TS/TSR is too then ok. 
> I don't think we need to talk about wsa:RelatesTo - its just implied
> by using WSA. 
> -Doug 
> 
> 

> 
> Anish Karmarkar <Anish.Karmarkar@oracle.com> 
> 01/26/2006 03:33 PM 
> 
> To
> 
> Doug Davis/Raleigh/IBM@IBMUS 
> 
> cc
> 
> ws-rx@lists.oasis-open.org 
> 
> Subject
> 
> Re: [ws-rx] Issue i078: Lost TerminateSequence -- proposal
> 
> 
> 
> 
> Doug,
> 
> The TerminateSequence/TerminateSequenceResponse message pairs are
meant 
> to be similar to the the CS/CSR pair. The reason for the 3rd bullet
was 
> to capture the discussion that we had during the F2F and because the
> previous proposal from Bob required it to be synchronous. I would
> imagine that nothing in the 3rd bullet needs to be translated into
> spec-ese. The intent of the bullet was to clarify the distinction
> between this proposal and the previous proposal. I should also clarify
> that the 'one-way/req-response' in the 3rd bullet refers to the 
> underlying transport protocol. I.e. one can use TS/TSR irrespective
of 
> whether you use a inherently one-way transport protocol such as UDP
or 
> inherently req-res transport protocol such as HTTP (same as CS/CSR).
> Make sense?
> 
> BTW, for the CS/CSR we don't say anything about the wsa:RelatesTo,
but 
> in the examples (appendix B) we do include it. Don't we need to specify
> that?
> 
> -Anish
> --
> 
> Doug Davis wrote:
> > 
> > Anish,
> >   One question, does the terminateSeqResponse have a wsa:RelatesTo?
> >  Your 3rd bullet makes me wonder about whether something
special is 
> > goning on here since we don't go out of our way to say anything
like 
> > this about the CreateSeq or CloseSeq messages.
> > thanks,
> > -Doug
> > 
> > 
> > 
> > *Anish Karmarkar <Anish.Karmarkar@oracle.com>*
> > 
> > 01/26/2006 02:47 PM
> > 
> >                  
> > To
> >                  wsrx
<ws-rx@lists.oasis-open.org>
> > cc
> >                  
> > Subject
> >                  [ws-rx]
Issue i078: Lost TerminateSequence -- proposal
> > 
> > 
> >                  
> > 
> > 
> > 
> > 
> > 
> > All,
> > 
> > Bob, Marc and I were given an AI at the last F2F to come up with
a
> > concrete proposal for i078 [1]. After some email discussion amongst
us
> > we came up with the following proposal to resolve i078. Please
note that
> > this does not contain exact text changes with line number. I
would like
> > to see if folks agree on the general direction before sending
out exact
> > text change.
> > 
> > 1) Introduce a new wsrm:TerminateSequenceResponse message:
> > <wsrm:TerminateSequenceResponse ...>
> >     <wsrm:Identifier ...> xs:anyURI </wsrm:Identifier>
> >     ...
> > </wsrm:TerminateSequenceResponse>
> > 
> > 2) A <wsrm:TerminateSequenceResponse> is sent in the body
of a response
> > message by an RM Destination in response to receipt of a
> > <wsrm:TerminateSequence> request message. I.e., it is not
sent to AcksTo
> > EPR but to the sender of the wsrm:TerminateSequence message (similar
to
> > the wsrm:CreateSequenceResponse message)
> > 
> > 3) We do *not* say anything about whether
> > wsrm:TerminateSequence/wsrm:TerminateSequenceResponse are
> > synchronous/asynchronous/one-way/req-res message-exchanges. That
would
> > depend on the underlying binding/transport that is used and not
relevant
> > to the WSRM spec.
> > 
> > 4) The RMS *may* retransmit the wsrm:TerminateSequence message
if it
> > does not receive the wsrm:TerminateSequenceResponse message (either
> > because the wsrm:TerminateSequence message got lost or the
> > wsrm:TerminateSequenceResponse message got lost).
> > 
> > 5) The RMD on receiving a wsrm:TerminateSequence message for
a Sequence
> > that has already been terminated (eg. for the case when RMD terminated
> > the Sequence and sent the wsrm:TerminateSequenceResponse message,
but
> > that message got lost) or for an unknown sequence, responds with
a
> > wsrm:TerminateSequenceResponse message or generates a fault.
> > 
> > Comments?
> > 
> > -Anish
> > --
> > 
> > [1] http://docs.oasis-open.org/ws-rx/issues/ReliableMessagingIssues.xml#i078
> >