ws-rx message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: Re: [ws-rx] PR issue 1 - WS-Addressing comment on ws-rm related to use ofextended anonymous uri
- From: Doug Davis <dug@us.ibm.com>
- To: ws-rx@lists.oasis-open.org
- Date: Fri, 29 Sep 2006 18:27:24 -0400
Paul Fremantle <paul@wso2.com> wrote on 09/27/2006
08:37:00 AM:
> > In my mind its the same component that checks for "none"
on any
> > outgoing message and take a specialized action.
>
> It would really help me understand your issue if you actually explained
> it! Its still not clear what you imagine is happening here.
I've been trying - please reread my full response
- I explain my
concerns in that note. You may not agree with
it, but its all there.
> I'm really not clear why any component other than the RM level component
> needs to look at this special reference parameter.
> Suppose I have a request-response with this EPR = (anon + rmpolling
ref
> parameter).
(RM EPR == and EPR that says we'll use MC to get the message)
Repeating, there are lots of ways to implement the
solution - and not all messages
necessarily go thru the RM component so detection
of messages targeted
to one of these RM EPRs may need to happen at the
same place the soap
engine detects 'anon' and 'none' - seems like a logical
place to add a check
for new 'special' URIs. Current impls would
need to check for WSA's special URIs and we worked
hard to make sure that
we leveraged code that would already be there and
we are asking them to
simply add another 'if' to the existing check. Adding
a check for
an outgoing message's ref-p's (which are supposed
to be opaque to it
because they're just for routing purposes at the final
node) was considered
to be quite a radical change for implementations.
Clearly its possible,
but during our discussions it was just another reason
why the use of
ref-p's was not chosen.
I thought you were part of these discussions, but
maybe I'm remembering
wrong.
> The server is trying to send a response back to the client. Before
it
> gets to the transport level handling, the RM handler stores the message
> in some message store.
> If there is a sequence already in existence, the RX handler also adds
a
> sequence header to it.
>
> Then the transport tries to deliver the message. If the correct
> transport backchannel exists, it should deliver that message right
then.
> If not, then the message is lost - just as if there was no RM handling
> in place.
>
> Of course if there is no sequence in play, I would expect that message
> to be stored and a CreateSequence message to be stored as well. At
some
> point we hope the MakeConnection will come in and pick up those messages.
> Paul
>
>
>
>
> Doug Davis wrote:
> >
> >
> > -Doug
> >
> >
> >
> > *Paul Fremantle <paul@wso2.com>*
> >
> > 09/27/2006 08:16 AM
> >
> >
> > To
> > Doug Davis/Raleigh/IBM@IBMUS
> > cc
> > Gilbert Pilz <Gilbert.Pilz@bea.com>, ws-rx@lists.oasis-open.org
> > Subject
> > Re: [ws-rx] PR issue 1 - WS-Addressing comment on
ws-rm related to
> > use of extended anonymous uri
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Doug
> >
> > > > Can you explain which bit of the processing chain you
are thinking of?
> > >
> > > I imagine that all soap engines have some if-statement,
someplace, that
> > > checks for anon/none - what I like about our current solution
is that we
> > > are asking them to simply add more if's (or else's) to that
existing
> > > logic - no matter where it may be. Checking for certain
ref-p's on an
> > > outgoing message is not something I feel comfortable with
assuming
> > > everyone
> > > does today, therefore I viewed this as a pretty radical
change.
> > I still don't understand exactly which component you expect to
be
> > checking for refp's on outgoing messages? Is the handling an
anonymous
> > response and deciding how to deal with it?
> >
> > Paul
> >
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]