RE: i028: Implications of the presence of ReplyTo

It is not that the sender controls the receiver's behavior, but the that
the sender's behavior must be consistent with its own expectations - in
this case the fact that it expects a possible reply.

For example, if the sender invokes a request response, then it is
reasonable to say that it expects a reply and thus it needs to encode a
replyTo. On the other hand, replies are often expected even if the WSDL MEP
does not necessarily imply them, so the sender may be encoding a replyTo in
cases where it is the semantics of the interaction, not the interface
definition, what drives the expectation of a reply.

Seems simple enough to me.

Paco




                                                                                                                                               
                      Brinild                                                                                                                  
                      <brinild@yahoo.com>             To:       public-ws-addressing@w3.org                                                    
                      Sent by:                        cc:                                                                                      
                      public-ws-addressing-req        Subject:  RE: i028: Implications of the presence of ReplyTo                              
                      uest@w3.org                                                                                                              
                                                                                                                                               
                                                                                                                                               
                      11/12/2004 08:25 AM                                                                                                      
                                                                                                                                               





In principle I agree.  However, as I stated in
another note, some of the definition text
for EPRs, such as FaulTo, would seem to imply that
the sender can control what it expects to
receive based on the presence or absence of
certain headers.  Perhaps it might be best if
these implications were not there?

--- Savas Parastatidis
<Savas.Parastatidis@newcastle.ac.uk> wrote:

>
> >
> > --- Martin Gudgin <mgudgin@microsoft.com> wrote:
> > >...And I don't think MEP semantics
> > > should be inferred from the presence/absence of
> wsa:
> > > headers ( although
> > > the set of such headers could be infered, or
> even
> > > explicitly stated, for
> > > a given MEP ).
> >
> > Too bad.  The idea of having a self-describing
> soap
> > envelope has its appeal.  Also, knowing if its a
> > request/
> > response MEP by looking at the message can
> eliminate
> > some ambiguity; for example in cases where there
> are
> > two port-types with the same operation, one as a
> > request/response and one as a one-way.
>
> Why would a message tell us whether it's part of a
> particular MEP or
> not? Couldn't the same message be part of multiple
> MEPs based on the
> semantics of a protocol, of a larger message-based
> interaction?
>
> When you see a letter in real life, you don't know
> from its envelope
> whether a reply should be sent. You know that if a
> reply is sent where
> it should go (if that information is captured) but
> you'll have to read
> the contents of the letter to figure out whether a
> reply is needed. And
> it may be the case that only the ultimate recipient
> (e.g. the service
> logic) will be able to make such a decision.
>
> Regards,
> .savas.
>
>
>


=====
Brinild@yahoo.com
http://brinild.blogspot.com



__________________________________
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com

Received on Friday, 12 November 2004 14:59:42 UTC