Re: [wsrm] [REL-XX]Proposal for POLL RM-Reply Pattern

[Sunil Kunisetty <Sunil.Kunisetty@oracle.com>]

Scott Werden wrote:

> > -----Original Message-----
> > From: Sunil Kunisetty [mailto:sunil.kunisetty@oracle.com]
> > Sent: Tuesday, October 07, 2003 4:33 PM
> > To: Scott Werden
> > Cc: wsrm@lists.oasis-open.org
> > Subject: Re: [wsrm] [REL-XX]Proposal for POLL RM-Reply Pattern
> >
> >
> >
> >  Scott,
> >
> > Scott Werden wrote:
> >
> > > I agree with Sunil's conclusion that the poll operation
> > should not be in the
> > > soap body but his discussion raises some other issues. The
> > issue of Body vs.
> > > Header is really orthogonal to WSDL. Having a WSDL
> > description of the RMP's
> > > poll service may be desribable even if the operation is in
> > the header. I can
> >
> >  If it is a Header, one can always define it in the (WSDL)
> > SOAP Binding section
> >  for that PortType/Port.
> >
> >  Alternatively, we should use the WSDL Annotation stuff which
> > is pending (I'll
> >  work on it after we close this issue) on my AI list.
> >
> > >
> > > see this being useful, for instance, if there is an
> > intermediary (or proxy)
> > > that needs to know about all the services that an endpoint will be
> > > providing. Additionally, how is the SOAPAction header being
> > set without a
> > > WSDL? We have no RM field to specify that.
> >
> >   Scott, could you clarify what SOAPAction has to do with RM?
> >
>
> If the client is making a poll request to the server using Soap/HTTP
> binding, SOAPAction must be set to something. What is it set to if there is

 That's correct. SOAPAction is must for SOAP 1.1/Http binding and is
 removed (replaced with optional 'action' attribute) in SOAP 1.2.

>
> no WSDL? WS-I basic profile says it must be a quoted-empty string if the
> wsdl:soapAction attribute is missing. Do we consider the attribute to be

 We can perfectly assume that it is absent and hence use "" or to that
 matter any literal string (if it will be used for firewall which was the actual
 intent anyway). It should't matter even if it is a literal string as the RMP
 will ignore it. Note that this attribute itself is optional in the first place.

>
> missing? I see some ambiguity here and that is my concern.

 Valid concern....

 -Sunil