Re: [wsn] Multiple endpoints in IRP?

[Steve Graham <sggraham@us.ibm.com>]

I  would expect this work to be
clearly delimited.

++++++++
Steve Graham
(919)254-0615 (T/L 444)
STSM, On Demand Architecture
Member, IBM Academy of Technology
<Soli Deo Gloria/>
++++++++


David Hull <dmh@tibco.com> wrote on 08/06/2004
02:37:41 PM:

> A point of clarification:  Is this "(small) normative specification"
expected to be
> clearly delimited as such, or will it be folded into the WSRF specifications?
 The 
> concern, of course, is that we be able to say things like "the
subscription 
> endpoint MUST conform to [whatever IRP becomes]" without having
to require anything
> about WSRP or WSRL or any other WSR*.
> 
> At this point I don't have a strong opinion whether such a spec merits
its own doc,
> or whether "Section 42 of the XYZZY doc" is fine.  I'm
more concerned that it be 
> clearly delimited and self-contained.
> 
> 
> Steve Graham wrote:
> 
> At the WSRF f2f, there was some discussion about IRP, whether a "singleton
pattern 
> was legal" whether non-WSA approaches were legal. 
> WSRF commissioned a task force to build a (small) normative specification
defining 
> the IRP concept and outlining a set of possible embodiments of IRP
in concrete 
> technology.  Using WSAddressing with reference properties and
using WSAddressing by
> encoding the resource disambiguator in the wsa:address will be two
embodiments we 
> will spell out.  I suspect there may be other embodiments we
outline.  The major 
> work here is to tease apart the IRP "concept" from any particular
embodiment. 
> 
> sgg 
> 
> ++++++++
> Steve Graham
> (919)254-0615 (T/L 444)
> STSM, On Demand Architecture
> Member, IBM Academy of Technology
> <Soli Deo Gloria/>
> ++++++++
> 
> 

> 
> "Patil, Sanjay" <sanjay.patil@sap.com> 
> 08/06/2004 11:22 AM 
> 
> To
> 
> wsn@lists.oasis-open.org 
> 
> cc
> 
> Subject
> 
> RE: [wsn] Multiple endpoints in IRP?
> 
> 
> 
> 
>   
> OK. It seems to me that for whatever it is, the semantics and mechanics
of IRP are 
> fairly clear. Could some one now please explain for my benefit (and
for those who 
> are part of WSN only), what did we mean during the F2F by  --
 pending 
> clarification of IRP under WSRF. Just a description of the problem
statement would do for me.
>   
> Thanks, 
> Sanjay 
> -----Original Message-----
> From: Steve Graham [mailto:sggraham@us.ibm.com] 
> Sent: Friday, Aug 06, 2004 5:12 AM
> To: Patil, Sanjay
> Cc: David Hull; 'Vambenepe, William N'; wsn@lists.oasis-open.org
> Subject: RE: [wsn] Multiple endpoints in IRP?
> 
> 
> Hi Sanjay: 
> IRP does not require WS REsources of the same type to share a single
endpoint.  So,
> in your example, it is completely legal for the Subscriptions created
by an NP to 
> have 3 or 4 or 20 or n different SubscriptionManager endpoints. 
> 
> ++++++++
> Steve Graham
> (919)254-0615 (T/L 444)
> STSM, On Demand Architecture
> Member, IBM Academy of Technology
> <Soli Deo Gloria/>
> ++++++++
> 

> 
> "Patil, Sanjay" <sanjay.patil@sap.com> 
> 08/05/2004 06:49 PM 
> 
>         
>        To:        "'Vambenepe,
William N'" <vbp@hp.com>, Steve 
> Graham/Raleigh/IBM@IBMUS, David Hull <dmh@tibco.com> 
>        cc:        wsn@lists.oasis-open.org
>        Subject:        RE:
[wsn] Multiple endpoints in IRP?
> 
> 
> 
> 
> 
> 
> I have a related question here  --
> 
> Does the IRP require that multiple WS Resources of the same type share
a single 
> endpoint? For example, is it required that the EPRs for all the Subscriptions
> created by a NotificationProducer have the same endpoint address.
Or would it 
> simply be considered as a common practice (isn't that really a pattern)?
> 
> Thanks,
> Sanjay
> 
> > -----Original Message-----
> > From: Vambenepe, William N [mailto:vbp@hp.com] 
> > Sent: Thursday, Aug 05, 2004 2:59 PM
> > To: Steve Graham; David Hull
> > Cc: wsn@lists.oasis-open.org
> > Subject: RE: [wsn] Multiple endpoints in IRP?
> > 
> > 
> > (with my TC member hat on, not my chair hat)
> > 
> > The statement you pointed out is really just a reinforcement
of the
> > rules of the IRP. And, as Steve says, it doesn't imply any 
> > capability to
> > address WS-Resources as groups. Or did I misunderstand your question
> > David?
> > 
> > William
> > 
> > 
> > ________________________________
> > 
> > From: Steve Graham [mailto:sggraham@us.ibm.com] 
> > Sent: Thursday, August 05, 2004 1:43 PM
> > To: David Hull
> > Cc: wsn@lists.oasis-open.org
> > Subject: Re: [wsn] Multiple endpoints in IRP?
> > 
> > 
> > 
> > hi David: 
> > I would think that one of our WSDM colleagues could comment further.
> > But from my read of the statement you quoted from WSDM MUWS,
it was
> > simply observing that a manageability endpoint (eg a Web service
> > endpoint) may be the front end for multiple resources.  Therefore
the
> > IRP must be used to identify any single one of those resources.
 I
> > didn't get the read that there was any capability implied to
access a
> > group of more than one resource with a single EPR.  This
is 
> > not possible
> > from what I understand of WS-Addressing. 
> > 
> > sgg 
> > 
> > ++++++++
> > Steve Graham
> > (919)254-0615 (T/L 444)
> > STSM, On Demand Architecture
> > Member, IBM Academy of Technology
> > <Soli Deo Gloria/>
> > ++++++++
> > 
> > 
> > 
> > 
> >                  David
Hull <dmh@tibco.com> 
> > 
> > 08/05/2004 04:14 PM 
> > 
> >                  
       
> >         To:        wsn@lists.oasis-open.org
> >         cc:         
> >         Subject:        [wsn]
Multiple endpoints in IRP?
> > 
> > 
> > 
> > While reading through WSDM MUWS, I ran across this interesting
> > statement:
> > 
> > "If the manageability endpoint corresponds to a variable
> > number (zero or
> > 
> > more) of manageable resources, then the WSRF Implied Resource
Pattern 
> > MUST be followed.  This means that the element(s) listed
in the 
> > ReferenceProperties of a WS-Resource qualified EPR must be 
> > included in 
> > the header of messages sent to such manageability endpoints."
> > 
> > This seems to imply a WS-Address with multiple ReferenceProperties
> > elements, one for each EPR.
> > 
> > Does the IRP allow for multiple resources to be addressed as
> > a group?  
> > Does this have to be done via repeated ReferenceProperties 
> > elements, or 
> > could one give a predicate instead, or bake something magic into
the 
> > Address URI?
> > 
> > I realize that the IRP is in a fluid state at the moment, but
I'd be 
> > interested to know current thinking.  Is this the sort of
> > issue the WSRP
> > 
> > subgroup is trying to address?
> > 
> > 
> >