W3C home > Mailing lists > Public > public-ws-addressing@w3.org > December 2004

RE: Another way of thinking about EPRs

From: Srinivas, Davanum M <Davanum.Srinivas@ca.com>
Date: Fri, 10 Dec 2004 09:31:40 -0500
Message-ID: <87527035FDD42A428221FA578D4A9A5B08907C83@usilms24.ca.com>
To: "Christopher B Ferris" <chrisfer@us.ibm.com>
Cc: "Martin Gudgin" <mgudgin@microsoft.com>, <public-ws-addressing@w3.org>

Let's take this use case then, Assuming that i am indeed using "refps
exclusively for application of the WS-RF implied resource pattern"
[possibly for monitoring a Grid environment, log into a web site, get an
EPR emailed to you, then use that EPR to access a grid service????].
wsa:To, wsa:From, wsa:MessageID, wsa:RelatesTo does not tell me which
resource is being used and I cannot collect aggregate statistics on how
many times it was used. And also assume that services are being added to
the grid on a dynamic basis (which means that I cannot keep entering all
Qname's of all possible RefProperties by hand). I want to do this
transparently and show a neat and nice UI with all resources
(automatically discovered, keyed by Address+ReferenceProperties) and
statistics about each resource.

Is this a valid use of WS-Addressing/EPR's/RefP's? or it is too
outland-ish?

-- dims

-----Original Message-----
From: Christopher B Ferris [mailto:chrisfer@us.ibm.com] 
Sent: Friday, December 10, 2004 9:02 AM
To: Srinivas, Davanum M
Cc: Martin Gudgin; public-ws-addressing@w3.org;
public-ws-addressing-request@w3.org
Subject: RE: Another way of thinking about EPRs

Dims,

You wrote:
>Section 2.4 of submission[1] talks about  comparing 2 EPR's. If I have 
>to write a completely transparent BPEL or  WS-Choreography monitoring 
>solution, I need to be able to look at the  wire and be able to figure 
>out who is talking to who and keep track of  interactions between 
>entities in the system. *IF* from the soap message  on the wire I can 
>figure out the "who" portion..then I am all set.

Again, I must ask; why aren't <wsa:To/> and <wsa:From/> sufficient to
achieve this objective? 

I could be using the refps for applying the implied resource pattern
(WS-RF), but that is not a required application of refps. Does your
monitoring of SOAP traffic have intimate knowledge about PONumber
embedded in the soap:Body? I doubt it would be even remotely practical
as it would require domain knowledge of all application-domains on the
part of the monitoring software.
Yet, a soap:Body//PONumber identifies a resource relevant to the service
just as a refp might do.

Bottom line, what is the "who" you think you are tracking? The
resource(s) behind a service endpoint or the service itself? If the
former, then frankly, unless refps are used exclusively for application
of the WS-RF implied resource pattern, you are probably making a false
assumption.

If you are tracking for purposes of matching requests with responses,
then you have wsa:MessageId and wsa:RelatesTo. The address becomes
immaterial. 

Cheers,

Christopher Ferris
STSM, Emerging e-business Industry Architecture
email: chrisfer@us.ibm.com
blog: http://webpages.charter.net/chrisfer/blog.html
phone: +1 508 377 9295

public-ws-addressing-request@w3.org wrote on 12/09/2004 10:13:24 PM:

> 
> Martin, All,
> 
> Yes, I get it...Here's a variation of the argument I have been making 
> for i008 (not sure about i001). Section 2.4 of submission[1] talks 
> about comparing 2 EPR's. If I have to write a completely transparent 
> BPEL or WS-Choreography monitoring solution, I need to be able to look

> at the wire and be able to figure out who is talking to who and keep 
> track of interactions between entities in the system. *IF* from the 
> soap message on the wire I can figure out the "who" portion..then I am
all set.
> According to 2.4, I can use the combination of [address] and 
> [reference properties] to figure out A is sending the message to B. 
> BUT if I don't have access to the EPR's themselves or list of all the 
> Qname's that are definitely reference properties, I cannot find out 
> the "who". IF there is some "annotation" (word used by Jonathan on 
> IRC) on the soap message on the wire then I can do this VERY easily.
> 
> So the basic question is - Is there consensus on whether we need some 
> "annotation"? (could be wrappers, could be attributes, could be 
> something else entirely).
> 
> [1] http://www.w3.org/Submission/2004/SUBM-ws-addressing-20040810/
> 
> Thanks,
> dims
> 
> -----Original Message-----
> From: public-ws-addressing-request@w3.org
> [mailto:public-ws-addressing-request@w3.org] On Behalf Of Martin 
> Gudgin
> Sent: Thursday, December 09, 2004 6:47 PM
> To: public-ws-addressing@w3.org
> Subject: Another way of thinking about EPRs
> 
> 
> I've noticed that whenever we talk about issues i001 and i008, that 
> there is an implicit assumption that we start with an EPR and go from 
> that to the message. I think that in many cases, the reverse is 
> actually true, that is, people start with SOAP messages with headers 
> in and then decide how to communicate to a potential sender what those

> messages should look like ( WRT the headers ).
> 
> For example, I want to have people send me messages that look like the

> three below.
> 
> <soap:Envelope>
>  <soap:Header>
>    <wsa:To>http://example.org/weather</wsa:To>
>    <m:ServiceLevel>Gold</m:ServiceLevel>
>    <m:TxId>1234</m:TxId>
>  </soap:Header>
>  <soap:Body>
>  . . .
>  </soap:Body>
> </soap:Envelope>
> 
> <soap:Envelope>
>  <soap:Header>
>    <wsa:To>http://example.org/weather</wsa:To>
>    <m:ServiceLevel>Silver</m:ServiceLevel>
>    <m:TxId>1234</m:TxId>
>  </soap:Header>
>  <soap:Body>
>  . . .
>  </soap:Body>
> </soap:Envelope>
> 
> <soap:Envelope>
>  <soap:Header>
>    <wsa:To>http://example.org/weather</wsa:To>
>    <m:ServiceLevel>Gold</m:ServiceLevel>
>    <m:TxId>4567</m:TxId>
>  </soap:Header>
>  <soap:Body>
>  . . .
>  </soap:Body>
> </soap:Envelope>
> 
> How can I communicate to my users that I want the messages to look 
> like this? Ah, I know, I'll send them an EPR. Here are the three EPRs 
> for the messages above ( my seperation between Props/Params is
arbitrary ).
> 
> <wsa:EndpointReference>
>  <wsa:Address>http://example.org/weather</wsa:Address>
>  <wsa:ReferenceProperties>
>   <m:ServiceLevel>Gold</m:ServiceLevel>
>  </wsa:ReferenceProperties>
>  <wsa:ReferenceParameters>
>   <m:ServiceLevel>1234</m:ServiceLevel>
>  </wsa:ReferenceParameters>
> </wsa:EndpointReference>
> 
> <wsa:EndpointReference>
>  <wsa:Address>http://example.org/weather</wsa:Address>
>  <wsa:ReferenceProperties>
>   <m:ServiceLevel>Silver</m:ServiceLevel>
>  </wsa:ReferenceProperties>
>  <wsa:ReferenceParameters>
>   <m:ServiceLevel>1234</m:ServiceLevel>
>  </wsa:ReferenceParameters>
> </wsa:EndpointReference>
> 
> <wsa:EndpointReference>
>  <wsa:Address>http://example.org/weather</wsa:Address>
>  <wsa:ReferenceProperties>
>   <m:ServiceLevel>Gold</m:ServiceLevel>
>  </wsa:ReferenceProperties>
>  <wsa:ReferenceParameters>
>   <m:ServiceLevel>4567</m:ServiceLevel>
>  </wsa:ReferenceParameters>
> </wsa:EndpointReference>
> 
> Does this make any sense?
> 
> Gudge
> 
> 
> 
> 
Received on Friday, 10 December 2004 14:32:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:35:00 GMT