wsrf message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: RE: [wsrf] Singleton Resource Pattern
- From: "Sedukhin, Igor S" <Igor.Sedukhin@ca.com>
- To: "Steve Graham" <sggraham@us.ibm.com>
- Date: Mon, 12 Jul 2004 17:25:08 -0400
It does not know what is the "key". It does not need
to. Using WS-RP it knows that there is a UserName property and its data
type (from the WSDL) and it just asks <GetProperty>UserName</GetProperty> and receives a response.
That is what the client wants to know. The client sends
WS-Security headers just like it would otherwise, without any relation to the
WS-RP.
I don't understand how an EPR would help this
scenario...
-- Igor
Sedukhin .. (igor.sedukhin@ca.com)
-- (631) 342-4325 .. 1 CA Plaza,
Islandia, NY 11788
how does the client
know that UserName is the proper lookup "key" to the stateful resource? How does
the client know which values of UserName are valid?
With the EPR approach, this is not an issue for the
client.
sgg
++++++++
Steve Graham
(919)254-0615 (T/L
444)
STSM, On Demand Architecture
Member, IBM Academy of
Technology
<Soli Deo Gloria/>
++++++++
| "Sedukhin, Igor S"
<Igor.Sedukhin@ca.com>
07/12/2004 04:22 PM
| To:
Steve Graham/Raleigh/IBM@IBMUS cc:
Rich Thompson/Watson/IBM@IBMUS,
<wsrf@lists.oasis-open.org> Subject: RE:
[wsrf] Singleton Resource Pattern |
All what the client needs to understand is that
there is a UserName property and that GetProperty message could be sent. The
rest is regular Web services techniques and mechanisms. What WS-RP could add to
that mix is a normative defiition of how to retreive a propery value and how to
find out what properties are there.
One could combine the above with
WS-Addressing or not. That should not affect semantics defined by WS-RP.
-- Igor
Sedukhin .. (igor.sedukhin@ca.com)
--
(631)
342-4325 .. 1 CA Plaza, Islandia, NY 11788
From: Steve Graham [mailto:sggraham@us.ibm.com]
Sent: Monday, July 12, 2004 4:15 PM
To: Sedukhin, Igor
S
Cc: Rich Thompson; wsrf@lists.oasis-open.org
Subject: RE:
[wsrf] Singleton Resource Pattern
Clearly we cannot prevent implementors from doing
this sort of thing.
However, by specifying the mechanism with an EPR, we standardize ONE
approach to combining the Web service address and the component that properly
identifies or disambiguates the stateful resource. The approach you
suggest requires much more understanding on the client's part on the association
between the WS-Security header, proper values of that header and the web
service. Having the EPR eliminates this additional burden on the client's
application.
sgg
++++++++
Steve Graham
(919)254-0615 (T/L 444)
STSM, On
Demand Architecture
Member, IBM Academy of Technology
<Soli Deo
Gloria/>
++++++++
| "Sedukhin, Igor S"
<Igor.Sedukhin@ca.com>
07/12/2004 03:50 PM
|
To:
Rich Thompson/Watson/IBM@IBMUS,
<wsrf@lists.oasis-open.org> cc:
Subject:
RE: [wsrf] Singleton Resource Pattern |
Why is X.509
certificate in a WS-Security header not a sufficient means for referring to a
proper "user" database record?
The property is shared, there is one service,
and there are multiple "records". Why can't I use WS-ResorceProperties spec to
access UserName property without having to deal with stateful/stateless
situation?
It seems that ANYTHING in a message (body or a header) could be such
"stateID". There may be no reason to set in stone what that is. Leave it to the
implementers.
-- Igor
Sedukhin .. (igor.sedukhin@ca.com)
--
(631)
342-4325 .. 1 CA Plaza, Islandia, NY 11788
From: Rich Thompson [mailto:richt2@us.ibm.com]
Sent: Monday, July 12, 2004 3:40 PM
To:
wsrf@lists.oasis-open.org
Subject: RE: [wsrf] Singleton Resource
Pattern
But
with the stateless character of web services, that UserName property would be
shared by all users of the web service. There may be some usefulness to that,
but it becomes far more useful when the web service interface fronts access to a
multiplicity of state, each storing such a property.
To make this a
concrete example, presume the UserName property is really a reference to a field
in a database. One could provide a separate web service interface for each
record within the database, but it would be a much lighter (and frequently more
useful) definition if there was a single web service interface that had some
well defined means for indicating what record of the database was the target of
the current invocation. WSRP had this type of need and had to define a means
that is idiosyncratic to our protocol for passing the stateful reference.
Defining a standardized means by which infrastructure can take more of the load
for handling such references is one of the outcomes I would like to see from the
WSRF effort.
Rich Thompson
OASIS WSRP TC Chair
"Sedukhin, Igor S"
<Igor.Sedukhin@ca.com>
07/12/2004 03:15 PM
|
To
| Rich
Thompson/Watson/IBM@IBMUS,
<wsrf@lists.oasis-open.org>
|
cc
|
|
Subject
| RE: [wsrf] Singleton
Resource Pattern |
|
Like I said,
GetProperty would mean for example:...
Web Service implements a UserName property.
Client sends a GetResourceProperty SOAP message with WS-Security headers
containing X.509 certificate. Web Service returns the UserName after matching
the certificate.
-- Igor
Sedukhin .. (igor.sedukhin@ca.com)
--
(631)
342-4325 .. 1 CA Plaza, Islandia, NY 11788
From: Rich Thompson [mailto:richt2@us.ibm.com]
Sent: Monday, July 12, 2004 3:03 PM
To:
wsrf@lists.oasis-open.org
Subject: RE: [wsrf] Singleton Resource
Pattern
What would
get/setProperty mean against the standard stateless character of the base Web
Services definition? Before these become useful, you need some manner of
modeling state. It seems to me the primary thrust behind defining WSRF is having
a standard definition of how to model such stateful web services and as someone
who has needed to idiosyncratically model state within the WSRP protocol, I see
a lot of value in that effort.
Rich Thompson
OASIS WSRP TC Chair
"Sedukhin, Igor S"
<Igor.Sedukhin@ca.com>
07/12/2004 02:55 PM
|
To
| Steve
Graham/Raleigh/IBM@IBMUS
|
cc
| <wsrf@lists.oasis-open.org>
|
Subject
| RE: [wsrf] Singleton
Resource Pattern |
|
Then what
value does the implied pattern add to the use case I have included?
Why does resource
pattern have to depend on WS-Addressing, EPRs and customizations of
those?
Why
can't GetProperty/SetProperety be defined independently?
-- Igor
Sedukhin .. (igor.sedukhin@ca.com)
--
(631)
342-4325 .. 1 CA Plaza, Islandia, NY 11788
From: Steve Graham [mailto:sggraham@us.ibm.com]
Sent: Monday, July 12, 2004 2:44 PM
To: Sedukhin, Igor
S
Cc: wsrf@lists.oasis-open.org
Subject: RE: [wsrf]
Singleton Resource Pattern
>It seems that whether it is an "implied resource" or an
"implied singleton" has nothing to do with GetResourceProperty (GetProperty for
that sake) operation. In >other words use of WS-Addressing does not add any
semantic value to the fact that one could retrieve a property by sending a
message to a service.
Unfortunately, I disagree. The implied resource pattern
clarifies a certain pattern relating a web service and a stateful resource. It
is important that this pattern is clarified to allow a single web service to act
as the Web services message processor for a plurality of stateful resources.
Therefore the pattern of the message is formed to disambiguate which of the
potentially many stateful resources is associated with the message is very
important.
sgg
++++++++
Steve Graham
(919)254-0615 (T/L 444)
STSM, On
Demand Architecture
Member, IBM Academy of Technology
<Soli Deo
Gloria/>
++++++++
| "Sedukhin, Igor S"
<Igor.Sedukhin@ca.com>
07/12/2004 11:56 AM
|
To:
Steve Graham/Raleigh/IBM@IBMUS,
<wsrf@lists.oasis-open.org> cc:
Subject: RE: [wsrf] Singleton Resource
Pattern |
Steve,
Your example was
interesting. How about this one:
Web Service implements a UserName property. Client
sends a GetResourceProperty SOAP message with WS-Security headers containing
X.509 certificate. Web Service returns the UserName after matching the
certificate.
-- Is this a singleton or an implied resource pattern? One could
claim that there is a user resource, however this interaction does not use
WS-Addressing and moreover it would not be possible to build WS-RF qualified
EPRs for such interaction.
The same use case could be modified to include
UserDiskQuota property, and the same argument would apply.
It seems that whether it is an
"implied resource" or an "implied singleton" has nothing to do with
GetResourceProperty (GetProperty for that sake) operation. In other words use of
WS-Addressing does not add any semantic value to the fact that one could
retrieve a property by sending a message to a service.
-- Igor
Sedukhin .. (igor.sedukhin@ca.com)
--
(631)
342-4325 .. 1 CA Plaza, Islandia, NY 11788
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]