[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [wsrf] Comments on wsrf-ws_resource-1.2-spec-wd-07
"Can one refer to a WS-Resource without a Resource identifier?"
On the face of it this appears to be a simple question but in fact hides a
degree of complexity, since the actual representation of the resource
identifier is determined by the WS-Resource provider (i.e the creator of
the EPR).
The requirement WS-RF has is that the provider of a WS-Resource must be
able to distinguish which specific resource should be used in the
processing of the messages defined in the WS-RF specifications.
Specifically, there must be enough information in the message for the
WS-Resource-provider to be able to do this. We have just deleted text that
made this requirement explicit on the grounds that it can be logically
deduced that IF an EPR identifies exactly one WS-Resource then messages to
the WS-Resource MUST contain (following the rules defined in WS-Addressing
) the information required for the WS-Resource to disambiguate the specific
resource. However, we make no statement that an EPR MUST identify exactly
one WS-Resource and so we have a problem.
Previously the spec stated:
A reference to a WS-Resource is represented by an endpoint reference, or
more precisely an XML element whose type is, or is derived (by
extension) from the complexType named EndpointReferenceType defined by
the [WS-Addressing] specification.
An identifier of the resource MUST be represented in any reference to a
WS-Resource. and MUST appear as part of any message to a WS-Resource to
allow the WS-Resource to disambiguate the resource targeted by the
message. The precise location of the resource identifier in a message to
a WS-Resource is dependent on the protocol binding used to interact with
the WS-Resource endpoint but is normatively defined in the appropriate
WS-Addressing binding specification. For example, [WSA – SOAP] defines
the binding of message addressing properties for the SOAP protocol.
and then we removed the 2nd bullet, which leaves us with a problem of
under-specificity.
We need to reinstate the requirement that this 2nd bullet conveyed. Perhaps
the following single bullet would be better:
A reference to a WS-Resource is represented by an endpoint reference, or
more precisely an XML element whose type is, or is derived (by
extension) from the complexType named EndpointReferenceType defined by
the [WS-Addressing] specification. A reference to a WS-Resource
encapsulates sufficient information to uniquely distinguish which
resource should be used by the WS-Resource when it processes messages.
There is no need for an upper case MUST in this because there is no
interoperability issue - it is simply helping the reader to understand that
WS-Addressing provides the appropriate mechanism for resource
disambiguation required by WS-RF.
If we go with the above, then the definition of "Resource Identifier"
becomes redundant and could be removed.
Regards,
Ian Robinson
STSM, WebSphere Messaging and Transactions Architect
IBM Hursley Lab, UK
ian_robinson@uk.ibm.com
"Wilson, Kirk D"
<Kirk.Wilson@ca.c
om> To
Ian Robinson/UK/IBM@IBMGB, "Murray,
07/09/2005 21:40 Bryan P." <bryan.murray@hp.com>
cc
<wsrf@lists.oasis-open.org>
Subject
RE: [wsrf] Comments on
wsrf-ws_resource-1.2-spec-wd-07
As I recall the previous discussion, the issue resolved around the
normative MUST in "An identifier of the resource MUST be represented in
any reference to a WS-Resource."
The issue seems to be (and I don't pretend to be able to answer this
question but would be VERY interested in the answer):
Can one refer to a WS-Resource without a Resource identifier?
If not, then use of an identifier is a necessity and not
normative (small "must").
If so, why are we requiring the use of an identifier? Is the
use of an identifier necessary for interoperability, in which case it
would be normative?
If so and it is not necessary for interoperability, then it is
a
recommendation and we should probably say MAY (with a strong
recommendation to do so).
Kirk Wilson
Architect, Development
Office of the CTO
802 765-4337
-----Original Message-----
From: Ian Robinson [mailto:ian_robinson@uk.ibm.com]
Sent: Wednesday, September 07, 2005 11:42 AM
To: Murray, Bryan P.; Wilson, Kirk D
Cc: wsrf@lists.oasis-open.org
Subject: RE: [wsrf] Comments on wsrf-ws_resource-1.2-spec-wd-07
One reason it is now difficult to relate the terms "resource identifier"
and "reference to a WS-Resource" is that I removed (per the resolution
to
issue 127) the following text which related these 2 concepts:
"An identifier of the resource MUST be represented in any reference to a
WS-Resource. and MUST appear as part of any message to a WS-Resource to
allow the WS-Resource to disambiguate the resource targeted by the
message.
The precise location of the resource identifier in a message to a
WS-Resource is dependent on the protocol binding used to interact with
the
WS-Resource endpoint but is normatively defined in the appropriate
WS-Addressing binding specification. For example, [WSA - SOAP] defines
the
binding of message addressing properties for the SOAP protocol."
It is not clear to me that this text repeats any part of WS-Addressing -
it
is a description of the WS-RF usage of EPRs.
I would agree with Kirk's observation and suggest that we reinstate some
or
all of the above text. At the very least, it would seem that we should
still state: "An identifier of the resource MUST be represented in any
reference to a WS-Resource. "
Comments?
Regards,
Ian Robinson
STSM, WebSphere Messaging and Transactions Architect
IBM Hursley Lab, UK
ian_robinson@uk.ibm.com
"Murray, Bryan
P."
<bryan.murray@hp.
To
com> "Wilson, Kirk D"
<Kirk.Wilson@ca.com>, Ian
06/09/2005 20:48 Robinson/UK/IBM@IBMGB,
<wsrf@lists.oasis-open.org>
cc
Subject
RE: [wsrf] Comments on
wsrf-ws_resource-1.2-spec-wd-07
The reason ResourceIdentifier is NOT used is that an EPR does not
provide "identity" for a resource. An EPR is only a reference to a
resource.
The use and rules associated with reference parameters is covered in
WS-Addressing and should not be covered in this spec.
Bryan
-----Original Message-----
From: Wilson, Kirk D [mailto:Kirk.Wilson@ca.com]
Sent: Tuesday, September 06, 2005 12:08 PM
To: ian_robinson@uk.ibm.com; wsrf@lists.oasis-open.org
Subject: [wsrf] Comments on wsrf-ws_resource-1.2-spec-wd-07
There now seems to be three concepts used with the space two pages:
Resource Identifier
Reference to a WS-Resource (which is structurally tied to
WS-Addressing)
And, in the SOAP example: SomeDisambiguatorElement
These three concepts are never explicitly related to one other.
Obviously, SomeDisabmiguator is part of the reference to a WS-Resource
(ReferenceParameters). Further, I assume, after carefully trying to
connect things that are in the text, that the Resource Identifier is at
least one possible "disambiguator element". If so, would be clearer to
use <ResourceIdentifier>R1</ResourceIdentifier> rather than
<SomeDisambiguatorElement> in the example? After all, the text does say
that "R1" "identifies" the resource
Kirk Wilson
Architect, Development
Office of the CTO
802 765-4337
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]