ws-rx message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: RE: [ws-rx] RE: [WS-RX] Action 0040
- From: "Patil, Sanjay" <sanjay.patil@sap.com>
- To: "Doug Davis" <dug@us.ibm.com>, <ws-rx@lists.oasis-open.org>
- Date: Wed, 5 Oct 2005 15:40:58 -0700
+1
I would also say that we should not spend too much time in
analyzing how certain past version of WS-Policy specification, etc, used
the term "observed".
Given that:
- This TC decided to adopt some text that uses the term
"observed"
- Some members see the need for clarifying the meaning
of "observed" and the TC accepted an issue in this regard
- The term "observed" is not normatively defined in
any other standard and
- Our intended semantics is not perfectly matching with the
definitions elsewhere,
we should come up with text that reflects our intended
semantics.
Ashok, if you agree with the above, could you please make a
formal proposal with self contained text for defining the term "observed".
I am not suggesting that this is the best approach but that will at least help
us have a focused discussion of the core issue.
Thanks,
Sanjay
Ashok,
would it be possible for you to write-up a formal
proposal for this. Its hard to know from this note exactly which kind of
change you're proposing. Having the exact spec changes you'd like to see will
speed up the discussions.
thanks
-Doug
Ashok Malhotra
<ashok.malhotra@oracle.com>
10/05/2005 06:20 PM
Please respond
to ashok.malhotra |
|
To
| ws-rx@lists.oasis-open.org
|
cc
|
|
Subject
| [ws-rx] RE: [WS-RX]
Action 0040 |
|
> Action: Ashok to write proposed clarification text on meaning of
> "observed" in this spec.
I sent earlier mail on this which
said:
"The word 'observed' in the above paragraph should be interpreted
in the sense of wsp:Observed which, in earlier versions of WS-Policy (see
ftp://www6.software.ibm.com/software/developer/library/ws-policy2003.pdf),
was
a value for the wsp:Usage attribute that identified how the assertion was to be
processed. Assertions that are 'Observed' become properties of the policy
subject but do not influence the content of messages."
If you read the
spec, though, it defines wsp:Observed as follows:
"This assertion will be
applied to all subjects and requesters of the service are informed that the
policy will be applied". This is not quite the semantic we agreed to and I
think we need a little more discussion. My apologies! I relied on memory
instead of checking the spec.
When the RM assertions are applied to a
WSDL, I'm assuming that this is the WSDL between the AS and the AD. These
are the two participants in the exchange and it's their messages that the WSDL
will define. If the messages are to be exchanged reliably the AS or AD
will add the RM assertions to the WSDL. But note that the WSDL defines
the messages in the reliable sequence and not the CreateSequence,
CreateSequenceResponse, etc. which are framing messages that manage the
protocol.
I think the semantic we require for this usage is that the
second half of the wsp:Observed quoted earlier "requesters of the service are
informed that the policy will be applied" but not the first half that says that
the assertion will be applied because these assertions are not applied to the
messages in the sequence.
If there is a requirement that the information
in the RM assertions should be carried in (some of) the framing messages, we
will need to devise a mechanism to say that.
All the best,
Ashok
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]