- From: Francisco Curbera <curbera@us.ibm.com>
- Date: Tue, 26 Apr 2005 13:57:54 -0400
- To: "Savas Parastatidis" <Savas.Parastatidis.Savas.Parastatidis@newcastle.ac.uk>
- Cc: humphrey@cs.virginia.edu, public-ws-addressing@w3.org, public-ws-addressing-request@w3.org, wasson@virginia.edu
The rationale behind those words is to protect client code from possible
changes in EPRs introduced by the issuer. The idea is that client code is
more robust if it is built in such a way that it takes no dependency on the
specific values, schema or overall structure of the reference parameter
elements.
On the other hand, there is no way (and probably no reason) to prevent
specific communities from defining ad-hoc conventions about the information
that these elements carry. This is very similar to the URI opaqueness
property: an architectural principle that is often violated for various
(sometimes legitimate) reasons.
Given that, I don't think anything like a MUST or a SHOULD is appropriate,
since it would just prevent the normal development of perfectly good usage
scenarios and likely end up confusing everyone (the WG itself spent a lot
of time debating the consequences of those words for example).
Paco
"Savas Parastatidis"
<Savas.Parastatidis@newca To: <public-ws-addressing@w3.org>
stle.ac.uk> cc: <humphrey@cs.virginia.edu>, <wasson@virginia.edu>
Sent by: Subject: Language for Reference Parameters
public-ws-addressing-requ
est@w3.org
04/26/2005 09:40 AM
Dear all,
This issue came up during some discussions with folks at the University
of Virginia (Marty and Glenn cc'd) while discussing an implementation of
WS-Transfer and how EPRs were used...
My understanding of the Ws-Addressing specification is that the
[Reference Parameters] information element item is opaque and that its
children should be included in messages as header information elements
without change. The specification states:
"Reference parameters are provided by the issuer of the endpoint
reference and are assumed to be opaque to consuming applications."
The "assumed" bit is the reason for this message. Should there be
normative language here? Either SHOULD or MUST? Are the consumers of an
EPR allowed to reason about the contents of the [reference parameters]?
My apologies if the above has already been discussed and a decision
already made.
Regards,
--
Savas Parastatidis
http://savas.parastatidis.name
Received on Tuesday, 26 April 2005 17:58:04 UTC