RE: [wsrf] Minutes of the telecon on Monday August 8th

["Wilson, Kirk D" <Kirk.Wilson@ca.com>]

Tim--great job on capturing the discussion on Issue 127.  As a relative
newbie to these issues, I especially appreciate your minutes.

There seems to be a number (possibly separate but definitely
interrelated) issues:
1.  Must there be an identifier in a message?  Except for one dissension
by William, the general consensus appears to be Yes.
2.  Is a normative reference to RAP required?  Except for the dissension
at the end by SteveG, it was generally agreed that the reference to RAP
could be eliminated in favor of a direct reference to WS-Addressing.
3.  So, is a normative reference to WS-Addressing required?  I will
agree with DaveS's comment that a URL will work.  This point seems
especially true following W3C's elimination of the ReferenceProperties.
We are just dealing with ReferenceParameters and they can just as well
occur in the URL.  But I think we should make WS-Addressing a
*structural* requirement on the definition of the WS-Resource reference.
References just ARE EPRs. ("Makes life easier.")  WS-Addressing is HOW
we require point #1 to be satisfied.  Thus, the requirement for
WS-Addressing is NOT normative in the normative sense of MUST, though a
reference to a WS-Resource must involve WS-Addressing.)
4.  Is there or is there not an interop problem in all this and, if so,
where is the interop problem?  (DaveS & William vs. Ian & SteveG).  As I
understand interoperability, if the use of WS-Addressing is structural,
then interoperability is not a problem wrt WS-Addressing. (It is matter
of compliance or non-compliance.)  I believe it is an interop problem
with respect #1--messages will break without the proper identifier.  Do
I understand interoperability correctly?  (So, as per the bulleted item
under dispute: "An identifier of the resource MUST be represented *in
any reference* to a WS-Resource to allow . . .[disambiguation]", but I
think it would help to avoid confusion if everything else in Ian's
proposed revision(about message structure and RAP)--essentially the
original bullet--is deleted.  I can see why the commenter objected to
the statement "We refer to *this* pattern..."  There's no *pattern*
there that an implementer could follow just to say the message must
contain the identifier.  If it is decided to keep the reference to RAP,
it should be part of the WS-R reference definition (which should follow,
IMHO, Ian's proposed revision).  
5.  William's last issue: Can we eliminate the WS-Resource spec and
distribute the definition of WS-Resource and the normative parts to
other documents?  (Oh my--probably, yes--technically.  But for user
readability it may be better to keep them separate.)

Hopefully, this provides a little help--perhaps only to focus the
issues.

Kirk Wilson
Architect, Development
Office of the CTO
802 765-4337
 
-----Original Message-----
From: Tim Banks [mailto:tim_banks@uk.ibm.com] 
Sent: Thursday, August 11, 2005 7:24 AM
To: wsrf@lists.oasis-open.org
Subject: [wsrf] Minutes of the telecon on Monday August 8th

The minutes are stored here:
http://www.oasis-open.org/committees/download.php/13996/WSRF%20TC%20%5B8
Aug05%5D%20notes%5B1%5D.pdf

And attached.


Regards, Tim Banks.