RE: [uddi-spec] Comments on the use of fragment identifiers in the "UsingWSDL in a UDDI Registry, Version 2"

[Tom Bellwood <bellwood@us.ibm.com>]

I would consider these changes to be editorial in nature, certainly in
keeping with what one would expect during a review period, so a repeat of
the 30-day review period should not be necessary.   We will, of course, all
vote on the ammended version.

Thanks,
Tom Bellwood       Phone:  (512) 838-9957 (external);   TL:  678/9957
(internal)
Co-Chair, OASIS UDDI Specification TC
STSM - Emerging Technologies
IBM Corporation

"John Colgrave" <colgrave@hursley.ibm.com> on 05/12/2003 07:15:07 AM

To:    <uddi-spec@lists.oasis-open.org>
cc:
Subject:    RE: [uddi-spec] Comments on the use of fragment identifiers in
       the "Using WSDL in a UDDI Registry, Version 2"





Luc,



Having checked the 1.08 BP again, I agree that replacing MUST with SHOULD
is in keeping with the wording in 1.08 so I have made the change.  I also
changed the MUST to SHOULD in Appendix C itself.



Surely this does not require another 30-day review though?  The TN has had
a 30-day review and the only issues that were brought up were minor
editorial ones with the exception of the issue that has become CR-032.  I
was hoping that, assuming CR-032 passes, I would make available one final
version that had all the changes (including this fragment identifier
change) applied and we would allow a week or so for people to check that
the issues had been satisfactorily addressed and then we would vote to
publish the TN.  If we have to start another 30-day review every time we
change a word we will never publish anything!



I will send you the Word file once I have made all the changes if you are
happy to produce the HTML and the PDF.



John Colgrave

IBM



-----Original Message-----
From: Luc Clement [mailto:lclement@windows.microsoft.com]
Sent: 10 May 2003 22:58
To: John Colgrave; uddi-spec@lists.oasis-open.org
Subject: RE: [uddi-spec] Comments on the use of fragment identifiers in the
"Using WSDL in a UDDI Registry, Version 2"



John,



Here's what I'd be content with: remove the use of "MUST" and replace it
with "SHOULD" so that section 2.3.6 reads as follows:

"...the optional fragment identifier is used, then the value of the
overviewURL SHOULD conform to the syntax described in Appendix C."

This would at least be consistent with 1.08 and not require the use of a WD
that is no longer in development and arguably
abandoned. As you suggest, when things settle down, we can incorporate the new developments (either
 those introduced in WSDL 1.1/1.2 or by the XML media registration which is
expected) when we transition the TN to a BP. We could update 1.08 and 2.0
at the same time.



If you agree, please send me the latest version of the documents with this
update and I'll have it posted to
http://www.oasis-open.org/committees/uddi-spec/doc/draft/ and begin the
30-day review (it might get accepted before its birthday after all).



Luc

Luc Clément
Microsoft

Note: I'll gladly produce the HTML and PDF version for you; the version of
Word that I'm using produces better quality HTML than earlier versions. Let
me know if you would like to take up this offer.



From:John Colgrave [mailto:colgrave@hursley.ibm.com]
Sent: Saturday, May 10, 2003 07:46
To: uddi-spec@lists.oasis-open.org
Subject: RE: [uddi-spec] Comments on the use of fragment identifiers in the
"Using WSDL in a UDDI Registry, Version 2"

Luc,



Given that:

1.      The use of fragment identifiers is optional, and not what we
recommend as a best practice.

2.      We updated the V1 BP to 1.08 just to add the same XPointer
fragments, so if we change the V2 TN we will introduce a
migration/compatibility problem.  (Or are you proposing a 1.09 BP as well
to change what we did for 1.08?)

3.      As you say there are no simple alternatives.

4.      The WSDL community is still discussing the issue for WSDL 1.2 and
it is not yet clear if the approach chosen will be retro-fitted to WSDL
1.1.

5.      I would like to publish the TN before its first birthday.

I propose that we leave the TN as it is.  If things settle down in the near
future we can incorporate the new developments when we transition the TN to
a BP.



There is a later Working Draft than [5], from December 2002, so it does not
seem to be entirely dormant, although the other pieces of XPointer have
advanced to Recommendation.



John Colgrave

IBM



-----Original Message-----
From: Luc Clement [mailto:lclement@windows.microsoft.com]
Sent: 10 May 2003 06:50
To: John Colgrave
Cc: uddi-spec@lists.oasis-open.org
Subject: [uddi-spec] Comments on the use of fragment identifiers in the
"Using WSDL in a UDDI Registry, Version 2"



John,

I’ve received feedback that [5] referenced in the TN is a Working Draft
that is now no longer in development because it was too big and complex to
be widely implemented.

In section 2.3.6 the TN states that if “the optional fragment identifier is
used, then the value of the overviewURL MUST conform to the syntax
described in Appendix C.” The use of “MUST” is obviously problematic if we
are going to continue to reference [5].

Unfortunately there are no simple alternatives and a perfect ready-made
solution for us at this time within the context of the XPointer Framework
[1] which represents the follow-on work to [5]. As it stands, the XPointer
Framework does not address our immediate needs:

o        The XPointer Framework [1] “#something” does not work for us given
that WSDL doesn’t support ids

o        The XPointer element() scheme [2] “#element(/1/2/6/5)” doesn’t
make much sense to us given that it only works if a WSDL document that
doesn’t change



We could consider the use of xPath and use an expression such as:

http://location/sample.wsdl#xmlns(wsdl=http://schemas.xmlsoap.org/wsdl/)xmlns(uddi=http://oasis-open.org/xpath)uddi:xpath(/wsdl:definitions/wsdl:portType[@name='StockQuotePortType"])

Doing so may prove to be more readability attainable given wide
availability of xPath processors, but would require us to define this
scheme. As an alternatives we could develop a scheme using the extension
mechanism “#xmlns(uddi=http://oasis-open.org/fragments
)uddi:portType(StockQuotePortType)”.  For example:
http://location/sample.wsdl#xmlns(uddi=http://oasis-open.org/fragments)uddi:portType(StockQuotePortType)

In short, there is no ready made solution. I’m told (i.e. I haven’t checked
for myself) that the XML media type registration (which should be updated
soon to point to the XPointer Framework), will allow people to use
element(), xpointer() or any other fragment scheme that comes along.
Constraining these extension points to improve interop can happen later on.

To summarize, at issue with section 2.3.6 of the TN are:

a. do we require a “MUST”? I propose no.
b. do we reference [5] given its state? I propose no.
c. do we use an xPath based scheme or the altenative extensions suggested
above? I think we need to investigate what is being considered in the XML
media type registration and/or what is currently being considered by wsdl
1.2 in terms of the "wsdl component designator" (which may or may not be
really stable).

John, I recommend that c. be considered and investigated.

Luc

[1]     XPointer Framework, W3C Recommendation 25 March 2003,
http://www.w3.org/TR/xptr-framework/
[2]     XPointer element() Scheme, W3C Recommendation 25 March 2003,
http://www.w3.org/TR/xptr-element/
[3]     XPointer xmlns() Scheme, W3C Recommendation 25 March 2003,
http://www.w3.org/TR/xptr-xmlns/

TN Reference:
[5]     XPointer xpointer() Scheme, W3C Working Draft, 10 July 2002.
http://www.w3.org/TR/2002/WD-xptr-xpointer-20020710/