RE: [wsn] AI from telecon 10/11

[Steve Graham <sggraham@us.ibm.com>]

Susan:
It is my understanding that there are
intermediaries in use that in fact do "modify" the namespace
declarations (often for efficiency).  Elements in scope that are typed
as QNames get fixed up (I think), but QNames contained within strings don't.
 That is what I believe is the heart of the matter.

Thanks indeed for cross posting, a very
good idea. 

++++++++
Steve Graham
(919)254-0615 (T/L 444)
STSM, On Demand Architecture
Member, IBM Academy of Technology
<Soli Deo Gloria/>
++++++++

Susan Malaika/Santa Teresa/IBM

10/18/2004 01:24 AM

To	"Patil, Sanjay" <sanjay.patil@sap.com>
cc	Steve Graham/Raleigh/IBM@IBMUS, wsn@lists.oasis-open.org
Subject	RE: [wsn] AI from telecon 10/11Link

I find the description slightly confusing. Do processors
sometimes modify the prefixes specified in namespace declarations and sometimes
modify the prefixes used in expressions? 
In any case, for interest - I am cross posting the
entry I just placed on the WSRF mailing list describing the approach the
XQuery group are taking to try to isolate the scope of namespaces through
the use of prologs. 

Susan Malaika
.................................................................................................................................................................................................................................................................................................

To:        wsrf@lists.oasis-open.org
cc:        
Subject:        [wsrf] Issue WSRF24: Should the XPath evaluations in WSRF specs be made in the context of some namespaces defined or not





I'd like to draw your attention to the approach the W3C XQuery group took in  http://www.w3.org/TR/xquery/
Note that XQuery uses XPath.

XQuery defines a host environment- What's outside the line in Figure 1 here http://www.w3.org/TR/xquery/#id-processing-model

In general, XQuery takes what the host environment gives it, adds or overrides with prolog directives, and then evaluates the XQuery expression.

Each prolog directive is different. For variables, adding a new binding will occlude a binding from the environment. (occlude : the inner scope hides the outer scope for these variables). See the tables here
http://www.w3.org/TR/xquery/#id-xq-context-components

The WSRF TC may want to contact the XQuery co-chairs before (or perhaps instead of) writing to the W3C as a whole. I've already spoken to Andrew Eisenberg (one of the XQuery co-chairs) about issue WSRF24 - so the XQuery group won't be surprised to hear from the WSRF TC.

Susan Malaika







................................................................................................................................................................................................................................................................................................
 

To:        Steve Graham/Raleigh/IBM@IBMUS, wsn@lists.oasis-open.org
cc:        
Subject:        RE: [wsn] AI from telecon 10/11




 
Following is my proposed text for this new issue. Please  review.
 
Thanks,
Sanjay
 

WSN2.31:  Namespace prefix change may result into unbound QNames

Some of the message components  (such as TopicExpression and Selector) may include QName strings in their value.   The namespace prefixes used in the  QNames are normally declared by some in-scope parent element.

It is possible that toolings  modify the namespace declarations and change the prefix values without  necessarily updating the QName strings appearing in the message content,  resulting into unbound QNames.

Specifications

• WS-BaseNotification

Proposed  Recommendations

Steve made the following proposal  via email -  http://www.oasis-open.org/archives/wsn/200410/msg00021.html

For each message exchange that  describes a component
that may include a QName embedded in a string, the  following normative
text be included in the description of that component:  

If the value of the component contains QNames, then the
namespace  declarations that specify the mapping of the QName's prefix
to an actual  namespace URI can be found amongst any namespace declaration
in scope for the  component.  Note: Some XML processors MAY modify
the namespace  declarations. Designers should be aware that such transforms
exist and may  render the expression incoherent, as it is likely the
change in namespace  declaration will not update a QName embedded
within a string.

Notes

http://www.oasis-open.org/archives/wsn/200409/msg00031.html

Status:   Open

Date:  Oct  11, 2004

Contact:  

Igor  Sedukhin

Cross  Reference: 

WSN2.12  raises a related issue.
-----Original Message-----
From: Steve Graham  [mailto:sggraham@us.ibm.com]

Sent: Monday, Oct 11, 2004 10:37  AM
To: wsn@lists.oasis-open.org
Subject: [wsn] AI from  telecon 10/11



I took an  AI to suggest resolution text to the proposed
problem: 
>Problem with the use of namespace prefixes in
topic and  selector expressions. Client tooling may modify the the
>prefixes  used
>in a web service request, so prefixes used in
these expressions  may become unbound. Igor has observed a similar
>problem in  WS-E[6] 

I propose the  resolution be as follows:

For each  message exchange that describes a component
that may include a QName embedded  in a string, the following normative
text be included in the description of  that component:


If the value of  the component contains QNames, then
the namespace declarations that specify  the mapping of the QName's
prefix to an actual namespace URI can be found  amongst any namespace
declaration in scope for the component.  Note: Some  XML processors
MAY modify the namespace declarations. Designers should be  aware
that such transforms exist and may render the expression incoherent, as
 it is likely the change in namespace declaration will not update
a QName  embedded within a string. 

sgg
++++++++
Steve Graham
(919)254-0615 (T/L 444)
STSM, On  Demand Architecture
Member, IBM Academy of Technology
<Soli Deo  Gloria/>
++++++++