RE: Its time to publish 2nd draft

Kirk,

Here's my take for the seemingly inconsistency.

For SML validators, as soon as they know an element E is a null reference 
(or E is not a reference at all), it can (and MUST) decide not to take 
further reference-related actions on E.

But deref() is a separate entity. It's defined as an extension XPath 
function, and may not be used *only* by SML validators. As a function, it 
MUST define its behavior for all circumstances, including when presented 
with a null reference, hence those bullets in 4.1.2.4.

Thanks,
Sandy Gao
XML Technologies, IBM Canada
Editor, W3C XML Schema WG
Member, W3C SML WG
(1-905) 413-3255 T/L 969-3255
 



"Wilson, Kirk D" <Kirk.Wilson@ca.com> 
Sent by: public-sml-request@w3.org
2007-09-18 07:00 AM

To
Valentina Popescu/Toronto/IBM@IBMCA
cc
"Pratul Dublish" <Pratul.Dublish@microsoft.com>, <public-sml@w3.org>, 
<public-sml-request@w3.org>
Subject
RE: Its time to publish 2nd draft






Valentina,
 
Thanks  for taking care of these points.
 
Regarding the point for which you could not find the reference: in section 
4.1.1.2:
 
If a reference element is recognized as null, then processors MUST NOT 
attempt to resolve it.
 
And in 4.1.2.3: A model validator MUST not attempt to resolve null 
references.
 
But section 4.1.2.4 (defining deref() ) says:
 
The output node set contains no element node corresponding to R if any of 
the following conditions is true
. . .
R is a null reference
Which seems to define an output (the same output for not finding a 
reference in a model) for dereferencing a null reference.
 
Kirk Wilson, Ph.D.
Research Staff Member
CA Labs
603 823-7146
 

From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On 
Behalf Of Valentina Popescu
Sent: Tuesday, September 18, 2007 12:07 AM
To: Wilson, Kirk D
Cc: Pratul Dublish; public-sml@w3.org; public-sml-request@w3.org
Subject: RE: Its time to publish 2nd draft
 

I fixed the following items : 

>> In the Introduction: a period is needed at the end of the bold sentence 
in item #2. 
>> I would recommend that the new example of using sml:nilref=?true? in 
section 4.1 include the <Name> and <Grade> elements from the example above 
to more clearly make the distinction between xsi:nil=?true? and 
sml:nilref=?true?  Perhaps some verbiage following the example would be 
helpful to further explain null references and contrast that with the 
sml:ref=?false?. 
>> Section 4.2.1 states, ?if a model validator chooses to represent 
references with the URI scheme?.  Since when does the validator ?choose to 
represent references??  Isn?t that the role of the document producer? 
<vp>replaced with " if a model author chooses to ...''</vp> 
>>In SML-IF, the following statement is made in section 2.2: 
 
Did not find the reference to : 
>> Why do we say that a validator MUST not try to deref a null reference 
when in fact we define an output (0 nodes) for a null reference? 

Left unchanged : 
>> I would suggest changing all the names ending in ??RefType? in the 
example in section 4.4.1 to simply ??Reference?. 
<vp>All types in this sample are called  somethingType, for that reason I 
guess we should keep the 'Type' suffix for the reference types as 
well</vp> 

The rest are non-editorial issues 


Thank you,
Valentina Popescu
IBM Toronto Labs
Phone:  (905)413-2412         (tie-line  969)
Fax: (905) 413-4850



"Wilson, Kirk D" <Kirk.Wilson@ca.com> 
Sent by: public-sml-request@w3.org 
09/17/2007 08:11 PM 


To
"Pratul Dublish" <Pratul.Dublish@microsoft.com>, <public-sml@w3.org> 
cc
 
Subject
RE: Its time to publish 2nd draft
 


 
 




Here are some comments?that I was able to complete on the plane.  (Again, 
regrets regarding this week?s meeting; I will not be able to make it. 
  
The Abstract still mentions XML Schema ?profile.?  (Also don?t we have to 
also clean up using ?profile? with regard to Schematron?) 
  
In the Introduction: a period is needed at the end of the bold sentence in 
item #2. 
  
IMHO, the definition of Rules in section 5 should restrict XML Schema to 
just version 1.0, since 1.1 now contains a simple constraint language. 
  
I would recommend that the new example of using sml:nilref=?true? in 
section 4.1 include the <Name> and <Grade> elements from the example above 
to more clearly make the distinction between xsi:nil=?true? and 
sml:nilref=?true?  Perhaps some verbiage following the example would be 
helpful to further explain null references and contrast that with the 
sml:ref=?false?. 
  
Two questions: 
1.        Something to think about for the future:  What does ?resolves? 
apply to: a scheme or a reference?  Well, both, which makes the use of 
?reference? float between these two levels.  I would suggest that 
reference elements refer while schemes point.  I think making this 
distinction will alleviate any confusion that may arise from the current 
text. 
2.        Given the discussion we had about validators vs. consumers with 
regard dereferencing behaviors, why MUST a validator provide an 
implementation for deref()?  A validator must apply ?strict? validation to 
references, which is different behavior from deref. 
  
Why do we say that a validator MUST not try to deref a null reference when 
in fact we define an output (0 nodes) for a null reference? 
  
Section 4.2.1 states, ?if a model validator chooses to represent 
references with the URI scheme?.  Since when does the validator ?choose to 
represent references??  Isn?t that the role of the document producer? 
  
I would suggest changing all the names ending in ??RefType? in the example 
in section 4.4.1 to simply ??Reference?. 
  
  
In SML-IF, the following statement is made in section 2.2: 
      See section 6, "Equivalence," and section 5 "Reference Resolution" 
in particular. 
      There currently are no sections 5 and 6 in the document. 
  
  
Kirk Wilson, Ph.D.
Research Staff Member 
CA Labs 
603 823-7146 
  
 


From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On 
Behalf Of Pratul Dublish
Sent: Friday, September 14, 2007 4:38 PM
To: public-sml@w3.org
Subject: Its time to publish 2nd draft 
  
All 
We are close to our target date for publishing the second draft (which is 
9/19). The editorial team is working on updating the spec to reflect the 
changes that were approved in yesterday?s call and expect to have a draft 
ready by Monday, 9/17.  They have already made several changes and these 
are ready for review by the members.  Please review the following changes 
made by the editorial team and send in  any request for changes  at the 
earliest and no later than 5 PM PST on 9/18 
  

4630 
Clarify relation of SML and XML 1.0 vs 1.1 
4638 
Conformance section needed 
4647 
smlerr:output - why exclude node sets including text node... 
4686 
Use schema terminorlogies to describe "xml schema valid" 
4802 
For bullet points 4,5 (...XPath expression must conform t... 
4803 
Edit bullet point (A pattern MUST be evaluated for an ins... 
4884 
update the definition of a null reference to include the ... 

  
  
Thanks! 
Pratul