RE: XML result of Schema validation

Thanks again for the responses from both of you! I was really looking to
find whether or not an enforced relationship exists between the PSVI and
the XPath data model for XSLT 1.0. I appreciate the help.

I had a related question: Is there a recommendation for an XML
representation of the PSVI? 

-Jack

-----Original Message-----
From: noah_mendelsohn@us.ibm.com [mailto:noah_mendelsohn@us.ibm.com] 
Sent: Monday, April 24, 2006 11:06 AM
To: Michael Kay
Cc: Matheson, Jack K; xmlschema-dev@w3.org
Subject: RE: XML result of Schema validation

Let me add a little to Mike's response, which is correct as far as it 
goes.    The way I look at it is this:  XML Schema Part 1 gives you a 
taxonomy of the information that can be discovered by validating an 
instance against a schema.  That information is called the Post Schema 
Validation Infoset (PSVI.  XSLT transforms what I'll loosely call an 
instance using a stylesheet.  In the case of XSLT, the instance is
modeled 
by the XPath data model abstraction. 

As Mike says, there is no normative dependence of XSLT 1.0 on XML
Schema, 
and the obvious way to build the XPath data model for XSLT 1.0 is to run
a 
simple (schema unaware) parser over your XML.  Then again, there is 
nothing preventing someone from writing a piece of software that builds 
the XPath data model for XSLT in a different way, and in particular by 
using the whitespace normalized information from a PSVI.  There is no 
Recommendation or other standard covering such a connection at this
time, 
and I'm not aware of any widely deployed software that does this for
you, 
but you wouldn't be violating any Recommendations if you built such 
software.  You'd just be making a connection that was a bit less than 
obvious and not covered (or probilited) by the standards, as far as I 
know.

As Mike also says or implies, XSLT 2.0 is schema-aware, and you can read

the proposed mappings [1] from PSVI to the XPath data model to see how 
schema will normatively affect XSLT processing in version 2.0.  Even 
there, you could use your own nonstandard mappings, but you or someone 
would have to write the necessary special purpose mapping software. 

Noah

[1] http://www.w3.org/TR/xpath-datamodel/#const-psvi

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------








"Michael Kay" <mike@saxonica.com>
Sent by: xmlschema-dev-request@w3.org
04/21/2006 11:13 AM
 
        To:     "'Matheson, Jack K'" <jack.k.matheson@intel.com>, 
<xmlschema-dev@w3.org>
        cc:     (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        RE: XML result of Schema validation


XSLT 1.0 is not schema-aware. It works only off the original InfoSet. 
Schema-validation should have no impact whatsoever.
 
Michael Kay
http://www.saxonica.com/

From: xmlschema-dev-request@w3.org [mailto:xmlschema-dev-request@w3.org]

On Behalf Of Matheson, Jack K
Sent: 21 April 2006 16:03
To: xmlschema-dev@w3.org
Subject: RE: XML result of Schema validation

I was trying to find my answer in XML Schema Part 1: Structures (and
also 
the primer).
When you say ?it will normally see the values before whitespace 
normalization?, would this apply mostly to XSLT 1.0 processors? Also,
does 
?normally? mean that behavior is slightly undefined?
 
Is it necessary to construct the XDM from the PSVI before acting as
input 
to an XSLT 1.0 processor?
 
Thanks again!
-Jack
 

From: Michael Kay [mailto:mike@saxonica.com] 
Sent: Friday, April 21, 2006 4:44 AM
To: Matheson, Jack K; xmlschema-dev@w3.org
Subject: RE: XML result of Schema validation
 
Which spec are you reading? For the effect on an XSLT transformation or
on 
XQuery, you need to understand
 
(a) the PSVI generated by the schema processor
 
(b) the mapping from the PSVI to the XDM: see 
http://www.w3.org/TR/xpath-datamodel/#const-psvi
 
Regarding your specific questions:
 
- if the stylesheet/query looks at text nodes or string values, it will 
normally see the values before whitespace normalization. If it looks at 
typed values it will normally see the values after such normalization. 
Processors are allowed to reconstruct the string value (and the text 
nodes) from the typed value, but I think this is only likely to happen
in 
practice when you store the data in an XML-enabled relational database.
 
- yes, default values are added to the XDM document.
 
Michael Kay
http://www.saxonica.com/
 

From: xmlschema-dev-request@w3.org [mailto:xmlschema-dev-request@w3.org]

On Behalf Of Matheson, Jack K
Sent: 20 April 2006 16:05
To: xmlschema-dev@w3.org
Subject: XML result of Schema validation
I?m having trouble finding information on the XML result of a Schema 
validation; in particular, what happens when the result of a validation 
acts as the input to a transformation. Does a stylesheet see 
whitespace-normalized versions of text values from the input document? 
Also, are default values for elements and attributes inserted for the 
stylesheet to see?
 
It?s certainly possible that I?m simply reading the spec incorrectly, so

any help is appreciated!
 
Thanks,
-Jack

Received on Monday, 24 April 2006 16:41:45 UTC