schema validation and elementFormDefault

[Patrick Dowler]

I am having trouble understanding the effect/correctness of elementFormDefault
being not set (or set to unqualified) in a schema I am using. In the current
draft of the schema, elementFormDefault is not set.

My test program build the document (SAXBuilder, xerces SAXParser) with and
without schema validation. Below are 3 tests I did with different documents.

In test #1 and #2 JDOM puts both the root and child element in the same
expected namespace, but SAX validation fails. With #3, JDOM puts the root and
child elements in different namespaces but SAX validation succeeds. #3 is
nominally equivalent to the examples for using elementFormDefault="unqualified"
in the W3C XML schema spec. So, it looks to me like JDOM and SAX (the schema)
disagree on the namespace of the child element.

If I change the schema to have elementFormDefault="qualified", test #1 and #2
pass and test #3 fails and everything seems consistent. So my problem seems to
stem from the behaviour of elementFormDefault="unqualified".

Can anyone shed any light on this?

What is up with elementFormDefault="unqualified"? It seems to change the rules
for determining which namespace an element belongs to... Is it fundamentally
broken?

thanks in advance,

Patrick

PS-I should note that the schema in question is just passing through a review
stage toward standardisation, so I can argue for elementFormDefault="qualified"
instead.


*** Document #1: terminology: is this default namespace usage?

<?xml version="1.0"?>
<VOTABLE version="1.2"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns="http://www.ivoa.net/xml/VOTable/v1.2">

<RESOURCE/>

</VOTABLE>

Test output #1:

building with validate=false...
[Element: <VOTABLE [Namespace: http://www.ivoa.net/xml/VOTable/v1.2]/>]
  is in [Namespace: prefix "" is mapped to URI  
  "http://www.ivoa.net/xml/VOTable/v1.2"]
[Element: <RESOURCE [Namespace: http://www.ivoa.net/xml/VOTable/v1.2]/>]
  is in [Namespace: prefix "" is mapped to URI
  "http://www.ivoa.net/xml/VOTable/v1.2"]
building with validate=false... [OK]


building with validate=true...
FAILED:
Error on line 6 of document file:
/home/cadc/pdowler/work/src/scratch/test/votable1.xml:
 cvc-complex-type.2.4.a:
 Invalid content was found starting with element 'RESOURCE'. One of
'{DESCRIPTION, DEFINITIONS, COOSYS, GROUP, PARAM, INFO, RESOURCE}' is
expected.

*** Test Document #2: explicit namespace prefixes

<?xml version="1.0"?>
<vot:VOTABLE version="1.2"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:vot="http://www.ivoa.net/xml/VOTable/v1.2">

<vot:RESOURCE/>

</vot:VOTABLE>

Test output #2:

building with validate=false...
[Element: <vot:VOTABLE [Namespace: http://www.ivoa.net/xml/VOTable/v1.2]/>]
  is in [Namespace: prefix "vot" is mapped to URI
"http://www.ivoa.net/xml/VOTable/v1.2"]
[Element: <vot:RESOURCE [Namespace: http://www.ivoa.net/xml/VOTable/v1.2]/>]
  is in [Namespace: prefix "vot" is mapped to URI
"http://www.ivoa.net/xml/VOTable/v1.2"]
building with validate=false... [OK]


building with validate=true...
FAILED:
Error on line 6 of document file:
/home/cadc/pdowler/work/src/scratch/test/votable2.xml:
 cvc-complex-type.2.4.a:
 Invalid content was found starting with element 'vot:RESOURCE'. One of
'{DESCRIPTION, DEFINITIONS, COOSYS, GROUP, PARAM, INFO, RESOURCE}' is
expected.:


*** Test Document #3: ugly hack?

<?xml version="1.0"?>
<vot:VOTABLE version="1.2"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:vot="http://www.ivoa.net/xml/VOTable/v1.2">

<RESOURCE/>

</vot:VOTABLE>

Test result #3:

building with validate=false...
[Element: <vot:VOTABLE [Namespace: http://www.ivoa.net/xml/VOTable/v1.2]/>]
  is in [Namespace: prefix "vot" is mapped to URI
"http://www.ivoa.net/xml/VOTable/v1.2"]
[Element: <RESOURCE/>]
  is in [Namespace: prefix "" is mapped to URI ""]
building with validate=false... [OK]


building with validate=true...
[Element: <vot:VOTABLE [Namespace: http://www.ivoa.net/xml/VOTable/v1.2]/>]
  is in [Namespace: prefix "vot" is mapped to URI
"http://www.ivoa.net/xml/VOTable/v1.2"]
[Element: <RESOURCE/>]
  is in [Namespace: prefix "" is mapped to URI ""]
building with validate=true... [OK]


--

Patrick Dowler
Tel/Tél: [phone]
Canadian Astronomy Data Centre
National Research Council Canada
5071 West Saanich Road
Victoria, BC V9E 2M7

Centre canadien de donnees astronomiques
Conseil national de recherches Canada
5071, chemin West Saanich
Victoria (C.-B.) V9E 2M7

_______________________________________________
To control your jdom-interest membership:
http://www.jdom.org/mailman/options/jdom-interest/youraddr@...

[Michael Kay]

> What is up with elementFormDefault="unqualified"? It seems to
> change the rules for determining which namespace an element
> belongs to... Is it fundamentally broken?

It's not broken, in that it's perfectly well defined, but it it produces
highly counter-intuitive results and there is never a good reason for using
it.

Regards,

Michael Kay
http://www.saxonica.com/
http://twitter.com/michaelhkay 

_______________________________________________
To control your jdom-interest membership:
http://www.jdom.org/mailman/options/jdom-interest/youraddr@...

[Patrick Dowler]

On Tuesday 20 October 2009 14:53:44 you wrote:
> > What is up with elementFormDefault="unqualified"? It seems to
> > change the rules for determining which namespace an element
> > belongs to... Is it fundamentally broken?
>
> It's not broken, in that it's perfectly well defined, but it it produces
> highly counter-intuitive results and there is never a good reason for using
> it.

When I parse valid document (valid according to a schema using
elementFormDefault="unqualified") JDOM puts the root element and the child in
different namespaces, eg:

<?xml version="1.0"?>
<vot:VOTABLE version="1.2"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:vot="http://www.ivoa.net/xml/VOTable/v1.2">

<RESOURCE/>

</vot:VOTABLE>

JDOM puts VOTABLE in the "http://www.ivoa.net/xml/VOTable/v1.2" namepsace and
it puts RESOURCE in the "" namespace.

SAX (xerces) schema validation says the document is valid, so it must think
RESOURCE is in the "http://www.ivoa.net/xml/VOTable/v1.2" namespace; that is
the intent of the schema authors.

Is JDOM wrong in the sense of not being capable of correctly assigning
namespaces in this case because it does not check this setting in the schema?

I'm not saying it is a bug... Is it a known (insurmountable?) limitation of
JDOM?

thanks in advance,

--

Patrick Dowler
Tel/Tél: [phone]
Canadian Astronomy Data Centre
National Research Council Canada
5071 West Saanich Road
Victoria, BC V9E 2M7

Centre canadien de donnees astronomiques
Conseil national de recherches Canada
5071, chemin West Saanich
Victoria (C.-B.) V9E 2M7

_______________________________________________
To control your jdom-interest membership:
http://www.jdom.org/mailman/options/jdom-interest/youraddr@...

[Michael Kay]

> When I parse valid document (valid according to a schema using
> elementFormDefault="unqualified") JDOM puts the root element
> and the child in different namespaces, eg:
>
> <?xml version="1.0"?>
> <vot:VOTABLE version="1.2"
>     xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
>     xmlns:vot="http://www.ivoa.net/xml/VOTable/v1.2">
>
> <RESOURCE/>
>
> </vot:VOTABLE>
>
> JDOM puts VOTABLE in the
> "http://www.ivoa.net/xml/VOTable/v1.2" namepsace and it puts
> RESOURCE in the "" namespace.

Quite correct. That's what the schema says it should do.
>
> SAX (xerces) schema validation says the document is valid, so
> it must think RESOURCE is in the
> "http://www.ivoa.net/xml/VOTable/v1.2" namespace; that is the
> intent of the schema authors.

By using elementFormDefault="unqualified" the schema authors were signifying
that locally-declared elements are in no namespace. If they intended them to
be in the target namespace, they should have used
elementFormDefault="qualified".
>
> Is JDOM wrong in the sense of not being capable of correctly
> assigning namespaces in this case because it does not check
> this setting in the schema?

No, JDOM is correct. As I said before, elementFormDefault="unqualified"
gives very counter-intuitive results, and for that reason is rarely used
(except by beginners who don't know better), but if it is used, this is what
it means and JDOM is handling it correctly.

Regards,

Michael Kay
http://www.saxonica.com/
http://twitter.com/michaelhkay 

_______________________________________________
To control your jdom-interest membership:
http://www.jdom.org/mailman/options/jdom-interest/youraddr@...