children element with namespace prefix were considered invalid by xerces

Question

I use xerces to valid a xml instance against schema:

parser.setFeature("http://xml.org/sax/features/namespaces", true);
parser.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
parser.setFeature("http://xml.org/sax/features/validation", true);
parser.setProperty("http://apache.org/xml/properties/schema/external-schemaLocation",
      schemaLocation);
parser.setFeature("http://apache.org/xml/features/validation/schema", true);
parser.parse(new InputSource(xml));

Here is my xml instance:

<?xml version="1.0"?>
<eml:eml packageId="tao.12926.1" system="knb" xmlns:eml="eml://ecoinformatics.org/eml-2.1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="eml://ecoinformatics.org/eml-2.1.0 eml.xsd">
  <dataset>
  .......
  </dataset>
</eml:eml>

This xml is considered valid.

However, if i added prefix "eml" to element "dataset":

<?xml version="1.0"?>
<eml:eml packageId="tao.12926.1" system="knb" xmlns:eml="eml://ecoinformatics.org/eml-2.1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="eml://ecoinformatics.org/eml-2.1.0 eml.xsd">
  <eml:dataset>
  .......
  </eml:dataset>
</eml:eml>

It give me an error: cvc-complex-type.2.4.a: Invalid content starting with element 'eml:dataset'. The content must match '((("":access){0-1},(((("":dataset)|("":citation))|("":software))|("":protocol))),("":additionalMetadata){0-UNBOUNDED})'.

I couldn't understand this. "dataset" has the default the namespace during our schema definition. "dataset" just a abridged version of "eml:dataset". Why xerces doesn't like ?

Would you please give me some clue?

Thanks!

Answer 1

I think any parser, not only xerces, would report an error regarding eml:dataset.
The reason is that the schema for eml:eml doesn't use the attribute elementFormDefault, which then defaults to "unqualified", hence requiring that "locally declared elements" names (such as dataset) do not receive a prefix (or an implied namespace by way of default namespace). Only global elements (such as eml, here), can have (in fact, require) a explicit namespace prefix.

To allow (or maybe to even require) that locally declared elements be prefixed, one would have to alter the schema, either by adding the elementFormDefault="qualified" attribute-value pair to the declaration of the underlying global element, or by adding a form="qualified" attribute to individual locally declared elements of the schema (dataset, for example).

All of this, and then some!, is explained in section 3.1, 3.2 of the W3C Schema Primer document.