EclipseLink/Development/ProjectXMLSchemaValidation

Schema Validation

EclipseLink 'native' Project XML is described in the EclipseLink jar in xsd\eclipselink_persistence_map_1.0.xsd, along with the previous TopLink versions in xsd\toplink-object-persistence_1x_y_z.xsd.

When a Project XML file is read in, EclipseLink attempts to validate it against the above schemas. However, schema validation in Java has mostly been a 'hit-&-miss' affair, with different XML parsers providing very different results. Of specific interest is the use of org.xml.sax.EntityResolver's to manage the lookup of xsds. The callback is:

public InputSource resolveEntity(String publicId, String systemId) throws SAXException, IOException

For example, the Oracle V2 XML Parser populates the systemId argument with a relative-path to the xsd directory in the eclipselink.jar file; however, the Xerces parser provided with Sun's JDK 1.5 populates it with a fully-resolved path. In addition, if the schema-to-be-resolved imports another schema (toplink-object-persistence_1x_y_z.xsd imports object-persistence_1_0.xsd), then this second-level of resolving does not work with either parser.

Requirements for Schema Validation

There does not appear to be a record of a user-driven requirement to perform schema validation on Project XML. Therefore, we should fix the current EclipseLink to not perform validation by default and enter an Enhancement Request (ER) for a new feature. It should be debated upon its merits and scheduled as normal.