This message is mostly an FYI; I wanted to report on XML schema validation tools and how they do and don't work with the proposed XML schema suite.
Those working with the XML schema suite are probably aware that the definition of the SimpleLiteral type in the dc schema, and the definition of restrictions of that type in dcterms do not pass muster with some of the schema validation tools.
My personal experience is this:
1. XSV *does* validate.
version="XSV 1.203.2.47.2.4.2.14/1.106.2.25.2.6 of 2002/06/15 18:59:35"
2. topologi *does* validate. (build 2001-10-25)
3. Xerces does *NOT* validate (version 2.0.1)
4. XMLSpy does *NOT* validate (version 4.4 and version 5).
The problem has to do with the validators interpretation of restriction of a mixed content type. According to _Definitive XML Schema_ by Priscilla Walmsley (Prentice Hall PTR 2002), the schemas are perfectly legal. (See Chapter 14, section 14.5.3, p 328):
"It is also possible to restrict a mixed content type to dervied ... a simple content complex type. ... Example 14-25 shows ... the only case where a restriction element may have both a base attribute and a simpletype child."
The example is very similar to the proposed qualified dc schema suite, but the two validators don't work. In fact, when I made the dc schema suite look MORE like what was in the book, XSV stopped working as well. Hmmmm.
Because of our needs at the NSDL, we have created a version of the schema suite that does NOT have a mixed content model for SimpleLiteral -- this will allow us greater flexibility in XML tool choices. All four tools above work with our schemas, but there's an oddity with XMLSpy: it will validate the document, but it will also open the schemas and complain about them. However, if you validate the schemas separately (as schemas) it says they're fine.
Ah yes, the bleeding edge.
- Naomi
|