> One of the the things that I'd like to have is the ability to find out > what pattern(s) matched a particular element/attribute and, in the > case of simple types, which simple type matched.
Hmm. Sounds a bit like a PSVI. Would it be sufficient if this works for a RELAX NG schema only if it is sufficiently deterministic that for a document valid against the schema you can determine which pattern an element or attribute matches just by looking at the names of the element or attribute and its ancestors? I think I would want to do this sort of thing with a separate infoset-augmentation process that is independent of RELAX NG validation even though it's controlled by the same schema. How would you want to get the information about which pattern matched an element/attribute? One possibility is to try and do something that fits into an XSD-ish model, where you get a QName representing the "type" of an element or attribute. The JAXP validation API is set up to support this, and I expect other XML technologies are as well. Maybe you could have a (configurable) annotation attribute on element/attribute/data patterns whose value gets reported to the application as the "type". I'm also curious what you want this for. Also, as between elements/attributes on the one hand and data on the other, for which are you more interested in getting the matching pattern? James
