Dare comments on my post re: RNG vs. XSD

Dare replied to my post on RNG...

He observes is that people use a schema langauge for different things: type-annotating an XML doc to facilitate OO and SQL mappings, validation, and documentation of a format. His point is that XSD is better for the first task and RNG is better for the second task. I certainly agree that RNG is better than XSD at validation. I don't know which is better for type mapping. There are loads of legal XSD things I can do that don't map at all well to the modern OO world, including, but not necessarily limited to: redefine, derivation of complexTypes (or even simple types) by restriction, element substitution groups (which pretty much require GEDs to be OO types too), the fact that I can apply both element and type substitution to the same element at the same time, final for stopping particular kinds of type construction, block for stopping runtime substitution of types and elements, and the fact that final and block options aren't inherited by subtypes. We work around that by not doing those things. There are, presumably, things that I can do with RNG that may be hard to map to other type systems, but as with XSD, I could avoid those things.