Cart
[Home] [By Thread] [By Date] [Recent Entries]
Thanks Dare. I appreciate that URL. Where we get into difficulties is when that set of types isn't adequate given a particular object model. The X3D XML Schema covers the XSD types, but it wasn't a great fit. The topic now becomes whether or not it is better for X3D systems to have their own datatypes or to mix and match. I suspect that as you say, it will depend on what they must interop with. But interoperation is in the object model, not the means of denoting the library. So putting validation up front seems to be putting the cart in front of the horse. One must first know what datatypes the object model specifies, then ensure the validation language includes these. The case for RNG is stronger. What happens to an RNG implementation if two datatype libraries are in conflict? Can that happen? One expects that not to happen to a given object model (say X3D) by definition, but intuitively, I can think of that happening given an aggregate or where two models must interoperate (say X3D inside or wrapping HTML). Must be a pain for the framework (say operating system). len From: Dare Obasanjo [mailto:dareo@m...] The ability to specify which datatype libraries are used by the validation language means that if you want to interop with other parties you have to use a standardized set of datatypes. The only [recognized] standardized set of datatypes we have in the XML world has been W3C XML Schema: Part 2. This is what most RELAX NG implementations support. PS: If you need an example, read http://www.oasis-open.org/committees/relax-ng/tutorial-20011203.html#IDA5UYR
|
