The example seems like it's actually trying to be two examples at once: a) the tax (abstract) type with two (concrete) subtypes -- PercentTax and FlatRateTax b) some type with 250 subtypes The modeling approach for the two different cases may very well be different. I suggest if we want to talk about both (a) and (b) that we're clear that they aren't necessarily solved by the same approach. For the tax example (a), couldn't I handle this trivial modeling problem with plain old XML schema using one of the Variable Content Container approaches
http://www.xfront.com/VariableContentContainers.pdf In particular Method 4: Implementing variable content containers using a dangling type describes how to create a variable content container containing a simple type. Is there some advantage to the schematron approach over Method 4 for the tax example? For problem (b), can you give an example -- because I don't think tax fits (with 250 subtypes). -Bill >