types and instances in DTDs and XML Schemas

[Alexander Nakhimovsky <sasha@cs.colgate.edu>]

In XML 1.0, the element's content model is its type.  Instances of the type 
are found in instance documents.  If one type contains another type, the 
containment relationship is visible in the instance document.

XML Schema "types" are used within the schema to define element content 
models.  They are twice removed from instance documents.  They are, in 
effect, tree-structured parameter entities, with relations such as 
"extention-of" or "restriction-of" defined on them.   These relations are 
completely invisible in instance documents.  Any computer program that 
relies on those relations either must have access to the XML schema 
document or work with (as yet unspecified) PSVI, which has to be produced 
by the validating parser.  There may well be useful programs that rely on 
Schema type relations, but it seems unreasonable to burden every validating 
parser with the task of understanding those relations and translating them 
into a "heavily-decorated DOM" data structure.

So, it would be a good thing, IMO, if XML Schema were *re-factored* into 
the validating part (such as RELAX NG) and a "complex-type-relations" part, 
for use in specialized applications.

Alexander Nakhimovsky
http://cs.colgate.edu/faculty/sasha/adn.htm