RNG vs. XSD : is the use of abstract types and polymorphism a goodor ba

["Costello, Roger L." <costello@mitre.org>]

Hi Folks,

James Clark and MURATA Makoto are the two principles that were involved in designing RELAX NG.

I noticed that they did not incorporate into RELAX NG these features:

- a reusable collection of elements and attributes that is designated "abstract" 

   -- XML Schema has this. It is a complexType with abstract="true"

- polymorphism (i.e., the ability to replace one thing with another provided both things descend from the same type)

   -- XML Schema has this. It is element substitution and type substitution


I wonder why James Clark and MURATA Makoto didn't incorporate abstract types and polymorphism into RELAX NG? 

James Clark and MURATA Makoto are both really, really smart guys. 

I am sure they had good reasons for not incorporating these things.

I can think of two possible reasons:

(1) They determined that abstract types and polymorphism are not appropriate for a schema language for XML.

(2) They determined that incorporating those features would have made RELAX NG too complicated.

If the reason is (1) then I would like to know why abstract types and polymorphism are not appropriate for a schema language for XML? Are there dangers in designing schemas for XML that use abstract types and polymorphism? Does James Clark and MURATA Makoto recommend, when using XML Schemas, avoiding its abstract types and polymorphism?

/Roger