Instead of inventing new attributes or syntaxes, why not just say that the subschema for a naming context can be discovered by reading the subschemaSubentry attribute in the naming context?