XML Schema Requirements

[Sorry -- I hit the Send button too early.  Let's try this again.]

Overall, I think the schema requirements are well-written and clear.  I 
particularly appreciate the level of detail that notes such things as 
defining the relationship between schema validity and XML validity.

Specific comments:

1) Please clarify the relationship between the capabilities of schemas and 
DTDs -- superset, subset, skew, not yet defined?  Design principle 1 states 
that schemas shall be "more expressive than XML DTDs."  Does this refer to 
readability or capability?  Conformance principle 4 states that the 
language must "define the relationship between schemas and XML DTDs."  Does 
this mean capability differences will be defined in the future or that the 
spec will define such things as how to handle a document with both a schema 
and a DTD?

2) Perhaps I am reading too much into the word "inheritance", but I would 
like to see a requirement for a reuse mechanism that is much simpler than 
inheritance.  That is, I would like to simply point at an element or 
attribute in another schema (as I can in DCD) and say, "I want to use 
that."  I believe this will be the most common reuse case.  (A more generic 
mechanism would allow me to point at anything, such as an enumerated type 
or sequence definition.)

3) At the risk of sounding like a broken record, I would like to see entity 
definitions removed from the schema language.  These are physical, not 
logical, definitions and have nothing to do with schemas.  If there is a 
need to define/process entities with an XML language, this should be done 
in a separate language.

4) How is datatype requirement 2 (define a type system that is adequate for 
import/export from database systems) different from primitive data typing 
(requirement 1) and user-defined datatypes (requirement 4)? Combined with 
use case 5 (a database can emit a schema the defines useful and legitimate 
queries) I am afraid that this implies such things as key relationships.  I 
believe these to be beyond the capabilities of a first-release schema 
language.

-- Ron Bourret

Received on Friday, 26 February 1999 05:03:02 UTC