RE: inconsistent use of "derived"

Berthold,

Thank you for your note.      

 

Speaking personally, I now feel that use of the word "derived" for the
creation of list and union types was

unfortunate.  Several people have complained that this usage is
misleading and seems to imply that,

for example, a list of decimal is a subtype of decimal, which is not the
case.  "Generated" or "constructed" would be better words to describe
the process of creating list and union datatypes.

All the best, Ashok 
=========================================================== 
Ashok Malhotra              <mailto: ashokma@microsoft.com
<mailto:%20ashokma@microsoft.com> > 
Microsoft Corporation 
212 Hessian Hills Road 
Croton-On-Hudson, NY 10520 USA 
Redmond: 425-703-9462                New York: 914-271-6477 

 

-----Original Message-----
From: Berthold Daum [mailto:berthold.daum@bdaum.de] 
Sent: Wednesday, July 10, 2002 3:52 PM
To: W3C XML Schema Comments list
Subject: inconsistent use of "derived"

 

XML Schema Part 1 (Structure) and XML Schema Part 2 (Datatypes) seem to
have different notions of "derived" for simple types.
 
According to Part1, setion 3.14.6, Schema Component Constraint: Type
Derivation OK (Simple), type unions and list extensions are NOT
"derived" from their respective member types (but their member types are
regarded as "derived" from the union type resp. list extension).
 
This is in contrast to Part 2, which defines union types and list
extensions as "derived" from their respective member types (2.5.2.2 and
2.4.2.3). 
 
The inconsistent semantics of "derived" can lead to confusion among
schema authors, in particular when working with substituion groups,
instance type overriding, and redefinitions.
 
We suggest to drop the term "derived" for type unions and list
extensions in XML Schema Part 2 and to replace it with the term
"constructed". This would also affect the classification of the built-in
types NMTOKENS, IDREFS, and ENTITIES, which are no longer "derived by
list" but "constructed by list".
 
Berthold Daum
bdaum industrial communications
email: berthold.daum@bdaum.de
web: www.bdaum.de