RE: catA014.xsd (test suite)

> -----Original Message-----
> From: Dare Obasanjo [mailto:dareo@microsoft.com] 
> Sent: Thursday, March 14, 2002 7:44 PM
> To: Stanley Guan; Schema XML
> Subject: RE: catA014.xsd (test suite)
> 
> 
> Interesting...
> 
> In the linked email, Henry states 
> 
> 	So in fact all the following are allowed:
> 
> 	final="#all"        (first branch of the union)
> 	final=""            (2nd branch of the union, length 0)
> 	final="extension"   (2nd branch of the union, length 1)
> 	final="restriction" (2nd branch of the union, length 1)
> 	final="extension restriction"   (2nd branch of the union, length
> 2)
> 
> However looking in the normative section in schema for 
> Schema(sForS) in appendix A of XML Schema Structures 
> recommendation and DataTypes recommendation we see the base 
> type "xs:derivationControl" derived from "xs:NMTOKEN" which 
> does not allow "" (see production for nmtoken[0] in xml 1.0 
> shown below) . So how can  final="" be allowed?
> 
> 
> Name 		::= (Letter | '_' | ':') ( NameChar)*
> Names       ::=  Name (#x20 Name)* 
> Nmtoken     ::= (NameChar)+

Jeni kindly pointed out in a private mail that I had overlooked the
(second branch of union, length 0) explanation which makes final=""
valid. 

> Secondly, athough the XML Schema WG has decided to make the 
> sForS a normative reference this does not mean that it should 
> define the behavior of the prose version of the 
> recommendation. I am currently not a 100% sure that the prose 
> version of the recommendation specifies that 
> block="restriction restriction" is a valid attribute for a 
> <complexType> although the XML Representation Summary of the 
> complexType Element Information Item[1] may be implying this with
> 
> 	block = (#all | List of (extension | restriction)) 

As for the above, I admit that it probably isn't the right thing for a
W3C test to state that parsers that allow block="restriction
restriction" are behaving incorrectly. Since both the prose version of
the XML Schema recommendation and the sForS imply that it should be
allowed. I'll confer with our W3C reps as to what should be the next
step. 



-- 
THINGS TO DO IF I BECOME AN EVIL OVERLORD #27
I will never build only one of anything important. All important systems
will
have redundant control panels and power supplies. For the same reason I
will
always carry at least two fully loaded weapons at all times.

Received on Friday, 15 March 2002 13:50:56 UTC