RE: XML schema extensibility : new facet ?

> -----Original Message-----
> From:	Cedric Thienot [SMTP:cthienot@acland.fr]
> Sent:	Thursday, October 05, 2000 2:42 AM
> To:	www-xml-schema-comments@w3.org; xmlschema-dev@w3.org
> Subject:	XML schema extensibility : new facet ?
> 
> Dear all,
> 
> I'm currently involved in the MPEG-7 DDL definition which
> relies on XML Schema. We need to define a new facet for list
> named 'dim' which specifies a multidimensionnal matrices.
> 
> It is desirable that a specific MPEG7 parser uses this new
> facet while keeping a certain level of interoperability with a
> classical XML-schema parser. At the very best, the XML
> Schema could simply ignore the facet.
> 
> I just want to know if the best way is to use <appInfo>:
> 
> <simpleType name="matrix2x2OfInteger">
>     <list base="integer">
>         <annotation><appinfo>
>                 <mpeg7:dim value="2 2"/>
>         </appinfo></annotation>
>     </list>
> </complexType>
> 
I suppose you'd have instances something like:

<myMatrix xsi:type='matrix2x2ofInteger'>1 2 3 4</myMatrix>

where you might have a psuedo-API which gave your MPEG-7 access of the form:
	myMatrix[0][0] = 1
	myMatrix[0][1] = 2
	myMatrix[1][0] = 3
	myMatrix[1][1] = 4

right?

If so, then yes, that is exactly what appinfo is for.  This way your schema
(and instance) documents are still 100% compliant with the spec, but your
special MPEG-7 processors know how to further interpret some of the data.

Hopefully, you will provide a schema for namespace in which the mpeg7:dim
element belongs, which might even help some non-MPEG-7 processors understand
your instance data better (see the hasFacetAndProperty schema [1] used in
the schema for datatypes [2] in the latest WD of the schema spec [3,4,5]).

pvb

References
[1] http://www.w3.org/2000/10/hasFacetAndProperty.xsd
[2] http://www.w3.org/2000/10/datatypes.xsd
[3] http://www.w3.org/TR/2000/WD-xmlschema-0-20000922/
[4] http://www.w3.org/TR/2000/WD-xmlschema-1-20000922/
[5] http://www.w3.org/TR/2000/WD-xmlschema-2-20000922/

Received on Thursday, 5 October 2000 14:49:56 UTC