Re: Key/IDREF checking as a pattern fragment

Why not using

<element doThis="DEF1" doThat="DEF2" doSomethingElse="DEF3"/>

?

Your intention is not feasable.

--Stefan

> 
> Hi there,
> 
> I have an element which, in its content, I wish to be able to construct a
> function sequence which references other elements.  I am currently using
> the
> following construct:
> 
> <xsd:element name="def">
>  <xsd:complexType>
>   <xsd:attribute name="id" type="xsd:ID"/>
>  </xsd:complexType>
> </xsd:element>
> 
> <xsd:element name="implement">
>  <xsd:complexType>
>   <xsd:choice minOccurs="0" maxOccurs="unbounded">
>    <xsd:element name="doThis">
>     <xsd:complexType>
>      <xsd:attribute name="use" type="xsd:IDREF"/>
>     </xsd:complexType>
>    </xsd:element>
>    <xsd:element name="doThat">
>     <xsd:complexType>
>      <xsd:attribute name="use" type="xsd:IDREF"/>
>     </xsd:complexType>
>    </xsd:element>
>    <xsd:element name="doSomethingElseEntirely">
>     <xsd:complexType>
>      <xsd:attribute name="use" type="xsd:IDREF"/>
>     </xsd:complexType>
>    </xsd:element>
>   </xsd:choice>
>  </xsd:complexType>
> </xsd:element>
> 
> Which allows me the following example content:
> 
> <def id="DEF1"/>
> <def id="DEF2"/>
> <def id="DEF3"/>
> <implement>
>  <doThis use="DEF1"/>
>  <doThat use="DEF2"/>
>  <doSomethingElseEntirely use="DEF3"/>
> </implement>
> 
> However, for the sake of brevity in my instance XML, I would like to be
> able
> to concatenate the do* elements and represent them as an attribute in the
> <implement> element, something like the following:
> 
> <implement do="doThis(DEF1),doThat(DEF2),doSomethingElseEntirely(DEF3)"/>
> 
> I can write a xsd:restriction pattern for the attribute simple type as
> follows:
> 
> <xsd:pattern
>
value="(doThis|doThat|doSomethingElseEntirely)\(\c+\)(,(doThis|doThat|doSome
> thingElseEntirely)\(\c+\))*"/>
> 
> which (I hope) enforces the function names, the parens, and a requirement
> that the content of the parens by characters permitted by xsd:NMTOKEN
> only,
> but (and I bet you saw this coming) is there any way that I can enforce
> the
> parens contents to be IDREFs?  At the moment I am contemplating either
> leaving it with my clunky element sequence, or using the above pattern and
> providing additional Schematron (or sim.) validation for the key
> validation.
> 
> Any suggestions would be gratefully received,
> 
> Stuart
>

Received on Friday, 29 November 2002 10:16:04 UTC