W3C home > Mailing lists > Public > xmlschema-dev@w3.org > March 2002

Re: Paper submission due!! please help!! Restriction with substitution???

From: Joey Coyle <joey@xcoyle.com>
Date: Wed, 13 Mar 2002 12:02:59 -0500 (EST)
Message-Id: <>
To: Jeni Tennison <jeni@jenitennison.com>
Cc: xmlschema-dev@w3.org
thanks Jeni,

I finally understand substitution groups.   Even though that works, it 
would be convenient if they added a way to restrict the wildcard method, 
because it is simple and is the way most programmers think about making 
arrays of a Parent class, which allows the array to have any subtype of 
that parent.   And then you would't have to declare global elements.  Maybe 
something like this ...

<complexType name="basetype">
                 <any restriction="myType" maxOccurs="unbounded"/>

This would allow only a sequence of "myType" or restrictions of "myType",

Then this would be valid ...

<complexType name="newtype">
      <restriction base="my:basetype">
          <element name="e1" type="myTypeRestriction1"/>
          <element name="e2" type="myTypeRestriction2"/>
          <element name="e3" type="myTypeRestriction3"/>

thanks again,

At 12:30 PM 3/13/2002 +0000, you wrote:
>Hi Joey,
> > I read in this list that this first example is OK, but is there a
> > way to do this with a type other than "any". Instead of any, I would
> > like my own type, then I would like to substitute derived types as
> > is done here. My example is after the $$$$$$$$$$$$$$$$$$$$$$$$$, and
> > I hope it is valid.
>I'm afraid that what you have definitely isn't valid. When you do a
>restriction, elements that are valid according to the restricted type
>must also be valid according to the base type. In your case, if you
>   <foo>
>     <e1 />
>   </foo>
>which is valid according to your newtype:
> > <complexType name="newtype">
> >   <complexContent>
> >     <restriction base="my:basetype">
> >       <sequence>
> >         <element name="e1" type="myTypeRestriction"/>
> >         <element name="e2" type="myTypeRestriction2" minOccurs="0"/>
> >         <element name="e3" type="myTypeRestriction3" minOccurs="0"/>
> >       </sequence>
> >     </restriction>
> >   </complexContent>
> > </complexType>
>it wouldn't be valid according to your base type, which only allows
>'item' elements within the element:
> > <complexType name="basetype">
> >   <sequence>
> >     <element name="item" type="myType" maxOccurs="unbounded"/>
> >   </sequence>
> > </complexType>
>The reason that you can do the restriction with the xs:any wildcard
>is that the base type then allows any element.
>What you could do, however, is create a substitution group for the
>four elements, with 'item' as the head element. The elements have to
>be declared globally, as follows:
><element name="item" type="myType" />
><element name="e1" type="myTypeRestriction" substitutionGroup="item" />
><element name="e2" type="myTypeRestriction2" substitutionGroup="item" />
><element name="e3" type="myTypeRestriction3" substitutionGroup="item" />
>[Note that myTypeRestriction must be derived from MyType.]
>You can then define your base type by referring to that item element:
><complexType name="basetype">
>   <sequence>
>     <element ref="item" maxOccurs="unbounded"/>
>   </sequence>
>and likewise your newtype by referring to the global declarations of
>the other elements:
><complexType name="newtype">
>   <complexContent>
>     <restriction base="my:basetype">
>       <sequence>
>         <element ref="e1" />
>         <element ref="e2" minOccurs="0"/>
>         <element ref="e3" minOccurs="0"/>
>       </sequence>
>     </restriction>
>   </complexContent>
>This works because if you take account of the substitution group, the
>basetype type definition is equivalent to:
><complexType name="basetype">
>   <sequence>
>     <choice maxOccurs="unbounded">
>       <element ref="item" />
>       <element ref="e1" />
>       <element ref="e2" />
>       <element ref="e3" />
>     </choice>
>   </sequence>
>Jeni Tennison
Received on Thursday, 14 March 2002 14:11:57 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:55:55 UTC