> ><complexType name="AttributeValueType">
> > <sequence>
>
> Do you mean <choice> here? Having these all in order doesn't
> seem right.
yeah
> > <any namespace="##any" processContents="lax" minOccurs="0"
> >maxOccurs="unbounded"/>
> > <element ref="saml:BooleanData" minOccurs=0
> >maxOccurs="unbounded"/>
> > <element ref="saml:IntegerData" minOccurs=0
> >maxOccurs="unbounded"/>
> > <element ref="saml:BooleanData" minOccurs=0
>
> (StringData, I assume)
again yeah
> >maxOccurs="unbounded"/>
> > </sequence>
> ></complexType>
> ><element name="BooleanData" type="boolean"/>
> ><element name="IntegerData" type="integer"/>
> ><element name="StringData" type="string"/>
> >
> >
> >Using generic slots the namespace can specify the use of the
> slot, for
> >example:
> >
> ><AttributeValue Namespace="urn:whatever:academic-agreement-2001"
> > Name="MemberOfFaculty">
> > <StringData>MIT.EDU</StringData>
> ></AttributeValue>
> >
> ><AttributeValue Namespace="urn:whatever:academic-agreement-2001"
> > Name="IsA Student">
> > <BooleanData>true</BooleanData>
> ></AttributeValue>
> >
> >
> >This leave the problem that the attribute slot is intended
> to be extensible
> >but an extension schema can't declare an extension that is
> only for use as
> >an attribute value.
> >
> >With generic data slots and an extension point:
> >
> >
> ><complexType name="AttributeValueType">
> > <sequence>
> > <any namespace="##any"
> processContents="lax" minOccurs="0"
> >maxOccurs="unbounded"/>
> > <element ref="saml:AttributeElement"/>
> > <element ref="saml:BooleanData" minOccurs=0
> >maxOccurs="unbounded"/>
> > <element ref="saml:IntegerData" minOccurs=0
> >maxOccurs="unbounded"/>
> > <element ref="saml:BooleanData" minOccurs=0
> >maxOccurs="unbounded"/>
> > </sequence>
> ></complexType>
> ><element name="AttributeElement" type="saml:AttributeElementType"/>
> ><complexType name="AttributeElementType"/>
>
> If you do this, then there's no reason to have the ##any wildcard, is
> there? We should pick one of the two ways, I think...
The problem is that the rest of the schema uses the 'strongly typed'
extension mechanism but when it comes to attributes people wanted to
be able to use #any...
I would prefer having the strongly typed extension point, the problem
with that is that if someone wanted to use an existing attribute
schema (e.g. name and address directory data) they would then have to
create a wrapper for that purpose alone.
On the other hand if you are creating something that is intended to be
used as an attribute extension then the 'principled approach' is
more desirable.
Phill
