Date:         Fri, 10 Dec 2004 21:46:42 -0500
Reply-To:     Metadata Object Description Schema List <[log in to unmask]>
Sender:       Metadata Object Description Schema List <[log in to unmask]>
From:         Bruce D'Arcus <[log in to unmask]>
Subject:      RELAX NG vs. XML Schema, and alternative MADS suggestion
Comments: To: Metadata Object Description Schema List <[log in to unmask]>
Content-Type: text/plain; charset=US-ASCII; format=flowed

I just threw together a quick demo of my thinking about the structure
of MADS, and of the beauty of RELAX NG (in contrast to XML Schema).

So I use the compact syntax, because it's easier to work with than the
verbose XML; so easy, in fact, that I can write the schema with the
simplest of text editors.

Anyway, here's the schema (again, just a demo):

===
default namespace = "http://www.loc.gov/mads"


## root is either a single mads element, or more than one MADS
## element wrapped in a madsCollection
start = MADS | MADS_Collection

## there is only a single main element, with a choice between
## person and work here (but would be extended in a full schema)
MADS = element mads { Person | Work }
MADS_Collection = element madsCollection { MADS+ }

## person element contains all related content
div {
   Person = element person { PersonName, Affiliation*, Address* }

   ## All main content is divided between a single authority (I'd prefer
"primary" myself)
   ## element and zero or more variants
   PersonName = element name { Name-Authority, Name-Variant* }
   Name-Authority = element authority { NameParts-Person }
   Name-Variant = element variant { NameParts-Person,
Variants-Attributes }
   NameParts-Person = NamePart-Given* & NamePart-Family*
   NamePart-Given = element givenname { text }
   NamePart-Family = element familyname { text }
   Affiliation = element affiliation { text }
   Address = element address { text }
}

## Work
div {
   Work = element work { Title }
   Title = element title { Title-Authority, Title-Variant* }
   Title-Authority = element authority { Titles }
   Title-Variant = element variant { Titles }
   Titles = TitleMain, SubTitle?
   TitleMain = element titleMain { text }
   SubTitle = element titleSub { text }
}
Variants-Attributes = attribute type { "abbreviation" | "alternate" |
"equivalent" }?

===

Here's the same schema automatically converted to (compliant!) XML
Schema via Trang:

<?xml version="1.0" encoding="UTF-8"?>
<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema"
elementFormDefault="qualified"
targetNamespace="http://www.loc.gov/mads"
xmlns:mads="http://www.loc.gov/mads">
   <xs:element name="mads">
     <xs:complexType>
       <xs:choice>
         <xs:element ref="mads:person"/>
         <xs:element ref="mads:work"/>
       </xs:choice>
     </xs:complexType>
   </xs:element>
   <xs:element name="madsCollection">
     <xs:complexType>
       <xs:sequence>
         <xs:element maxOccurs="unbounded" ref="mads:mads"/>
       </xs:sequence>
     </xs:complexType>
   </xs:element>
   <xs:element name="person">
     <xs:complexType>
       <xs:sequence>
         <xs:element ref="mads:name"/>
         <xs:element minOccurs="0" maxOccurs="unbounded"
ref="mads:affiliation"/>
         <xs:element minOccurs="0" maxOccurs="unbounded"
ref="mads:address"/>
       </xs:sequence>
     </xs:complexType>
   </xs:element>
   <xs:element name="name">
     <xs:complexType>
       <xs:sequence>
         <xs:group ref="mads:Name-Authority"/>
         <xs:group minOccurs="0" maxOccurs="unbounded"
ref="mads:Name-Variant"/>
       </xs:sequence>
     </xs:complexType>
   </xs:element>
   <xs:group name="Name-Authority">
     <xs:sequence>
       <xs:element name="authority" type="mads:NameParts-Person"/>
     </xs:sequence>
   </xs:group>
   <xs:group name="Name-Variant">
     <xs:sequence>
       <xs:element name="variant">
         <xs:complexType>
           <xs:complexContent>
             <xs:extension base="mads:NameParts-Person">
               <xs:attributeGroup ref="mads:Variants-Attributes"/>
             </xs:extension>
           </xs:complexContent>
         </xs:complexType>
       </xs:element>
     </xs:sequence>
   </xs:group>
   <xs:complexType name="NameParts-Person">
     <xs:choice minOccurs="0" maxOccurs="unbounded">
       <xs:element ref="mads:givenname"/>
       <xs:element ref="mads:familyname"/>
     </xs:choice>
   </xs:complexType>
   <xs:element name="givenname" type="xs:string"/>
   <xs:element name="familyname" type="xs:string"/>
   <xs:element name="affiliation" type="xs:string"/>
   <xs:element name="address" type="xs:string"/>
   <xs:element name="work" type="mads:Title"/>
   <xs:complexType name="Title">
     <xs:sequence>
       <xs:element ref="mads:title"/>
     </xs:sequence>
   </xs:complexType>
   <xs:element name="title">
     <xs:complexType>
       <xs:sequence>
         <xs:group ref="mads:Title-Authority"/>
         <xs:group minOccurs="0" maxOccurs="unbounded"
ref="mads:Title-Variant"/>
       </xs:sequence>
     </xs:complexType>
   </xs:element>
   <xs:group name="Title-Authority">
     <xs:sequence>
       <xs:element name="authority" type="mads:Titles"/>
     </xs:sequence>
   </xs:group>
   <xs:group name="Title-Variant">
     <xs:sequence>
       <xs:element name="variant" type="mads:Titles"/>
     </xs:sequence>
   </xs:group>
   <xs:complexType name="Titles">
     <xs:sequence>
       <xs:element ref="mads:titleMain"/>
       <xs:element minOccurs="0" ref="mads:titleSub"/>
     </xs:sequence>
   </xs:complexType>
   <xs:element name="titleMain" type="xs:string"/>
   <xs:element name="titleSub" type="xs:string"/>
   <xs:attributeGroup name="Variants-Attributes">
     <xs:attribute name="type">
       <xs:simpleType>
         <xs:restriction base="xs:token">
           <xs:enumeration value="abbreviation"/>
           <xs:enumeration value="alternate"/>
           <xs:enumeration value="equivalent"/>
         </xs:restriction>
       </xs:simpleType>
     </xs:attribute>
   </xs:attributeGroup>
</xs:schema>
===

And here's a simple example instance created against the above schema:

<mads xmlns="http://www.loc.gov/mads"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="http://www.loc.gov/mads
file:/Users/darcusb/Desktop/mads/mads-alt.xsd">
   <person>
     <name>
       <authority>
         <givenname>John</givenname>
         <familyname>Doe</familyname>
       </authority>
       <variant type="alternate">
         <givenname>Jonathan</givenname>
         <familyname>Doe</familyname>
       </variant>
     </name>
   </person>
</mads>

I think the structure of the schema more sense.  Consider the xpath
expression for the name authority:

        mads:mads/mads:person/mads:name/mads:authority

... and for all variants:

        mads:mads/mads:person/mads:name/mads:variant

Also, the validation is MUCH tighter.  From what I gather of the
current MADS draft, it would be perfectly valid (though also perfectly
wrong) to have an authority for a name, but a variant element that
contained a title.

With this structure, there is no such sloppiness; content is tailored
to each top-level element.

Consider the technical issues too: the RELAX NG source is much easier
to work with (there's a reason why major schemas like OpenOffice and
DocBook are developed in RELAX NG and not XML Schema), and the same
source produces a clean XML Schema without the obvious bugs in XMLSpy.

Bruce