XML Design Guidelines for Facilitating the Creation and Processingof Com

["Costello, Roger L." <costello@mitre.org>]

Hi Folks,

What guidelines would you recommend to someone who: 

   - wants to create an XML vocabulary, 

   - that will be combined with other XML vocabularies to create a compound instance document, 

   - which will be processed by applications?

In other words, what XML designs will facilitate the creation and processing of compound documents?

Below is a list of guidelines that I have come up with. Are there other guidelines that you recommend?  


----------------------------------------------
   GUIDELINES FOR FACILITATING THE
      CREATION AND PROCESSING
       OF COMPOUND DOCUMENTS
---------------------------------------------- 

Namespaces are key to creating compound documents that can be effectively processed. Applications can partition compound documents along the lines identified by namespaces, thus enabling module-specific processing.


1. When designing a schema, associate the XML vocabulary it creates with a namespace. Thus the first guideline is to identify each XML vocabulary via a namespace. Here's an example of a Book XML vocabulary identified by the http://www.book.org namespace: 

    <Book xmlns="http://www.book.org";>
         <Title>Illusions</Title>
         <Author>Richard Bach</Author>
         <Date>1977</Date>
         <ISBN>0-440-34319-4</ISBN>
         <Publisher>Dell Publishing Co.</Publisher>
    </Book>

Conversely, do not identify the XML vocabulary by embedding a unique identifier inside an element or attribute. Here's an example of the Book XML vocabulary identified using a GUID embedded inside an element: 

    <Book>
         <Identifier>fc48ee30-0a6b-11de-8c30-0800200c9a66</Identifier>
         <Title>Illusions</Title>
         <Author>Richard Bach</Author>
         <Date>1977</Date>
         <ISBN>0-440-34319-4</ISBN>
         <Publisher>Dell Publishing Co.</Publisher>
    </Book>


2. If the schema language used to create the XML vocabulary provides the option of namespace-qualifying all elements or namespace-qualifying only global elements, choose the former. For example, the W3C XML Schema language allows you to specify "all elements must be qualified in the instance document" by setting:

    elementFormDefault="qualified"

Conversely, it allows you to specify "only qualify globally declared elements in the instance document" by setting:

    elementFormDefault="unqualified"

Don't use the latter as it destroys the value of namespaces. Instead, use the former as it maximizes the value of namespaces.


3. Use standard filename suffixes: .xsd (W3C XML Schema), .rng (ISO RELAX NG), .sch (ISO Schematron), .dtd (W3C DTD). Although it is technically accurate to, say, suffix an XML Schema file with .xml, it is preferable to use .xsd as the latter enables applications to more easily recognize an XML vocabulary's schema language.


4. Design your schema maximizing the use of globally declared elements. Each globally declared component is a potential compound document component. For example, with this schema both BookStore and Book can be used as a component in a compound document:

    <element name="BookStore">
        <complexType>
            <element ref="Book" maxOccurs="unbounded" />
        </complexType>
    </element>

    <element name="Book"> ... </element>

Conversely, this schema only allows BookStore to be used as a component in a compound document: 

    <element name="BookStore">
        <complexType>
            <element name="Book" maxOccurs="unbounded"> ... </element>
        </complexType>
    </element>


What else would you add to this list?

/Roger