Abstract

XML is successful because it can be used in many different scenarios, and because it is easy to define a schema (such as a DTD) for new scenarios, producing a tailored XML data model for this scenario. This means that names in XML documents must be interpreted as belonging to a certain schema. As long as a document uses names from only one schema, this can be done rather easily. However, in many scenarios today documents combine names from different schemas, and XML Namespaces provide a mechanism how the names in an XML document can be associated with a namespace.

Outline (How to think about Namespaces)

How to think about Namespaces [5]
Using Namespaces [5]
Defining Namespaces [5]
Processing Namespaces [3]
Conclusions [1]

Namespaces are Simple

XML Namespaces are often misunderstood
- the biggest problem is to get rid of some assumptions
- XML Namespaces are too simple and thus confusing
Instincts of Web users
1. URIs identify something that can be retrieved by a browser
2. URIs identify something that can be displayed by a browser
3. if I cannot get it and cannot look at it, what good can it be?
However, these assumptions are not always true
1. URIs identify resources which often, but not always, can be accessed over the Web
2. URIs identify resources which often, but not always, have a Web-accessible representation
3. sharing URIs means sharing an identity, which can mean sharing semantics (associated with this identity)

Simple Examples

<html xmlns="http://www.w3.org/1999/xhtml">
 <body>
  <p>Some MathML Formula:</p>
  <math xmlns="http://www.w3.org/1998/Math/MathML">
   <mfrac>

<html xmlns="http://www.w3.org/1999/xhtml">
 <body>
  <p>Some MathML Formula:</p>
  <math:math xmlns:math="http://www.w3.org/1998/Math/MathML">
   <math:mfrac>

<html xmlns="http://www.w3.org/1999/xhtml" xmlns:math="http://www.w3.org/1998/Math/MathML">
 <body>
  <p>Some MathML Formula:</p>
  <math:math>
   <math:mfrac>

<html:html xmlns:html="http://www.w3.org/1999/xhtml" xmlns:math="http://www.w3.org/1998/Math/MathML">
 <html:body>
  <html:p>Some MathML Formula:</html:p>
  <math:math>
   <math:mfrac>

Name Spaces

Names are one form of identification
Identification is essential for communications
Names in XML are not suitable for identification
- they are local to their context (where they are defined)
- if the context is uniquely identified, the names would be, too
Name Spaces: Put names into spaces
- how to identify the space? Web things are identified by URIs

URI Philosophy

Uniform Resource Identifier (URI) uniquely identify resources
URIs often provide access information
- pretty clear in http://dret.net/lectures/xml-fall06/
- less clear in urn:ietf:rfc:2648 (RFC 2648)
- very (and purposely) unclear in tag:9327493874329 (RFC 4151)
URIs often return resource representations
- the resource itself is never returned (how to return a lecture?)
- some representation often is useful (HTML, PDF, maybe video/audio)
- the resource exists and is useful without a representation!
URIs are much more than just addresses of HTML pages

The Namespace Problem

People assume that URIs point to Web pages
- a namespace name (a URI) may point to a Web page
- it may also have no Web page associated with it
- it may even use a URI scheme which cannot be retrieved
- but it is still possible to compare URIs!
People assume some standardized content format
- friendly namespaces provide HTML portals (XHTML)
- some namespaces just give you the schema (SOAP)
- less friendly namespaces provide minimal information (XSLT)
- very unfriendly namespaces may return a 404 or even use inaccessible schemes
- but they all are valid, because no resource representation is required!
Namespaces are used by comparing URIs
- anything else maybe useful, but is not strictly required
- when searching for a namespace definition, use Google (string search)

Outline (Using Namespaces)

How to think about Namespaces [5]
Using Namespaces [5]
Defining Namespaces [5]
Processing Namespaces [3]
Conclusions [1]

Declaring Namespaces

Using a namespace means referencing names from it
- unfortunately, there is no really standard way of writing these names
- the Clark notation is useful: {http://www.w3.org/1999/xhtml}html
- Namespaces are declared and then used
xmlns-prefixed attributes are used for declaring namespaces
- Default: html xmlns="http://www.w3.org/1999/xhtml"
- Prefix: xhtml:html xmlns:xhtml="http://www.w3.org/1999/xhtml"
Namespace declarations are inherited and can be overwritten
- the default namespace can be undeclared
- Namespace declarations can be used in a myriad of ways

Unhealthy Namespace Usages

Namespaces can be (and are) used in very weird ways
- these are syntax variations of identical structures
- without a good (i.e., conforming) parser, interpretation is very hard
- copy/paste can become hard or impossible
Namespaces can be neurotic, psychotic, borderline, or normal
Each of the insane cases complicates processing
None of these has any real technical inaccuracies
XML should be used with humans in mind

Unhealthy Namespace Usages in Practice

<neurotic>
 <prefix:elem xmlns:prefix="http://ns1.com/">
  <prefix:child>content</prefix:child>
 </prefix:elem>
 <prefix:elem xmlns:prefix="http://ns2.com/">
  <prefix:child>content</prefix:child>
 </prefix:elem>
</neurotic>

<borderline>
 <prefix1:elem xmlns:prefix1="http://ns1.com/">
  <prefix1:child>content</prefix1:child>
 </prefix1:elem>
 <prefix2:elem xmlns:prefix2="http://ns1.com/">
  <prefix2:child>content</prefix2:child>
 </prefix2:elem>
</borderline>

<psychotic xmlns:prefix1="http://ns1.com/" xmlns:prefix2="http://ns1.com/">
 <prefix1:elem>
  <prefix2:child>content</prefix2:child>
 </prefix1:elem>
 <prefix2:elem>
  <prefix1:child>content</prefix1:child>
 </prefix2:elem>
</psychotic>

Elements and Attributes

Namespaces often apply to elements and attributes
- if an element name has no prefix, it has no namespace or the default namespace associated
- if a name has a prefix, the prefix must be bound to a namespace name
- names like this are called Qualified Names (QNames)
Elements and Attributes are treated differently
- the default namespace only applies to unprefixed element names
- unprefixed attribute names are in no namespace
- XML Schema deals with this by keeping attributes local
Applications should interpret QNames
- naïve implementations will break when processing unhealthy instances
- the mechanics of implementing namespaces are not very hard

Other Usages

Increasingly, QNames are used in content
- XSL Transformations (XSLT) was the first specification using this
- many other technologies have followed

<xsl:template match="section" xmlns:mathml="http://www.w3.org/1998/Math/MathML/">
<xsl:if test="exists(.//mathml:*)">

Technically, everything is well-defined
- for processing, the namespace bindings must be known
- copy/paste on a textual basis may not work or even work wrong

Outline (Defining Namespaces)

How to think about Namespaces [5]
Using Namespaces [5]
Defining Namespaces [5]
Processing Namespaces [3]
Conclusions [1]

Any URI is Possible

A namespace name is a URI, that's all!
- it may not be accessible (because of the URI scheme)
- when retrieving it, nothing may be returned
- when retrieving it, something may be returned
The only important thing is the name
- the name is mentioned in the documentation
- if you know the documentation, you known the name
- shared names mean shared knowledge

Namespace Definitions

Namespaces can be defined by a DTD (XHTML)
Namespaces can be defined by an XML Schema (SOAP)
Namespaces can be defined by RELAX NG (XHTML 2.0)
Namespaces can be defined by prose (XSLT)
If schemas are provided, additional information is required
- it is unlikely that a namespace can be fully described by a schema
- additional constraints and semantics are specified in prose

Structured Namespaces

Namespaces have no structure
- a collection of names grouped by their namespace name
- inside the namespace, names have local meaning
Namespace definitions to make up their own rules
- but then they must also make rules how to deal with conflicts
XML Schema structures the namespace defined by a schema
- the different parts of the namespace are called symbol spaces
- all XML Schema components have their own symbol space
- simple and complex types share the same symbol space
- locally defined elements/attributes are in sub symbol spaces

Fixed or Extensible?

Can a namespace change over time?
- may the namespace description become outdated? extended? replaced?
- this should be clearly documented in the namespace description
The XML XML Namespace was widely believed to be defined by XML
- xml:lang and xml:space defined by XML
- xml:base was added by XML Base
- xml:id was added by xml:id
When defining namespaces, plan ahead and publish everything
- dependencies, change management, and versioning issues are important
- there still is no accepted standard for namespace descriptions

Namespace Descriptions

Erik Wilde, Structuring Namespace Descriptions, 15th International World Wide Web Conference (WWW2006), Edinburgh, UK, May 2006.

Outline (Processing Namespaces)

How to think about Namespaces [5]
Using Namespaces [5]
Defining Namespaces [5]
Processing Namespaces [3]
Conclusions [1]

Namespaces and Validity

Namespaces define an additional layer on top of XML
- they define additional semantics (assignment to namespaces)
- they define additional constraints (declaration and usage of namespaces)
Namespace-awareness is a basic requirement for XML tools
- XML not compliant with XML Namespaces will break most tools
- processing namespaces should be done by tools
- a namespace-aware parser translates namespace declarations into nodes

Namespaces in the Document

<?xml version="1.0" encoding="UTF-8"?>
<html:html xmlns:html="http://www.w3.org/1999/xhtml" xmlns:math="http://www.w3.org/1998/Math/MathML">
 <html:body>
  <html:p>Some MathML Formula:</html:p>
  <math:math>
   <math:mfrac>
    <math:mn>2</math:mn>
    <math:mi>x</math:mi>
   </math:mfrac>
   <math:mo>=</math:mo>
   <math:mfrac>
    <math:mn>3</math:mn>
    <math:mi>y</math:mi>
   </math:mfrac>
  </math:math>
 </html:body>
</html:html>

Namespaces in the Tree

Outline (Conclusions)

How to think about Namespaces [5]
Using Namespaces [5]
Defining Namespaces [5]
Processing Namespaces [3]
Conclusions [1]

Name Spaces

Bags of Names with a URI as a label
The URI does not necessarily return anything
Namespaces can be defined in any way (e.g., schemas)
If you want to be cool, simply say I hate XML Namespaces

XML Namespaces

Information Systems and the World Wide Web

International School of New Media
University of Lübeck

Erik Wilde, UC Berkeley School of Information
2007-01-09

Abstract

Outline (How to think about Namespaces)

Namespaces are Simple

Simple Examples

Name Spaces

URI Philosophy

The Namespace Problem

Outline (Using Namespaces)

Declaring Namespaces

Unhealthy Namespace Usages

Unhealthy Namespace Usages in Practice

Elements and Attributes

Other Usages

Outline (Defining Namespaces)

Any URI is Possible

Namespace Definitions

Structured Namespaces

Fixed or Extensible?

Namespace Descriptions

Outline (Processing Namespaces)

Namespaces and Validity

Namespaces in the Document

Namespaces in the Tree

Outline (Conclusions)

Name Spaces

XML Namespaces

Information Systems and the World Wide Web

International School of New MediaUniversity of Lübeck

Erik Wilde, UC Berkeley School of Information2007-01-09

Abstract

Outline (How to think about Namespaces)

Namespaces are Simple

Simple Examples

Name Spaces

URI Philosophy

The Namespace Problem

Outline (Using Namespaces)

Declaring Namespaces

Unhealthy Namespace Usages

Unhealthy Namespace Usages in Practice

Elements and Attributes

Other Usages

Outline (Defining Namespaces)

Any URI is Possible

Namespace Definitions

Structured Namespaces

Fixed or Extensible?

Namespace Descriptions

Outline (Processing Namespaces)

Namespaces and Validity

Namespaces in the Document

Namespaces in the Tree

Outline (Conclusions)

Name Spaces

International School of New Media
University of Lübeck

Erik Wilde, UC Berkeley School of Information
2007-01-09