Question

I saw various cases where elements in XML file are prefixed with x: or something else.

What is the purpose for this?

Also, what is the meaning of the “namespace” for the xml? (usually some URL)? how is it being used in any way?

Answer 1

The x: is a namespace identifier in your XML. An XML subtree can have one »default« namespace for elements and attributes (set by xmlns='...'). If you need elements or attributes from other namespaces you need to define a prefix, such as xmlns:x='...'. Elements and attributes you use from that namespace then need to be prefixed with x:. A common one I have is

<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"

which then requires me for the use element to do the following:

<use xlink:href='#foo'/>

In this case the use element comes from SVG, while the href attribute comes from XLink. Both are different, but in some places compatible specifications.

And that’s basically what namespaces are for:

• Group different XML uses into coherent parts
• Allow them to be used in a single XML
• Potentially avoid name clashes (if you have two namespaces and both allow for an element called foo, then you need to specify which of the two foos you want.

In that way namespaces are not much different from how they work in many programming languages. You could view the things you import by namespaces a little like libraries in above SVG case. The people writing SVG noticed that there is already a specification allowing linking to arbitrary XML elements (XLink) and they simply re-used it. In other cases, such as WPF, tehre exist namespaces to separate the declarational UI stuff of the presentation framework and the glue attributes needed to interface with code. E.g. a window there might look like this:

<Window
    x:Class="W"
    Height='500'
    Width='500'
    xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
    xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml">

The Class attribute is only needed for the compiler to know the appropriate codebehind class, it doesn’t have any visible difference on the window.

Another example is XSLT, where you define transformations on XML documents. Both the XSLT instructions and the templates for the output are XML. You usually use an explicit namespace for XSLT’s elements:

<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">

so that you can emit generated XML like so:

<xsl:if test="Content/Chapter">
  <xsl:variable name="text"><xsl:value-of select="$trans//item[@name='content'][lang($lang)]" /></xsl:variable>
  <h2><xsl:value-of select="$text"/></h2>
    <ol>
      <xsl:apply-templates select="Content" mode="TableOfContents"/>
    </ol>
</xsl:if>

If I were to make the xsl: namespace the default namespace, then the XSLT processor couldn’t tell my to-generate HTML from the actual instructions. That’s a problem if both your code and data use the same format and live in the same place. I believe Lisp solved that by marking data with a single-quote.

As for namespaces being an URI, that’s probably just a W3C thing. There is no requirement for anything to live at those URIs. They are just identifiers. They are not meant to be retrieved. A program consuming XML with namespaces must know the namespaces it can handle and act appropriately.