As we have already said, it is a pretty good rule of thumb to consider anything outside of tags to be character data and anything inside of tags to be considered markup. But alas, in one case this is not true. In the special case of CDATA blocks, all tags and entity references are ignored by an XML processor that treats them just like any old character data.

CDATA blocks have been provided as a convenience measure when you want to include large blocks of special characters a character data, but you do not want to have to use entity references all the time. What if you wanted to write about an XML document in XML! Consider the following example in which you would have an example tag in your XML Guide written in XML:

<EXAMPLE>
&lt;DOCUMENT&gt;
&lt;NAME&gt;Coleen Merriman&lt;/NAME&gt;
&lt;EMAIL&gt;cm@mydomain.com&lt;/EMAIL&gt;
&lt;/DOCUMENT&gt;
</EXAMPLE>

As you can see, you would be forced to use entity references for all the tags. YUCK!

To avoid the inconvenience of translating all special characters, you can use a CDATA block to specify that all character data should be considered character data whether or not it "looks" like a tag or entity reference.

Consider the following example:

<EXAMPLE>
<![CDATA[
<DOCUMENT>
<NAME>Coleen Merriman</NAME>
<EMAIL>cm@mydomain.com</EMAIL>
</DOCUMENT>
]]>
</EXAMPLE>