Name: skT45625 Date: 07/31/2000


java version "1.2.2"
HotSpot VM (1.0.1, mixed mode, build f)



The JAXP 1.0.1 validating parser seems to "expand" CDATA sections
when it is parsing them. Taking a look at the following example:

test.dtd:
<!ELEMENT A(#PCDATA)>

test.xml:
<!DOCTYPE A SYSTEM "file:test.dtd">
<A> <![CDATA[Hello]]> abcd <![CDATA[World]]> &lt; <![CDATA[xyz]]> </A>

When the JAXP validating parser parses the above and the output
is write out, the product is:

<A> <![CDATA[Hello abcd ]]><![CDATA[World ]]>&lt; <![CDATA[xyz ]]></A>

It seems that the CDATA sections are expanded to include all characters
before the next CDATA section, a character reference, or an end tag.

p.s. It would be useful if you add an XML category to the bug list
above (and on the previous page). I could not find an appropriate
place to file this bug.

ALso, I tried to include the character reference "&lt;", but it might
be converted to the character "<" when submitted.
(Review ID: 107768)
======================================================================