CDATA content incorrectly encoded in version 0.90.x

[florentsorel]

In version 0.90.x, strings inside <![CDATA[ ... ]]> are being encoded, which should not happen.
The content of a CDATA section is supposed to be treated as raw text and preserved as-is, without escaping characters like &.

Example:

Expected:

<![CDATA[?param1=value1&param2=value2]]>

Actual:

<![CDATA[?param1=value1&amp;param2=value2]]>

This breaks interoperability with systems expecting the original unescaped string inside CDATA sections.

[pdvrieze]

Thank you for the report. I've confirmed the issue (it is a bug in the generic writer - if you use the jvm/dom specific writers that will avoid it/work around it). It should now be fixed in the dev branch (a snapshot would come along - but maven central support for snapshot uploads is excruciatingly slow).
I have also fixed incorrect handling of ]]> in the actual cdata content (it will now split this over two consecutive cdata sections).

[florentsorel]

Thanks for the fix. I'll wait the next release.