Lucene RSS search feed malforms output with <pre>

Lucene's search results can be exported as RSS, like on xwiki.org:
http://www.xwiki.org/xwiki/bin/view/Main/Search?xpage=plain&text=fail

It looks like valid RSS code but in fact sarts with <pre> followed by html-escaped tags.
So you see:

<?xml version="1.0" encoding="ISO-8859-1"?>

But it actually is:

<pre>&lt;?xml version="1.0" encoding="ISO-8859-1"?&gt;