"alex.loubyansky(a)jboss.com" wrote : The attributes don't belong to the CDATA, so it's logical that they are not set in this case. And it would even look confusing to me if we "fixed" it so that they are set. | | Instead, the key should be unmarshalled to an object on which we would set its children that the attributes and the CDATA are. Yes, but it is more complicated than that. Its polymorphic and a mixed type. The way I've done the wrapper for the CDATA is obviosly wrong. It doesn't work if the wrapper has attributes like this case. We should do like the old parsing did, which is to assume the wrapper (StringValueMetaData) but then replace if we see a nested element (e.g. ListValueMetaData). I don't think its just a case of creating a new ParticleHandler to create the wrapper because we don't want the ListValueMetaData injected onto the StringValueMetaData we want to replace it if we see a subelement. I'll look at it once I've resolved the other issues I've found. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4134566#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...