Why does Barracuda choose the way to parse the whole page (*ML) to a doc?

I am trying to understand Barracuda/XMLC right now. I
am wondering why XMLC parse the whole *ML to a
Document. The doc tree is so big. Why don't just parse
to a smaller doc tree? So the perfomance will be
improved significantly.

______________________________________________________________________ 
Post your free ad now! http://personals.yahoo.ca