The tricky part (in all of these schemes really) is knowing how to handle the coordinates. They'd have to be put in the msgid together with the actual stuff to translate. But in fact you can't know what to set the coordinates to unless you see a graphical representation of the dialog... This would mean we need a real resource editor :-/
Having an open source graphical resource editor would be usefull for more than just translating WINE dialogs :)