search for: hasn_map

...e locks already, not that you can't take out locks. In fact, it seems to me it must be telling you this precisely so you know that it is safe to take out locks without worrying about existing locks the thread might already have. > I don't know if that can be helped, since access to stl hasn_map must > be synchronized. > On the other hand, I understand that reconstructing java objects on > "C-side" is probably not a good idea either. > > My first thought was to implement explicit "destroy" methods on java > objects in org.xapian so that it will be an...