To be more clear, I am not wondering about the direction of Mozilla's code base and future directions as a goal, but rather about the amount of work involved.
I know enough about python to C/C++ language binding to be dangerous via boost::python. To me the two advantages of using boost::python over other methods is not having to handle reference counting (unless one is invoking some voodoo), and if it compiles it just works (unless I messed up on managing the memory in my own code ...). I am not suggesting that boost::python is even a relevant tool for xulrunner bindings (or whatever you want to call the translation between the DOM and python), but I am wondering why it has to be so difficult. I understand requiring reference counts in C, but such details should be hidden from javascript or any other "higher" language.
