A couple of quick questions for all the incredibly brainy and wise monks out there, how does TT2 implement stream parsing, and why is it good practice to run an arbitary interpretted templating language on top of another interpretted language, and why is it better to limit the system to the reduced scope of the interpretted template language, when one can just use eval and have the full power and expressiveness of perl at ones fingertips... Can't quite figure that one out myself.