* Optimize the datastream generator
* Split the processing code (setup & dray) from the datastreem generator => library