This is not very difficult. The best way is to take a running processor (eg. ExampleProcessor stored in processors/luiza_example/) and adopt it to your needs. If you start a new analysis it is best if you create a dedicated catalogue for it, store all related code in this catalogue and compile it as a separate shared library (as is done for ExampleProcessor, see processors/luiza_example/Makefile).

Rules we tried to follow so far:

• document your code while writing.
  Use doxygen tags to describe all methods in header file. This will provide documentation for other users...
  
  
• Use initial '_' for names of class data members. This helps to distinguish them from local variables in the code.