Programming Style Based Program Partition

Program partitioning is a task of splitting a large, complex software system into functionally independent program modules. It is a key step in program understanding, software maintenance and software reuse. Traditional program partitioning methods are nonlinear. In most cases, the computational efforts needed for partitioning a source program will increase exponentially with the size of the source program. The NP-hard complexity constitutes a computational barrier for partitioning legacy software systems properly and efficiently. In this paper, we propose a new method that can partition a source program into program modules within a timescale that is linear with the size of the program. Our method uses special heuristic knowledge, based on psychological analysis on human programming styles, to partition a source program into domain-oriented program modules. A case study on a legacy C program that consists of 92 functions is reported to demonstrate the efficiency and effectiveness of this method.