Enhanced Code Generation from UML Composite State Machines

UML modeling tools provide poor support for composite state machine code generation. The generated code is typically complex and large, especially for deeply-nested composite state ma-chines. Existing approaches either do not handle this case at all or handle it by flattening the com-posite state machine into a simple one with a large number of states, possibly leading to a combinatorial explosion that reflects on the size of the generated code. This paper presents a new code generation approach for composite state machines. Our approach transforms a composite state machine into an equivalent set of simple state machines before code generation. This effectively avoids the combinatorial explosion of states and leads to a more concise and scalable generated code. We implemented our approach in the context of the model-oriented programming language, Umple. We also report on a case study where we compare our approach to others in terms of code size and scalability.