A Design Unit Based Code Generation Technique for Object-Oriented Software Development

Generating executable code from some high-level design notation has been an active research area. Carlson [1] and Hurlbut [4] proposed an adaptive use case methodology for software development. It is the deployment of design techniques that lead to semiautomatic processes for code generation and test plan generation. The focus of this paper is on the development of a code generation technique that achieves several quality control properties supported by the use case methodology. Our code generation process starts from such high level design as interaction diagrams and eventstate tables because we want to get the control over the quality of the code and establish a manageable relationship between design and actual implementation. Its foundation is the object-oriented software design approach that partitions design schemata into a layered architecture of functional components called design unit. During the design phase, interaction diagrams and event state tables are developed from which design units are then identified. The skeletal code is generated from event-state table based on design units. Code segments are organized based on current state-logic for each transition, current event-logic for each current state, and next state-logic for each current state/event transition to next state. The proposed code generation technique provides a programmer with guidelines to write source code. In addition, it provides reusability, traceability, and testability properties.