Publications

An object-oriented methodology is presented that is based on two sets of Data Flow Diagrams (DFDs): one for the functional view, and one for the behavioral view. The functional view presents the information flow between shared objects. These objects map to the classes identified in the structural view (e.g., Information Model). The behavioral view presents the flow of information between control components and relates these components to their state models. Components appearing in multiple views provide a bridge between the views. The top-down hierarchical nature of the DFDs provide a needed overview or road map through the software system.

Current CASE technology provides sophisticated diagramming tools to generate a software design. The design, stored internal to the CASE tool, is bridged to the code via code generators. There are several limitations to this technique: (1) the portability of the design is limited to the portability of the CASE tools, and (2) the code generators offer a clumsy link between design and code. The CASE tool though valuable during design, becomes a hindrance during implementation. Frustration frequently causes the CASE tool to be abandoned during implementation, permanently severing the link between design and code. Current CASE stores the design in a CASE internal structure, from which code is generated. The technique presented herein suggests that CASE tools store the system knowledge directly in code. The CASE support then switches from an emphasis on code generators to employing state-of-the-art reverse engineering techniques for document generation. Graphical and textual descriptions of each software component (e.g., Ada Package) may be generated via reverse engineering techniques from the code. These reverse engineered descriptions can be merged with system over-view diagrams to form a top-level design document. The resulting document can readily reflect changes to the software components by automatically generating new component descriptions for the changed components. The proposed auto documentation technique facilitates the document upgrade task at later stages of development, (e.g., design, implementation and delivery) by using the component code as the source of the component descriptions. The CASE technique presented herein is a unique application of reverse engineering techniques to new software systems. This technique contrasts with more traditional CASE auto code generation techniques.