Program Maintenance Aid Using Static Data Flow Analysis Techniques

摘要

Research reveals that data flow and data dependencies in a program significantly contribute to its complexity. However, data dependency research has produced only single-valued measures of that complexity; those complexity measures provide little to help the maintenance programmer understand data flow in a program. Needed is a maintenance tool that will reveal to the programmer sources of the data flow complexity. This thesis presents a specification for a static intraprocedural data flow analysis tool. It computes six data flow properties of programs: definition without a corresponding use, possible use without a corresponding definition, two distinct uses of a variable, dependency slice of a variable, influence slice of a variable, and influence of a predicate. The first three properties reveal potential program errors and the last three show how statements in a program influence the outcome of other statements. A modified Generalized Program Graph is used to model the program control flow and data flow information needed to compute the properties. Algorithms are presented to compute each property from the Generalized Program Graph. (Theses)