Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

Doris L. Carver


The process of maintenance and enhancement of legacy software systems is a laborious and unavoidable task. Often these systems lack structure or modularity, as they were developed using programming languages and paradigms that do not incorporate object-oriented features and sound design principles. The software engineer's task can be simplified if tools are available to identify object like features in the code. These tools can help transform the non-object-oriented code to object oriented code. This research describes a comprehensive and systematic process for transformational reengineering of legacy systems. Research in reengineering is mainly focused on clustering techniques that group procedures present in the legacy system into candidate objects. These clustering approaches are limited to systems with well-defined data structures and procedures. Several of these approaches are either not comprehensive, limited to certain types of systems, or depend extensively on engineer knowledge of the system. Unlike these approaches that analyze legacy systems at the procedural level, the reengineering process we present analyzes systems at the statement level. This process results in the identification of object operations. These operations, along with the state variables and the user defined data structures, are arranged in a hierarchy that represents the object structure of the reengineered variant of the legacy system. From this system hierarchy, objects are identified and encapsulated by streamlining the interfaces. The reengineering process is incorporated in a tool, ReArchitect. Programs are statically analyzed and represented as a statement dependence graph (StDG) for further processing. The StDG is a fine-grained representation with modular representation for functions and program slices. It can adapt to program changes, unlike other representations. The StDG is restructured by merging cohesive components in the graph. The restructured graph is used to build the object structure, which is used to identify the objects. The StDG is a theoretically sound framework that provides support for many problems found in the reengineering domain. We show the value of the StDG in two such domains: program slicing and maintenance. The StDG is restructured differently for different requirements (space/time), and for different types of applications.