Evidence driven object identification in procedural code

Abstract

Software evolution is an integrated part of software maintenance. It may take the form of porting a legacy system to a new hardware platform operating system, translating the system to a new language or rearchitecting the system to take advantage of new programming paradigms. This paper presents techniques for the identification and recognition of object-oriented structures in legacy systems that have been implemented using a procedural language. The paper examines methods for the selection of object classes and the recovery of the possible associations between the recovered classes.