Applying Program Comprehension Techniques to Improve Software Inspections

Software inspections are widely regarded as a cost-effective mechanism for removing defects in software, though performing them does not always reduce the number of customer-discovered defects. We present a case study in which an attempt was made to reduce such defects through inspection training that introduced program comprehension ideas. The training was designed to address the problem of understanding the artifact being reviewed, as well as other perceived deficiencies of the inspection process itself. Measures, both formal and informal, suggest that explicit training in program understanding may improve inspection effectiveness. The software technical review is a widely -recommended mechanism for software defect removal. Such reviews go by many names--inspections, Fagan-style inspections, code reviews, peer reviews, formal reviews---and exhibit significant variations among organizations [Fagan, Freedman, Gilb]. All such review methods rely on the self-evident notion that software professionals are likely to find defects in software if they actually look at the products they produce. A software technical review is a meeting--along with its preparation--in which a group of software professionals (peers) does exactly that. Types of reviews are distinguished from one another by the rules governing how that examination takes place and how it relates to the overall software development or maintenance process. Impressive claims are made for the efficacy of reviews [Humphrey]. What follows is a case study in which developers were given, along with traditional (and nontraditional) instruction, explicit instruction in program comprehension concepts and techniques. The case study suggests that software engineers often have poor strategies for understanding the artifacts they are called upon to review and that providing training in comprehension skills can improve their performance significantly. A Training Opportunity One of the authors (Rifkin) was engaged by a manufacturing firm that we will call Widget, Inc. ] Widget management, having read the literature on software inspections, had expected the introduction of this practice to produce a significant decline in customer-discovered defects. The anticipated decline had not occurred, however, either in the number or percentage of defects identified by customers. 1 The firm wishes to remain anonymous and does not want to divulge raw data on defects, which it considers proprietary. The data in this paper are presented in a manner intended to respect those wishes. SEW Proceedings 115 SEL-94-006 https://ntrs.nasa.gov/search.jsp?R=1995002482