An Experimental Study on Detecting Semantic Defects in Object-Oriented Programs using Software Reading Techniques

Abstract

Software defects are categorized into two main types: Syntax and Semantic defects. Syntax defects are easily to capture using compliers (or interpreters) that programming languages have. However, compliers are incapable to detect semantic defects, and this makes programmers re-read the source code a number of times to figure out where the defects locations are. Semantic defects are unable to be detected by compilers, as they are logical errors, and need the code file to be inspected carefully to catch them. This paper presents an approach to help programmers in discovering semantic defects in object oriented programming. As software code reading techniques play the most important role in capturing semantic defects, we have conducted a number of controlled experiments using three reading techniques, in order to test their efficiency and effectiveness.