An Investigation of Java Faults Operators Derived from a Field Data

Abstract. The knowledge of real software faults representativeness is important to allow the emulation of software faults in a more accurate way through software fault injection techniques. This paper presents a field data study to analyze the representativeness of Java software faults, including security faults. The faults are classified according to a previous field study of C faults representativeness and new types of faults are identified due to the specific characteristics of the Java language structure. Results are compared and show that the mistakes most commonly made by programmers follow a pattern, independently of the programming language. 1