Detecting errors in software using a parameter checker: An Analysis

This paper will discuss a study of parameter errors that occurred in a software product developed using the C language. Through the use of a parameter checker that was developed from an existing compiler, the frequency and type of parameter errors over several releases of an application are studied. In addition to parameter checking, other static analysis data was collected concerning function usage (or non-usage) and function return value comparisons. This paper may serve as a useful complement to previous static analyses done for the FORTRA, COBOL and Pascal languages. The study presents statistics that show the large number of parameter errors caused by the weak type-checking of the C language. This paper also discusses the cost associated with detecting and repairing parameter errors that might otherwise have been detected by static analysis methods. In summing up the presence of these parameter errors the author discusses the impact of the ANSI C standard on this issue. The code used to build the data analysis portion of the Parameter Checker is presented in the appendices