Reducing Screened Program Points for Efficient Error Detection
Ricardo Manuel Nascimento Castilho · 2012
Despite extensive testing in the development phase, residual defects can be a great threat to dependability in the operational phase. Lately there have been some studies that proved the utility of generic invariants, often dubbed fault screeners as error detectors within a spectrum-based fault localization (SFL) approach aimed to diagnose program defects in the operational phase. The generic nature of this generic screeners allows them to be automatically instrumented without any programmer-effort, while training is straightforward given the test cases available in the development phase. Experiments based on the Siemens program set demonstrate diagnostic performance that is similar to the traditional, development-time application of SFL based on the program pass/fail information known before-hand. In this document we present a study that researches the possibility of further improving the performance of this automatic error detection technique by reducing the number of generic invariants used to screen a program. Here we describe the adopted metrics in order to evaluate the diagnosis quality of each screened program point. In the same baseline, we investigate about the variables that have the highest diagnosis effectiveness, the so-called collar variables, and how the current false negative rate of the techniques used hindered their discovery in our benchmark set up.