An Abstract Operational Framework for Dependence Models in Software Debugging

In this article we introduce an abstract operational framework employing Aspect´s notion of abstract dependences. We exemplify that this framework allows one for motivating certain desirable model properties in an intuitive and straightforward way. Moreover, we propose an algorithm that allows for obtaining a static program slice from an abstract execution trace. This computation resembles computing a dynamic slice for a certain test case, but unlike to a dynamic slice, does not suffer from error masking since it covers all possible executions. We show how to employ abstract dependences to determine those variables upon which computation of the slice is most beneficial. Moreover, we show how to best use the obtained static slices to compute diagnoses, that is - fault candidates, in terms of hitting sets.