Comparing Two Models for Software Debugging

Fault localization models in debugging

“…However, the later approach demands considerably large amount of computational power and memory space allocation compared to the earlier one Another lightweight techniques, commonly known as program slicing, has seen successful application in fault localization [15,16. The researchers of [12,16] rather made use of notion of dependences in order to perform fault localization. Thus, the models introduced in [10,11] suffers from a major shortcoming that they are unable to deal with pre-and post conditions or assertions, in a straightforward way.…”

“…In contrast, detailed fault localization capabilities are achieved in the detailed value-level models [,13, 14]. However, the later approach demands considerably large amount of computational power and memory space allocation compared to the earlier one Another lightweight techniques, commonly known as program slicing, has seen successful application in fault localization [15, 16. The researchers of [12,16] rather made use of notion of dependences in order to perform fault localization. Thus, the models introduced in [10,11] suffers from a major shortcoming that they are unable to deal with pre-and post conditions or assertions, in a straightforward way.…”

Modeling and Diagnosing Buggy Programs with Points-To Analysis

“…Model-based diagnosis [2,3] can be viewed as a process to locate faulty components in a technical system solely on the basis of its structural and behavioral model. The model-based approach to diagnosing faults is a general framework allowing the static, formal and automatic debugging of programs (see [4] for an overview and references therein), whereas statistical fault localization is to compare successful runs and failing runs of buggy program to find points of failing runs that deviate from successful runs [16]. A model checker produce error trace to indicate an error in programs, how to localize the cause in error traces is discussed in [17].…”

“…A model checker produce error trace to indicate an error in programs, how to localize the cause in error traces is discussed in [17]. And there is a body of work on model-based debugging approach, which uses model-based diagnosis for locating faults in software [4]. In [11] F. Wotawa shows model-based debugging in the context of functional dependencies has the similar capabilities as program slicing in finding faults in programs.…”

“…In the past, our colleagues have developed two categories of models in the Jade project [5,6,7,8,9,12]: the functional dependency model (FDM) [5] and the value-based model (VBM) [6]. Since the VBM can eliminate wrong diagnoses by using additional run-time semantic information like the values of variables, in general the VBM achieves better results than the FDM [7] . However, whereas VBM heavily relies on low-level information, FDM can reason at a higher level of abstraction [12].…”