Software fault localization in spreadsheets

Spreadsheets are well known and used in many companies, such as for financial reporting, price calculations, and forecasting. This chapter explains how to apply spectrum-based fault localization on spreadsheets. It shows how to convert a spreadsheet into a set of constraints so that model-based debugging can be used for localizing faulty cells. The chapter defines test cases for spreadsheets and formulates the spreadsheet debugging problem. Cones in spreadsheets are the equivalent of slices in programs. Slices are computed by considering several different types of dependencies, such as data and control dependencies. In the spreadsheet domain, the similarity coefficient expresses the correlation of erroneous values and the cells that contribute to the computation of these values. In the value-based model, the values of the spreadsheet's formulas are propagated. The chapter discusses two repair techniques, namely goal-directed debugging and mutation-based spreadsheet debugging.