Failed Error Propagation


In order for a faulty statement s in program p to be found during testing, a test case used must lead to statement s being Executed, the resultant program state must be erroneous (Infection occurs), and this erroneous program state must Propagate to the output. Previous experiments have demonstrated that failed error propagation (FEP) occurs in practice and therefore reduces test effectiveness. In this line of work we are exploring the conceptual similarities between FEP and Quantified Information Flow (QIF) and this reseach has resulted in QIF based metrics that have been found to correlate strongly with the proportion of tests that suffer from FEP. The longer term goals are to develop a more general information theory based foundation for software testing and to devise test methods that utilise tractable estimates of QIF-based metrics.



Back to Rob Hierons' home page

Last updated: February 2014.


Disclaimer The contents of this page falls outside the responsibility of Brunel University.