Eghbali and Tahvildari examine a previously uncovered technique that orders test cases based on their coverage of code components, the additional technique (AT). They determine that ties can occur in AT-type techniques and then propose a heuristic alternative based on the AT that can efficiently break the ties that occur. AT is one technique for measuring test coverage, which is the amount of testing performed by a set of test cases.

The authors zero in on the problem with AT, namely the possibility of having a situation where more than one test case has the highest coverage of entities that have not yet been covered. Scheduling test cases in this fashion can lead to additional test cases that cannot add additional coverage, which means a tie can occur. In AT, when a tie occurs, all remaining test cases are considered equally important and are selected at random, which is ineffective and therefore a drawback of AT. Based on this drawback, the authors embark on an experiment to show that a variation of the AT that uses lexical ordering would be superior to basic AT. The main focus of their proposed technique, termed generalized additional technique, is to break the ties that AT-type techniques suffer from. To address it, the authors present an algorithm that takes care of the ties that occur in the proposed technique by taking into account all possible candidates and evaluating each one to find the best test case. The real intellectual kernel of the proposed solution is how, unlike other traditional coverage-based alternatives for resolving tie cases, the authors ranked different software entities “based on the number of times they are covered until the current step.” This made it possible for all entities to be “considered in the selection of the next test case,” while at the same time giving higher priority to entities that are covered less.

It must be said that the authors provide a compelling case as to why their proposed technique is superior to other alternatives to AT for test case prioritization. Their analysis of the time complexity of the AT basic algorithm and the algorithm of their proposed technique is very interesting. Despite the modifications to the basic algorithm to derive the algorithm for the proposed technique, the fundamental functionality did not change and still resulted in a lower computational cost. They also evaluated the results of other techniques, compared them to the proposed technique, and determined that their algorithm outperforms every other technique in all cases, except AT. However, their technique still outperforms AT in four out of six cases. Based on this, it is not hard to see why the authors concluded that while it is very likely for AT-type techniques to face ties when prioritizing test cases, their proposed technique can break ties more efficiently than any other known technique.