I’ve just this morning read an interesting article on combinatorial or pairwise testing (http://www.testingeducation.org/wtst5/PairwisePNSQC2004.pdf and http://shrinik.blogspot.com/2007/03/boundary-value-exploration-bve.html .) This article has been around for while but it covers a very valid point. Pairwise testing is Domain testing or as ISTQB call it Equivalence Partitioning. Equivalence Partitioning “…is based on the premise that by testing one value within a class the result is representative of all values within that class.”
The technique, on the surface, has merit but how do we test for the unknown? What if our test data doesn’t cover that exception? Pairwise testing is a valid technique BUT as with all testing, it should be combined with the testers skill and judgement. To follow one technique for the sake of the technique is but to limit one’s effectiveness.
Judgement and skill are hallmarks of a good tester. Education and knowledge of the ‘craft’ are more important than the expertise in a particular system or application. There are many good subject matter experts but not alot of really good testers…testers that can think, change and adapt.
As Robert Sauborin likes to frequently quote in his course slides “No! Try not, Do. Or do not.There is no try.” – (Master Yoda instructing young Luke Skywalker in the ways of the force.)
In otherwords, look beyond what you do know…there may be another way to test your product…do it and evaluate for yourself. In is only after a period of exploration and experimentation that we can become aware of what the system can do.
Just because you (or the team) have always tested the system the same way doesn’t necesarily mean that it is the be all and end all of testing. Look outside the square and think!!