Abstract

In subdomain testing, the input space of the program being tested is divided using a criterion of “sameness,” grouping input points such that it seems unnecessary to try many in the same group. These subdomains are then sampled: it is required that each be covered by at least one test. There are a number of pitfalls in this procedure because the points in a subdomain are never truly “the same”; nevertheless, the method is the only imaginable way to systematize testing.