predicatecoverage

Predicate coverage is a software testing coverage criterion focused on the boolean sub-expressions within conditional statements. It requires that each atomic predicate—the individual boolean sub-expression—takes on both the true and false values across the test suite. The goal is to detect faults that arise when a particular sub-expression behaves differently than expected, independent of the overall decision outcome.

In practice, predicate coverage examines expressions in conditions such as if (A && B) or while (X || Y). Each predicate A, B, X, Y should be evaluated to both true and false in some test cases. Short-circuit evaluation can complicate this: for example, in A && B, B is only evaluated when A is true, so test cases must be chosen so that both A and B are observed in both truth values when feasible.

Predicate coverage sits between other criteria in the family of test coverage measures. It is more focused than general condition coverage, which looks at the truth value of the entire expression, but often weaker than multiple-condition coverage or modified condition/decision coverage (MC/DC), which require more stringent testing of how each predicate influences the overall decision.

Limitations include its inability to guarantee all meaningful interactions among predicates, especially in complex expressions with short-circuit behavior. It may miss faults arising from specific combinations of sub-expressions that do not change the overall decision in the tested cases. Consequently, predicate coverage is typically used in conjunction with other criteria to achieve comprehensive test adequacy.