Thou shalt have no other gods before the ANSI C standard 1606

Douglas A. Gwyn

How does that work again? I don't see how you can deduce path coverage, given only statement coverage figures.

Consider this code snippet: if (a) f1(); else f2(); if (b) g1(); else g2(); How do you tell whether your tests have attained complete path coverage? For instance, suppose that you run 4 tests, and you observe that every then branch is executed exactly 2 times, and the same for every else branch. Can you conclude that complete path coverage has been attained?

I see two scenarios:

Scenario #1: The first test executed the path f1() g1(). The next test executed f1() g2(), the third f2() g1(), the fourth f2() g2(). This gives complete path coverage, and it is consistent with your statement coverage observations.

Scenario #2: The first two tests followed the path f1() g1(). The last two tests followed f2() g2(). This does not give complete path coverage, but it is still consistent with your statement coverage observations.

Conclusion: If all you have are source-line usage counters, there are situations that are ambiguous: you cannot infer, given only statemetn coverage information, whether complete path coverage has been attained.