Practice · Verification and Validation I: Testing Fundamentals

Q1

Omar misunderstood the late-penalty rule and wrote if (daysLate >= 7) return 0;. A student 7 days late then received 0. Which term describes the wrong line of code?

A Failure B Error C Fault (defect) D Oracle

Q2

What is a test oracle?

A A tool that generates random inputs B The source of the expected result — the mechanism that decides whether an output is correct C The test log D A senior tester

Q3

Dijkstra said testing "can show the presence of bugs, but never their absence". What is the practical consequence?

A Only formal proofs should be used. B Tests should use random inputs only. C Since we cannot test every input, tests must be chosen systematically where defects are most likely (partitions, boundaries, risky combinations). D Testing is useless.

Q4

Which is the healthy shape of an automated test suite according to the test pyramid?

A Only manual tests B Many fast unit tests, fewer integration tests, few end-to-end tests C Exactly equal numbers at each level D Mostly end-to-end UI tests, few unit tests

Q5

A post title must have 5 to 100 characters. Which set is the best equivalence-partitioning representatives set?

A 50 only B 2, 50, 150 C 5, 6, 7 D 10, 20, 30

Q6

For the same rule (5–100 characters), which lengths does boundary value analysis add?

A 0, 50, 1000 B 1, 2, 3 C 4, 5, 6, 99, 100, 101 D 5 and 100 only

Q7

A group-size rule: "a study group has 2 to 6 members". A test suite checks sizes 1, 2, 6 and 7. Which fault could it miss?

A Rejecting size 7 B A rule that wrongly rejects size 4 because of a typo size != 4 C Using size > 6 instead of size >= 7 D Using size < 2 instead of size <= 1

Q8

When is a decision table the most useful test-design technique?

A When the output depends on combinations of several conditions B When there is a single numeric input C When testing performance D When there is no specification

Q9

Using the decision table in this chapter: a student who is not the author and not a TA opens a hidden post. What must happen?

A The table does not say. B They cannot view it (rule R3). C The post is deleted. D They can view it.

Q10

A post can be DRAFT, PUBLISHED, HIDDEN or DELETED. Which test is an important invalid-transition test?

A Hide a published post B Publish a draft C Try to restore a DELETED post and check that it is rejected D Create a new draft

Q11

For the badge(posts, endorsed) method in this chapter, the only test is badge(12, true). What are the statement and branch coverage?

A 50% statement, 100% branch B 100% statement, 50% branch C 100% statement, 100% branch D 50% statement, 50% branch

Q12

A module has 100% branch coverage. What can we conclude?

A It needs no more tests. B It meets all requirements. C It has no bugs. D Every branch outcome was executed by some test — but not that results were checked, nor that required behaviour that was never coded exists.

Q13

In assertEquals(26, result); which argument is the expected value in JUnit?

A Neither B Both are treated the same, so it never matters C The second (result) D The first (26)

Q14

How do you test in JUnit 5 that a method throws IllegalArgumentException for a negative score?

A assertThrows(IllegalArgumentException.class, () -> m(-1)); B try { m(-1); } catch (Exception e) { } C System.out.println(m(-1)); D assertTrue(true);

Q15

A test checks "Deadline reminder is sent 24 hours before" by reading the real system clock. It passes in the morning and fails at night. Which FIRST property is violated?

A Repeatable B Fast C Timely D Self-validating

Q16

testB passes only if testA ran first, because testA inserts a student into a static list. Which property is violated, and what is the fix?

A Independent; build fresh fixtures in @BeforeEach and avoid shared static state B Fast; run tests in parallel C Timely; write the tests earlier D Nothing is wrong

Q17

To test the answer cache without calling the real AI provider, the team uses an object that answers with fixed strings and counts how many times it was called. This double is best described as…

A an oracle B a boundary value C a mutant D a spy (a fake that records its use)

Q18

What is the main risk of over-mocking?

A Tests become too slow. B There is no risk. C Mocks cannot be used in Java. D Tests verify the implementation's call sequence rather than behaviour, pass even when results are wrong, and break at every refactoring.

Q19

In TDD, why must the new test be seen failing before writing the code?

A It proves the test can detect the missing behaviour; a test that never fails might test nothing. B It is tradition. C Failing tests make the build faster. D So that the code coverage drops.

Q20

What is the correct order of the TDD cycle?

A Refactor → red → green B Green → red → refactor C Red → green → refactor D Code → test → deploy

Q21

An AI generated this test. What is its problem?

@Test
void testScore() {
    int r = LatePenalty.scoreAfterLatePenalty(85, 7);
    assertEquals(LatePenalty.scoreAfterLatePenalty(85, 7), r);
}

A It is too long. B It uses the wrong assertion method. C Nothing. D The expected value comes from the code itself, so the test can never fail — it has no real oracle.

Q22

Which AI-generated test is weakest?

A assertNotNull(penalty(85, 7)); on a method that returns int B assertEquals(0, penalty(85, 8)); C assertEquals(26, penalty(85, 7)); D assertThrows(IllegalArgumentException.class, () -> penalty(-1, 0));

Q23

You want to know quickly whether an AI-generated test suite is strong. What is a good check?

A Count the tests. B Check that the test names are long. C Run the tests twice. D Introduce small deliberate bugs (e.g., change > to >=) and see whether at least one test fails.

Q24

Which level of testing checks that StudyBuddy's PostService and the real MySQL database work together correctly?

A Integration testing B Acceptance testing C Static analysis D Unit testing

Q25

A test for the reminder feature needs "now" to be 9 October 23:59. What is the best approach?

A Skip the test. B Change the computer's system time before running the tests. C Inject a Clock into the code and pass a stub clock fixed at 9 Oct 23:59 in the test. D Run the test only at that exact time.

Q26

Which is the best way to combine TDD with an AI coding assistant?

A Ask the AI to delete failing tests. B Write the tests yourself from the requirement, see them fail, then let the assistant propose code until they pass — and review it. C Skip tests because the AI is careful. D Let the AI write both the code and the tests, then merge.

Q27 Short answer

Design test cases for this rule using equivalence partitioning and boundary value analysis: "A study group has between 2 and 6 members, and each member must be at least 16 years old." List partitions (valid and invalid) for both inputs and at least 8 test cases with expected results.

Q28 Short answer

Explain why 100% branch coverage does not prove a method correct. Give two concrete examples.

Q29 Short answer

Draw the state-transition model for a StudyBuddy post (states DRAFT, PUBLISHED, HIDDEN, DELETED; events publish, hide, restore, delete) and list tests that cover every valid transition plus two invalid ones.

Q30 Short answer

Explain the difference between a stub, a fake and a mock, with one StudyBuddy example each, and say when a fake is better than a mock.