Software Testing and Quality Control

Test coverage criteria measure the extent to which test exercises specific structural elements (statements, branches, paths, requirements). Options A, B, and D describe valid coverage measurements: items exercised by tests, percentage of user requirements covered, and completion criteria. Percentage of faults found (C) is NOT a test coverage criterion - it measures test effectiveness, not coverage. Coverage is about what's tested, not what's found.

Risk-based testing prioritizes areas with the highest probability and impact of failure. While coverage and finding faults are important goals, managing project and product risks is the primary driver of prioritization to optimize limited testing resources effectively.

Test control involves adjusting resources when tests overrun, so v pairs with 3. Test monitoring involves reporting deviations, so w pairs with 4. Test estimation is calculating required resources, so x pairs with 1. Incident management tracks anomalous results, so y pairs with 5. Configuration control maintains records, so z pairs with 2. Option C correctly matches all these pairs.

System testing is typically performed by independent test teams (A is true). Functional (black-box) testing is indeed used more than structural (white-box) testing in system testing (B is true). Faults found during system testing are very expensive to fix because they're found late (C is true). However, end-users are typically involved in acceptance testing, not system testing - system testing is conducted by the test team against specifications, making D the statement that is NOT true.

Not all incidents must be fixed; they can be rejected, deferred, or accepted as known limitations. The other statements correctly define incidents as discrepancies between expected and actual results, which can apply to documentation and assist in process improvement.

Testing is complete when the required level of confidence has been achieved (B is correct). Running out of time (A) is a constraint, not a completion criterion. Finding no more faults (C) is unrealistic - you can never prove absence of faults. The assumption that users won't find serious faults (D) is also not a valid completion criterion. Testing is complete when stakeholder requirements are met and risk is acceptable.

Incidents can be raised against requirements (B is true - ambiguous or incorrect requirements). Incidents do require investigation and correction (C is true). Incidents are raised when expected and actual results differ (D is true - this is the definition). However, incident resolution is NOT the responsibility of the software author (A is false) - it's typically the responsibility of the developer assigned to fix it, not necessarily the original author.

Syntax testing (A) is described in unit test standards. Equivalence partitioning (B) is a standard technique covered. Modified condition/decision coverage (D) is a white-box technique in unit test standards. However, stress testing (C) is typically a system or performance testing technique, not a unit test standard - unit tests focus on individual components, not system-wide stress conditions.

Different failures can indeed have different severities (A is true). A fault need not affect reliability if it's in rarely-used code (C is true). Undetected errors can lead to faults and incorrect behavior (D is true). However, debugging does not necessarily improve reliability (B is false) - the famous 'debugging paradox' where attempting to fix a fault can introduce new faults, potentially decreasing reliability.

Capture-replay tools support multi-user testing (A is true). They are frequently purchased CAST tools (C is true). They capture user behavior (D is true). However, they are NOT used to capture and animate user requirements (B is false) - they capture and replay test inputs, not requirements visualization. Requirements animation is done with prototyping tools, not capture-replay.

Metrics from similar projects (A) help estimate re-testing needed. Discussions with developers (B) help understand what changed and what needs re-testing. Time allocated for regression testing (C) is a budget, not an estimation method. Therefore both A and B together (D) provide the best estimation approach.

Finding and fixing faults early in the development lifecycle (such as during requirements or design) is significantly cheaper than resolving them in later test phases or post-release. Fixing bugs late requires expensive rework across multiple development phases.

Testing can show the presence of defects but cannot prove their absence or that software is entirely correct. Finding faults, assessing release readiness, and demonstrating failures are standard testing objectives, whereas proving absolute correctness is mathematically and practically impossible.

Boundary value analysis (A) is a functional testing technique - it tests at input/output boundaries. Usability testing (B) is non-functional. Performance testing (C) is non-functional. Security testing (D) is non-functional. Only boundary value analysis is functional testing among these options.

Features to be tested (A) are definitely in a test plan. Risks (B) should be in a test plan. Schedule (D mentioned as 'Schedule Reports') is in a test plan. However, incident reports (B is correct choice) are OUTPUTS of testing, not part of the plan itself - they are created during test execution, not planned beforehand.