# Debug a Test-Driven C++ Project A supplied C++ repository includes a README and test script. Diagnose and repair the project when tests fail because of possible syntax errors, incomplete functions, or logic defects. Explain how you would use a constrained AI assistant without asking it to produce the answer directly, and leave enough time for a clean submission. ### Constraints & Assumptions - Treat the repository and its tests as the source of truth. - Do not make broad rewrites before identifying the first failure. - The final state must build, pass the relevant tests, and contain only intended changes. ### Clarifying Questions to Ask - What does the README define as expected behavior? - Does the project fail to compile, link, start, or satisfy an assertion? - Which files and tests are in scope, and are hidden tests likely to exercise boundaries? ```hint Fix the first broken boundary Classify the earliest failure before searching later logic; one syntax or build error can hide every downstream test. ``` ### What a Strong Answer Covers - A disciplined read-build-test-inspect loop. - Separation of syntax, incomplete implementation, and behavioral failures. - Small patches plus targeted regression tests and edge-case reasoning. - Diff review, clean build, full test rerun, and submission-time management. ### Follow-up Questions 1. How would you proceed if the visible tests all pass but hidden tests fail? 2. What evidence would justify refactoring instead of making a local repair?