CCAR-F Context and Reliability Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Request the entire migration in a single response to preserve the model's global understanding of the module.

  Why it's wrong here

  Very large single responses strain the output limit and increase the chance of truncation or silent omission, and they are harder to validate piecewise. Long monolithic generations also make error isolation difficult. For verifiable conversion, decomposing the module into analyzable units with per-unit checks yields stronger guarantees than one giant response, even if global context feels appealing.

• ✓

  Add an automated test harness that compiles and executes the generated code against representative legacy inputs and compares outputs.

  Why this is correct

  Compiling and running generated code against known inputs converts the migration from a text task into an empirically verified one. Behavioral equivalence with the legacy module is the strongest available signal of correctness. This practice catches semantic errors that schema validation alone cannot, making it essential for a pipeline where stakeholders demand verifiable, high-reliability conversion results.

• ✗

  Maximize the model's creativity by raising temperature so it can find novel modernization opportunities.

  Why it's wrong here

  Creativity is counterproductive when the goal is faithful behavioral equivalence. Higher temperature introduces variation that can alter control flow or business rules, producing output that compiles yet behaves differently. The stakeholders want verifiable migration, not reinvention, so loosening sampling undermines both reliability and the ability to validate results consistently across runs.

• ✗

  Allow the model to omit edge-case handling when the legacy code is ambiguous, to keep outputs concise.

  Why it's wrong here

  Silently dropping edge-case handling hides risk exactly where migration is most dangerous, and it defeats verifiability because the omission may not be visible in the output. Ambiguity should be surfaced explicitly for human decision, not discarded for brevity. Concise output is worthless if behavioral equivalence is broken, so this practice conflicts with the reliability requirement.

• ✓

  Define a fixed output schema for migrated code plus a structured migration report, and validate every response against that schema.

  Why this is correct

  A fixed schema makes each response machine-checkable, so malformed or incomplete migrations are caught automatically rather than trusted blindly. The migration report gives reviewers a structured account of assumptions and mappings. For a reliability-critical conversion pipeline, this turns subjective review into deterministic validation, which is precisely what verifiable output requires in this scenario.