LAB14 — A defensible fault report

Start CP06, then use a separate FAULT-stale folder. Suggested time: 60 minutes. No physical sensor is involved.

First run .\scripts\replay.cmd fixtures/invalid.csv, preserve exit 2 / line 3 diagnostic, and propose competing failure locations. Copy the fixture and correct only 28C to 28. Run the copied fixture and then the unchanged original. Explain why accepting the corrected number while retaining rejection supports an input-format cause.

Next run the stale-recovery fixture from FAULT-stale. At timestamp 4001, a sample from 1000 must be stale and unavailable. Preserve the bad observation, trace poll and fail, then restore only the disabled expiry guard. Rerun the same fixture and confirm both the expiry boundary and the interrupted recovery at 6000–8000.

Submit a report for each fault with source identity, command, expected/actual result, hypotheses, chosen test, correction, regression evidence, and limitations. Attempt the repaired shared-source firmware build with the recorded target and preserve the actual result. Do not describe an unexecuted check as passing.

Recover by preserving the current folder and extracting the checkpoint into another folder, or use the appropriate already-taught Git operation for committed versus uncommitted changes. Chapter 14 supplies worked diagnoses and explanatory questions.