Run a Fail-Then-Pass Replay Before an OSS Fix

Описание и применение

Создайте воспроизводимый «fail‑then‑pass» шлюз, чтобы гарантировать, что исправление действительно исправляет исходную ошибку, и только после этого проходит тест.

Задача

Maintainers cannot verify that a patch fixes a bug if the failure is not reproducible locally, leading to speculative diffs and unreliable reviews.

Результат

The workflow creates a reproducible gate that ties a patch to a concrete failing command and a passing outcome, improving review reliability and reducing speculative changes.

Рабочий процесс

• 1. Isolate a worktree at the issue tag and record the SHA.
• 2. Write a regression test that reproduces the failure and captures the fail needle.
• 3. Run the test to prove the failure and capture the fail log.
• 4. Apply the minimal patch to the production code.
• 5. Re-run the test to produce a pass log and verify the pass needle while ensuring the fail needle is absent.
• 6. Automate the gate with a script that checks both logs for the required needles.

Доказательства

This workflow pins one issue to one command.

Источник

The artifact is a fail-then-pass replay gate.

Источник