Python ETLHardofficialBenchmark eligiblepro
Checkpoint after commit
The job crashed mid-batch. checkpoint.json is corrupt or points past uncommitted work. Detect corruption, resume from the last good hash, write a new checkpoint only after COMMIT.
| Field | Detail |
|---|---|
| Creator | PipelineArena |
| Version | 2.0.0 |
| Time | 55 min |
| Scored checks | 17 |
Objective
No duplicate committed batches; no skipped uncommitted batches; final counts match.
| # | Success condition |
|---|---|
| 1 | Corrupt checkpoint detected |
| 2 | Committed batches skipped |
| 3 | Uncommitted batches replayed |
| 4 | New checkpoint written after commit only |
| # | Visible instruction |
|---|---|
| 1 | Validate checkpoint.json (hash + batch id). |
| 2 | Replay from last good commit. |
| 3 | fsync/write checkpoint after COMMIT. |
Scoring checks the leftover tables or files, replay safety, and seeded variations — not the exact syntax you used. Check implementation is never shown to participants.
Signed-in workspace: Open in app