[MrScripty]

This extends the existing T8 current-version restore witness across three controlled interruptions: after real PostgreSQL population before commit, after committed metadata and a partial inventory of complete copied files, and after the actual final destination rename before its return. Each retry reconciles the same transfer. All 43 raw table states, sequences, files, POSIX modes, immutable coordinator acknowledgement, retained timings and recovered publication Ref remain intact; cold SDK and separate CLI collection launch no compute.

The initial source passed one focused test in 17.615 seconds. After independent review strengthened backup permission checks and sequence-fixture provenance, the final test passed independent reproduction in 17.013 seconds, with zero failures, errors or skips. Exact owned schema/database absence was independently queried, and the task PostgreSQL container is stopped. Raw runners/logs/receipts and review are sealed in the acceptance report.

Published head: 0134a753a1ae54702af8dd0a559c6b0756e20ca1, parent main 8b1f78eb06c17e3dba2a73dc0acc9568e53ccc7e, tree 6fa29f84c442a126eea20c8484cef69855a59a97. The working GitHub app created a normal commit with new metadata; its tree exactly matches reviewed local a28034e, and its test SHA256 remains 50ac2975b216052dc0d07248db9f21a8618f9cf096b22214c2c9a317c302d199, independently reproduced at d02f034. Final report/source review receipt SHA256: 6e48aa5664089703e2a906e226d7a899a691d5e5594aeb37e556701249da8e93. The applicable published-head lifecycle workflow completed successfully at that exact head. Source and report/tree-binding reviews are clear; current main's composed merge tree matches the qualified tree exactly. Main's original post-merge run and PR118 are monitored separately.

This changes only the existing test, plan/ledger and scoped evidence. No production/schema/README/workflow change, benchmark rerun or old-version witness is added. Existing collection measurements and the default individual collection decision remain unchanged.

The scope is controlled exception/retry integrity, not host crash, power loss, torn payload or storage-device durability. The nontrivial sequence high-water mark is authored fixture state with empty job-order rows. This does not claim full M2/M3/M5/T8 or science/performance acceptance. Older project owners perform their own one-time migration; current-version integrity remains required. PUB-1 and SCI-1 still need their genuine original failure and scientific comparison inputs.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

• Configuration used: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: a2685ae2-51da-4398-9e02-d8b493ada4cf

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}