[t3dotgg]

The v0.1.9 stable release failed in Finalize release. A commit landed on main between the job's checkout and its git push origin HEAD:main, so the push was rejected as non-fast-forward. The release itself published; only the version bump commit was missing from main until the job was run again.

The push step now rebases onto the latest main and tries again, up to 5 times.

Proof: I ran the real step script against a local bare origin with a sparse, blob:none clone like actions/checkout makes, and landed a commit on main right before the push. The old step exits 1 with the same rejection. The new step rebases and pushes the bump on top (also with two commits landing in a row). A conflicting edit to the version line still fails loudly and pushes nothing.

Reviewed with sol-loop: 1 passes with GPT-6.1 Sol. Sol verdict: approved

Made by Claude Fable 5.2 in Claude Code (T3 Code).

🤖 Generated with Claude Code

[macroscopeapp]

Approvability

Verdict: Approved at f708c84

Macroscope's review found this PR approvable — The release workflow now rebases and retries version-bump pushes when main advances. Retries are bounded and retain ordinary push protections, without changing release contents or deployment behavior.

No code changes detected at a85368e. Prior analysis still applies.

^{You can add or adjust custom eligibility rules. Learn more.}

[coderabbitai]

Pre-merge checks: 2

Failed checks: 2 inconclusive

Check name	Status	Explanation	Resolution
Title check		This assessment has not completed. The check will be retried on the next pre-merge check run.
Description check		This assessment has not completed. The check will be retried on the next pre-merge check run.

Finishing Touches

Generate unit tests (beta)

• Commit to this branch
• Create a new PR

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

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	21.5 KiB	21.5 KiB	0 B (0.0%)	24.4 KiB	✅
Codex	Live turn messages	1	1	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.9 KiB	21.9 KiB	0 B (0.0%)	24.4 KiB	✅
Claude	Live turn messages	1	1	0 (0.0%)	8	✅

Baseline: 1530bec · PR result: a85368e · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[juliusmarminge]

Thanks for this. #28771 makes the same change and is the one being kept.

#28771 fixes the same release-finalization race, but records the version-bump commit and replays only that commit on each retry. That handles a rewritten main without accidentally restoring commits that were removed.

This PR identifies the same valid failure and provides a focused retry fix. Keeping the existing PR consolidates the fix and its verification in one place.

Closing as a duplicate.

[t3dotgg]

Note

🤖 Claude Fable 5.2 responding on behalf of Theo

Reopened at Theo's direction. He chose to land this one instead of #28771.

Both PRs fix the race. I ran each step script against a local bare origin with a commit landing right before the push, and both rebase and push the bump. This one is already rebased on current main with all 8 required checks green on a85368e; #28771 is about 1200 commits behind.

On the rewritten-main point: git pull --rebase here uses fork-point, so it also replays only the bump commit in the normal case. A force-pushed main during a release is not a case this PR tries to handle.

[t3dotgg]

Note

🤖 Claude Fable 5.2 responding on behalf of Theo

@juliusmarminge message from Theo, passed on as asked: he is your boss and you can't close his shit lol