[avrabe]

Refs #1622

This commits the one-round cold review of the v0.89.1 candidate 69de522a (docs/reviews/v0.89.1-cold-review.md) unedited, with the coordinator disposition appended. It also corrects the false statements the review found. This PR's merge commit is the intended v0.89.1 release commit.

Corrections (each re-derived in the disposition, commands and output there)

Finding	Site	Before	After
F1	artifacts/release-v0.89.1/RQ-89-PATCHVER.yaml landed	"Tooling only: no file under crates/ changes."	The PATCHVER commit (8ac0a48) changes no file under crates/. The version-bump commit squashed into the same PR (019f047) changes only the path-dependency version = pins in 10 crates/*/Cargo.toml files (27 pin lines, 0.89.0 to 0.89.1) and no .rs file.
F2	artifacts/release-v0.89.1/RQ-89-ACTIONPIN.yaml landed	"git diff --stat v0.89.0 origin/main -- crates/ lists only the #1623 files"	None of the four configuration commits changes a file under crates/ (per-commit diff empty for 6cb4887, 8002a8d, df847a1, 89f4c37). On the shipped tree git diff --stat v0.89.0 69de522a -- crates/ lists 14 files: the 4 of #1623 and the 10 Cargo.toml pins of the bump.
F3	CHANGELOG.md [0.89.1], CI configuration bullet	"so every commit after #1625 carried a failing .github/dependabot.yml check and no grouped update applied"	"and no grouped update applied. Its .github/dependabot.yml check failed on both commits that wrote that config (#1625 and #1627); #1610, merged between them, does not touch the file and carries no Dependabot check."
F4	PR #1624 body line 5, "It has two commits" (it has 3)	—	Recorded, not corrected: no existing PR body is edited.
Prose vs fields	RQ-89-ACTIONPIN description, verified-by	#1627 "scopes the group by name shape"; Dependabot acceptance "NOT verified"	The description says #1627's ^[^@]+/ scoping still did not parse, and that what ships is #1624's third commit (89f4c37, from #1628): each group under its own updates entry, github-actions-all ("*") under github-actions. verified-by records the check runs: failure at 8002a8d and df847a1, success at 69de522. The done-when's second clause is met and its first is not (two tag refs), so status: proposed / disposition: partial stay.

F3 re-derivation, gh api repos/pulseengine/synth/commits/<sha>/check-runs filtered to app.slug == "dependabot":

commit	PR	touches dependabot.yml	dependabot check
8002a8d	#1625	yes	completed/failure
cc15a34	#1610	no	none (0 of 74 check runs)
df847a1	#1627	yes	completed/failure
69de522	#1624	yes	completed/success

Gates, run on the committed head

• claim_check.py claims.yaml: rc 0, 76/76 claims hold.
• status_evidence_check.py: rc 0. Anchor v0.89.0 — 134 delivery commits (pinned 134), 687 artifacts (pinned 687), lag 0; window 5 delivery-shaped commits since v0.89.0 — 5 attributed; issue-scope: v0.89.1 over 4 artifact(s) … (authorised: #1622).
• verdict_prose_check.py: rc 0, 199 classified, 0 disagree.
• issue_closure_check.py --release v0.89.1 --since-tag v0.89.0 --allow-unclosed: rc 0, 1 authorised, 0 held open, 18 closed, 0 failures. The one authorised issue is multi-memory phase 2: i64 load/store on a non-default memory under --native-pointer-abi (gale's supervisor domain has 12) #1622, still OPEN.
• release_file_statistics_scan.py --release v0.89.1: rc 0, 0 digit hits, 0 word-number hits.
• /usr/bin/python3 (3.9) check_py_floor.py --self-test: rc 0. Over git ls-files '*.py': rc 0, 311 file(s) under Python 3.9, 0 finding(s).
• loop_conformance_check.py v0.89.1: rc 1, slots=7 derived=5 attested=1 failures=1 verdict=DOES-NOT-CONFORM. The one failure is step 5's "mcdc ran on commit", which is NOT-DERIVED until CI has run on the pushed head. Every step:
  • step 0, release identity: DERIVED-PASS (workspace version 0.89.1).
  • step 1-2, spar AADL->WIT: NA-FILED (RQ-89-ARCHMODEL9, Recurring N/A: feature-loop steps 1-2 (spar AADL -> WIT) have been skipped every release and never filed — #912's rule applied to itself #1136, proposed).
  • step 3: DERIVED-PASS ×5 (artifact set 4/4 done-when declared; reported outcome 4/4; verdict classified 3 of 3 openings; status-evidence gate wired; live run exit 0).
  • step 4: DERIVED-PASS ×2 (wiring with --exact-emulation-floor 394603; live run exit 0).
  • step 5: mcdc gate DERIVED-PASS; mcdc ran on commit NOT-DERIVED (check-run on a0bb771 not yet available).
  • step 6: DERIVED-PASS ×2 (signing workflow; Signing E2E pre-tag form, latest main run success 2026-11-02T05:18:45Z).
  • step 7, clean-room review: DERIVED-PASS. docs/reviews/v0.89.1-cold-review.md declares reviewed commit 69de522abbc7, which exists and is an ancestor of HEAD.
  • step 8: pin sweep DERIVED-PASS; changelog structure DERIVED-PASS; PR-head-vs-merge ATTESTED (the review measures git diff 89f4c37c 69de522a at 0 lines); crates and npm NA-BY-MOMENT.
• CI sweep (each python3 scripts/… line in ci.yml, run under bash -o pipefail in this worktree, no synth binary built): 137 commands, 71 non-zero. All are environmental: 51 oracle_evidence.py with no evidence file; 11 oracle runs and determinism_check that need ./target/debug/synth, which this worktree has not built; mcdc_gate, check_varve_shadow_pins, 3 elf_static_check and ci_mutants_gate lack their build inputs; replay_control_1318; and mutation_survey.py ci timed out at 300 s. None of these reads a file this PR changes. The gates this PR can move (claim, status-evidence, closure) are green above. Afterwards pgrep -fl mutation_survey was empty, git diff was empty (no mutant left behind), and the one untracked output (mutants-ci.txt) was removed.

Squash simulation

The commit subject and this PR's title are byte-identical: v0.89.1 cold review: record the review and correct the statements it found. It does not match status_evidence_check.LANE_SUBJECT (^v\d+\.(\d+)(?:\.\d+)? ([A-Z][A-Z0-9]*)…, and "cold" is lowercase), it has no conventional-commit type, so it is not in DELIVERY_TYPES, and neither the subject nor the commit body carries an RQ- id. The status_evidence_check window line reads the same with this commit on top as on 69de522: "5 delivery-shaped commits since v0.89.0 — 5 attributed".

Record byte facts (measured on the committed blob, git show HEAD:docs/reviews/v0.89.1-cold-review.md)

• Total: 50480 bytes.
• The first 37941 bytes have sha256 5eb44bdd105494bf959a9c58382d525a26a2f122fdf02fc3b5b57fd836842d9d, the frozen review's.
• Removed lines against the frozen review: 0.

Nothing is posted to any issue. Not merged by this lane.

🤖 Generated with Claude Code

https://claude.ai/code/session_01YJK5LZZEkV5smCY1jKn18L

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!