[SamGu-NRX]

fleet-packet: burst17-bs-review-verify-recon

What changed

Codex and Greptile left ten unresolved review findings on PRs #240 and #242, six of them P1, on files in verification/hsverify/ and recon/recon/. This branch works each finding to a written disposition against alpha at 969d4e9 and closes the one test gap the audit found: public_assumption_warnings marks an unreported rule override as unverified for every assumed key, but only battery_width had a test; the existing test now parametrizes over corner_allowance too. Every real fix was already on alpha, so no production code changes here.

Why

Nobody had acted on the ten findings, and some were expected to be stale or wrong. Reading each thread in full and the current code at alpha showed all eight real bugs were fixed by later commits that carry their own regression tests, and two were reviewer misreadings. The disposition table below records the commit or reasoning behind each, so the threads can be resolved from this PR instead of re-derived. Repo access used an existing repo sandbox (cmp_l3e2i2Fu) with a fresh clone in its own directory, per the fleet fallback order.

How it works

#	Finding	Disposition	Fix commit / evidence	Test that covers it
1	codex #242, coverage.py:168 — reject free-space rays blocked by fused occupied voxels	already-fixed-on-alpha	1ed45b65: seen_through now applies vol.blocked with the same stop-short tolerance as seen_by	recon/tests/test_coverage_views.py::test_free_space_rays_respect_fused_obstructions_but_not_the_sample_surface; bite-checked
2	codex #242, coverage.py:406 — measure every vertical band from local ground	already-fixed-on-alpha	ba158ea1: free_space, wall_front and _slabs take the fitted ground and offset each vertical sample by ground.height_at(...)	recon/tests/test_coverage_local_ground.py (3 tests, slope + uphill eave + wall band); bite-checked
3	codex #242, resultcheck.py:1045 — sample every pre-end start in each sweep run	not-a-bug	At review, run_start_count returned pre-end count plus 1 (the endpoint), so range(count - 1) already sampled every pre-end start; c557e3a examples confirm. c32ee975 made the count explicit (range(count)) without changing the samples and cites both threads	verification/tests/test_sweep_sampling_contract.py (yields, failure at a required pre-end start, charge-per-start); bite-checked
4	codex #242, resultcheck.py:1045 — sample the first start before stepping through sweep runs	not-a-bug	Same misreading as #3; the sub-step run [0, 0.4] step 1 samples [0, 0.4], both endpoints included	verification/tests/test_sweep_sampling_contract.py::test_pre_end_count_plus_endpoint_matches_every_yield; bite-checked
5	codex #240, resultcheck.py:1145 — compare reported battery width with the active rules	already-fixed-on-alpha	a7b3169f + db1e4c2f: public_dimension_problems compares every reported spot width and battery.width_ft threshold with rules.width_ft whenever the response's policy does not identify a different policy	verification/tests/test_policy_admission.py::test_public_policy_width_is_checked_without_case_assumptions, test_matching_public_policy_keeps_loaded_dimensions; bite-checked
6	codex #240, resultcheck.py:1893 — verify reported dimensions against matching public rules	already-fixed-on-alpha	db1e4c2f: reported_rules overrides the loaded rules only when the reported fingerprint/private source names a different policy; matching-public responses are validated against the loaded rules	verification/tests/test_policy_admission.py (test_matching_public_policy_keeps_loaded_dimensions, test_missing_policy_cannot_override_loaded_dimensions, test_private_marker_cannot_override_matching_fingerprint); bite-checked
7	codex #240, depth.py:226 — recompute caches that claim encryption	already-fixed-on-alpha	a7b3169f (carried by daa36a24): the repair block catches RuntimeError (damaged ZIP encryption flag) and NotImplementedError (unsupported compression) and deletes the cache	recon/tests/test_review_debt.py::test_corrupt_moge_cache_is_recomputed (encrypted-flag and corrupt-compressed params); bite-checked
8	codex #240, e2e.py:169 — accept scene.json inside a single top-level folder	already-fixed-on-alpha	717e6da8 + 67b10e56: read_bundle mirrors the API's wrapped layout and refuses competing scene.json paths	verification/tests/test_e2e.py::test_bundle_reads_one_wrapped_capture_and_ignores_macos_metadata, test_bundle_accepts_public_api_finder_fixture_shape, test_bundle_rejects_root_and_wrapped_scene_together; bite-checked
9	codex #242, e2e.py:603 — retain rule-independent property checks	already-fixed-on-alpha	ba158ea1: the properties check always runs and passes rule_checks=not warnings; identical-input, mirrored, coverage and requests-captured resends stay on, only rule-consuming transforms are withheld	verification/tests/test_unreported_rule_properties.py::test_unreported_width_keeps_identical_input_comparison; bite-checked
10	codex #242, resultcheck.py:1593 — mark unreported private overrides as unverified	already-fixed-on-alpha	a7b3169f + db1e4c2f: the warning fires whenever the response used non-public rules, not only when the assumed value differs from the public default	verification/tests/test_reported_rules.py::test_unreported_override_is_unverified_even_when_case_matches_public — widened in this PR to corner_allowance; bite-checked before and after widening

Bite-checked means: the cited fix's hunk was reverted in a scratch working tree at alpha, the named test failed, and the tree was restored clean. All eight reverts failed their tests as expected.

UI changes

N/A — test-only change, nothing visible.

Verification

Both suites were run twice: first on pristine alpha (969d4e93, the baseline), then on this branch's head (c542eccc). make test in each suite runs ruff check, ruff format --check and pytest -q. There is no separate typecheck for these suites; ruff is the repository's static check.

Suite	alpha baseline	this branch
verification	594 passed, lint clean	596 passed, lint clean (+2 new params)
recon	168 passed, lint clean	168 passed, lint clean (recon untouched)

Test logs

$ cd verification && make test          # on pristine alpha 969d4e93
uv run ruff check .
All checks passed!
uv run ruff format --check .
46 files already formatted
uv run pytest -q
......................................................................  [ 12%]
594 passed, 2 warnings in 36.64s

$ cd recon && make test                 # on pristine alpha 969d4e93
uv run ruff check .
All checks passed!
uv run ruff format --check .
46 files already formatted
uv run pytest -q
............................................................           [100%]
168 passed, 49 warnings in 7.71s

$ cd verification && make test          # on this branch c542eccc
All checks passed!
46 files already formatted
596 passed, 2 warnings in 33.52s

$ uv run pytest -q tests/test_reported_rules.py::test_unreported_override_is_unverified_even_when_case_matches_public
....                                                                     [100%]
4 passed in 0.12s

Bite checks (fix reverted, named test run, tree restored clean each time):
M1 seen_through without vol.blocked            -> FAILED-AS-EXPECTED (test_coverage_views)
M2 _slabs without local-ground offset          -> FAILED-AS-EXPECTED (test_coverage_local_ground)
M3 run_starts dropping the last pre-end start  -> FAILED-AS-EXPECTED (test_sweep_sampling_contract)
M4 public-dimension admission skipped          -> FAILED-AS-EXPECTED (test_policy_admission)
M5 nonpublic unreported override silent        -> FAILED-AS-EXPECTED (test_reported_rules)
M6 cache repair missing RuntimeError           -> FAILED-AS-EXPECTED (test_review_debt)
M7 wrapped capture refused                     -> FAILED-AS-EXPECTED (test_e2e -k wrapped or finder)
M8 property checks skipped when unverified     -> FAILED-AS-EXPECTED (test_unreported_rule_properties)

Risk

Low. The only code change is a parametrized test; production behavior is unchanged. The dispositions rest on cited commits and revert experiments that anyone can rerun: reverting any commit in the table should fail the named test.

Checklist

• The scope is focused; related changes belong together
• I explained what changed and why
• N/A — no before/after images: nothing visible changed
• N/A — no video: no motion or interaction change

---

Opened by the maintainer from the Obvious thread's own description, so every review bot reviews it (CodeRabbit skips PRs authored by bots). The work and its commits are by Obvious.

Co-authored-by: obvious-autobuild[bot] <262744130+obvious-autobuild[bot]@users.noreply.github.com>

Summary by CodeRabbit

• Tests
  • Expanded coverage for override warnings across battery_width and corner_allowance assumptions, including private sources and fingerprints that differ from public rules.

[SamGu-NRX]

@codex review

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-17T23:46:52.576879Z	c542ecc	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: d96cbd66-4ebc-4fc1-b2aa-391e98a187e4

📥 Commits

Reviewing files that changed from the base of the PR and between 969d4e9 and c542ecc.

📒 Files selected for processing: 1

• verification/tests/test_reported_rules.py

Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.

📝 Walkthrough

The unreported-override test now checks battery_width and corner_allowance assumptions against two policy identity cases. It continues to expect one warning containing the public-rules override message.

Changes: Reported rules test coverage

Layer / File(s)	Summary
Expand override test cases verification/tests/test_reported_rules.py	The test checks both assumptions against a private source or a fingerprint that differs from the public rules. It continues to expect one public-rules override warning.

Priority: ⬇️ Low

Estimated code review effort: 1 (Trivial) | ~3 minutes

Change: Other

Merge Risk: ⚪ Minimal · up to c542e

The added test cases protect the unverified-warning behavior for corner_allowance under both nonpublic policy identities. No actionable merge risk remains.

Pre-merge checks: 4 · 1

Failed checks: 1 warning

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 1 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

Passed checks: 4 passed

Check name	Status	Explanation
Title check		The title clearly identifies the verification test change and the specific corner allowance warning behavior.
Description check		The description provides complete sections for the change, rationale, implementation details, UI impact, verification results, risk, and checklist. It also records the test-only scope and completed te…
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Full details: Description check

Explanation

The description provides complete sections for the change, rationale, implementation details, UI impact, verification results, risk, and checklist. It also records the test-only scope and completed test results.

• Fix all pre-merge checks with AI

✨ Finishing Touches: 💡 1

📝 Generate docstrings: 💡

• Commit to this branch
• Create a new PR

• 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.}

[codemetrics-ai]

Note

Review skipped

Codemetrics is installed on this repository, but it hasn't been added to your team yet.

To enable AI code reviews for this repo, go to your dashboard and add BaseScanning to your team's repositories.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Swish!

Reviewed commit: c542eccc08

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[greptile-apps]

Confidence Score: 5/5

[Low impact] This test-only change appears safe to merge.

Summary

Expands the existing warning test to cover corner_allowance alongside battery_width.

• The override test covers corner allowance as well as battery width.

_{Reviews (1) · Last reviewed commit: "test(verification): pin unverified-overr..." · Reviewed by Greptile}

[SamGu-NRX]

@coderabbitai review

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.