fix(geometry): carry the no-void probe reading through the census reclassification list (#3366)

[BIMvoice]

Summary

• Refs Census: a no-void probe that starts failing reads as an improvement and the lane stays green #3366: HostRow::is_torn_solid reads open, rep, far, and pre (the no-void probe reading). reclassifications — the list of is_torn_solid inputs that move the gated defect population without any count moving — carried rep and far, but not pre.
• When a host's no-void probe starts or stops failing, pre flips Open(n) <-> Failed while open, tris, collapsed and alt all hold, so only the derived is_torn_solid predicate saw it move. A probe going dark landed in classify's "no longer a genuine watertightness defect" arm with nothing else to attribute it to, so the host filed under improved, requires_bless() stayed false, and the lane stayed green while the census lost the ability to distinguish "arrived torn" from "the boolean tore it" for that host.
• Fix: carry pre in reclassifications, per the issue's suggested fix. Both directions of a no-void flip now route into changed, which requires a bless. The probe-recovers direction changes label too (previously regressed, now changed) — both are red and both require a bless, so this does not weaken the gate; it updates the pre-existing test (a_no_void_pass_that_stops_failing_makes_the_host_a_gated_defect, renamed) that encoded the old label.
• Grepped for the same shape (a symmetric better/worse classify pair over a derived predicate) elsewhere in the repo's census/benchmark comparators; rust/geometry/tests/census_golden/mod.rs is the only occurrence.

Test plan

• RED: added a_probe_that_starts_failing_is_a_reclassification_not_an_improvement, confirmed it fails against the pre-fix code (d.improved non-empty, requires_bless() == false).
• GREEN after carrying pre in reclassifications.
• Both directions asserted: probe going dark AND probe starting to run both require a bless (d.changed, not d.improved/d.regressed).
• Updated the pre-existing test that had encoded the old (buggy) expectation for the probe-recovers direction.
• Full census_golden suite green: cargo test -p ifc-lite-geometry --test triangulation_invariance --features triangulation-alt census_golden — 19 passed.
• cargo clippy -p ifc-lite-geometry --tests --features triangulation-alt clean.
• Test-only change in rust/geometry/tests/; no changeset (matches this repo's convention for test-only fixes, e.g. e2c67f0).

https://claude.ai/code/session_01QPHChk3Ve9N519A4kY7436

Summary by CodeRabbit

• Bug Fixes

  • Improved handling of no-void probe failures and recoveries by classifying them as population reclassifications.
  • Transition reports now provide clearer reasons when these changes occur.
  • Both failure and recovery transitions now require explicit approval instead of passing silently.

• Tests

  • Added coverage to verify correct classification and approval requirements in both directions.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 354e1674-77a4-406f-8785-caf83c17787e

📥 Commits

Reviewing files that changed from the base of the PR and between 5a431e5 and 8e5c385.

📒 Files selected for processing (1)

• rust/geometry/tests/census_golden/mod.rs

Included review availability: Your plan provides up to 2 included reviews per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The census golden tests now classify no-void probe failures and recoveries as PreVoid reclassifications. Both transition directions appear as changes and require blessing.

Changes

PreVoid reclassification

Layer / File(s)	Summary
Record PreVoid transitions rust/geometry/tests/census_golden/mod.rs	Reclassification reporting now includes HostRow.pre changes and documents their effect on the gated defect population.
Validate transition blessing rust/geometry/tests/census_golden/mod.rs	Tests cover Open-to-Failed and Failed-to-Open transitions. Both are classified as changed, not improved or regressed, and require blessing.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: ⚪ Minimal · up to 8e5c3

This localized test-only change corrects census handling for no-void probe state transitions, with both directions covered and the relevant test suite passing; no actionable merge-blocking risk remains beyond normal checks and review.

Suggested reviewers: louistrue

Poem

A rabbit checks the probe at dawn

From Open to Failed, the fear is gone
From Failed to Open, records align
Both changes need a blessing sign
The census hops along just fine 🐇

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the geometry fix and the no-void probe change in the census reclassification list. It accurately reflects the main change.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 4 functions across 1 files.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

---

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

[github-actions]

Viewer benchmark

✅ No threshold regressions detected.

01_Snowdon_Towers_Sample_Structural(1).ifc

Baseline recorded 2026-07-01T20:31:05.538Z on github-actions ubuntu-latest, viewer-benchmark-ci (headless Chrome, SwiftShader ANGLE), production build.

Metric	Current	Baseline	Delta	Threshold	Status
firstBatchWaitMs	2068ms	2905ms	-28.8%	+50%	✅
firstVisibleGeometryMs	3178ms	3652ms	-13.0%	+50%	✅
streamCompleteMs	4059ms	3598ms	+12.8%	+50%	✅
spatialReadyMs	1411ms	1032ms	+36.7%	+50%	✅
metadataCompleteMs	2128ms	3063ms	-30.5%	+50%	✅
totalWallClockMs	4200ms	3700ms	+13.5%	+50%	✅

AC20-FZK-Haus.ifc

Baseline recorded 2026-07-01T20:30:59.972Z on github-actions ubuntu-latest, viewer-benchmark-ci (headless Chrome, SwiftShader ANGLE), production build.

Metric	Current	Baseline	Delta	Threshold	Status
firstBatchWaitMs	313ms	1075ms	-70.9%	+50%	✅
firstVisibleGeometryMs	2054ms	1572ms	+30.7%	+50%	✅
streamCompleteMs	1091ms	1980ms	-44.9%	+50%	✅
spatialReadyMs	1126ms	915ms	+23.1%	+50%	✅
metadataCompleteMs	1206ms	1392ms	-13.4%	+50%	✅
totalWallClockMs	2100ms	3300ms	-36.4%	+50%	✅

Refresh the baseline from a CI run: dispatch the Benchmark workflow with record_baseline, download the benchmark-baseline artifact, and commit baseline.json (see tests/benchmark/README.md).

[louistrue]

@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.