fix(#319): Intf_Interference O(1) tangent-zone + checkpoint breaker — v1.15.1

CLAUDE.md

@@ -116,6 +116,7 @@ suite into these targets (each `Tests/OCCT<Domain>Tests/`, declared in `Package.
 - `ShapeFix_Face::FixPeriodicDegenerated` (invoked from `Shape.face(from:boundary:)`/`face(from:boundary:innerWires:)` and the `FaceFixer` API whenever a face's sole boundary wire is a single closed edge belting a `Surface.cone`'s full period, apex outside the wire's V range — e.g. a rivet/boss-rim seam fit as one periodic curve) used to SIGSEGV (uncatchable) on the ordinary `ShapeFix_Face fixer(face); fixer.Perform();` construction — a null-Context dereference at the function's final `Context()->Replace(myFace, myResult)`, the only one of twelve such call sites in the file missing the `if (!Context().IsNull())` guard every sibling uses. A standalone circle repro is negative (`wireFromEdges` alone never reaches `FixPeriodicDegenerated`); the minimal repro needs the wire trimmed to a periodic conical surface via `face(from:boundary:)`. **Fixed in v1.12.7**: bridge now calls `fixer.SetContext(new ShapeBuild_ReShape)` before `Perform()` at all three `ShapeFix_Face` call sites (immediate fix, any xcframework); kernel patch also carried (`Scripts/patches/0005-*`, xcframework rebuilt) and filed upstream as [OCCT#1378](https://github.com/Open-Cascade-SAS/OCCT/issues/1378) (repro) / [OCCT#1380](https://github.com/Open-Cascade-SAS/OCCT/pull/1380) (fix). #317.
 - `BRepGProp_EdgeTool::IntegrationOrder` (invoked from `BRepGProp::LinearProperties`, which backs `Shape.analyze(tolerance:)`'s small-edge scan) used to SIGSEGV (uncatchable) on an edge whose sole geometry is a Bezier/BSpline-type curve-on-surface pcurve (no 3D curve) — the common shape of a degenerate edge `BRepBuilderAPI_Sewing` produces reconciling near-coincident vertices between two faces that don't share an edge outright (surfaced sewing two real mesh-derived candidate faces from `kof_ii_engine_cover.stl`). `IntegrationOrder` correctly reads the pcurve's type via `BAC.GetType()` (the curve-on-surface-aware virtual dispatch) but then re-derives the pole count by hand from `BAC.Curve().Curve()` — a completely different, non-virtual accessor that's null whenever there's no 3D curve; down-casting that null handle and calling `->NbPoles()` on it crashes. A from-scratch synthetic degenerate edge (`BRep_Builder` + a hand-built `Geom2d_BSplineCurve` pcurve on a plane, no 3D curve) reproduces the identical crash trace — no real fixture needed to pin the mechanism. **Fixed in v1.12.8**: bridge's small-edge scan now skips degenerate edges outright (`OCCTShapeAnalyze`, immediate fix, any xcframework — a degenerate edge's zero 3D extent isn't a "small edge" defect to flag); kernel patch also carried (`Scripts/patches/0006-*`, xcframework rebuilt): `IntegrationOrder` now calls the adaptor's own (correctly-dispatching) `BAC.NbPoles()` instead. Filed upstream as [OCCT#1381](https://github.com/Open-Cascade-SAS/OCCT/issues/1381) (repro) / [OCCT#1382](https://github.com/Open-Cascade-SAS/OCCT/pull/1382) (fix). #318.
 - Three more upstream crash/hang fixes carried proactively (audit of OCCT PRs since our p1 baseline, not discovered via our own crashes — #323, v1.12.9): (1) `ShapeAnalysis_FreeBounds::connectWiresToWiresImpl` (the same helper as the #310 fix above) left a stale `lwire` index when a skipped-loop candidate wire had zero edges (e.g. a wire wrapping a single internal-orientation edge), so the outer loop's termination check never fired and it read invalid memory — `Scripts/patches/0007-*`, backports open third-party [OCCT#1331](https://github.com/Open-Cascade-SAS/OCCT/pull/1331) (fixes OCCT#1330, the "still open" sibling bug noted above). (2) `Geom_BSplineCurve::PeriodicNormalization` used an O(N) `while`-loop to bring an out-of-range parameter back into a periodic curve's range, and could infinite-loop outright once the parameter's magnitude vastly exceeded the period (floating-point no-op on `Parameter -= Period`) — hung `BRepAlgoAPI_Section` on cylindrical shapes; `Scripts/patches/0008-*`, backports merged [OCCT#1329](https://github.com/Open-Cascade-SAS/OCCT/pull/1329) (rewritten to O(1)). (3) `StepData_StepWriter::AddString` looped forever writing a single unbroken raw string longer than the 72-char line buffer (`StepLong`) — no amount of flushing ever made room; `Scripts/patches/0009-*`, backports open [OCCT#1318](https://github.com/Open-Cascade-SAS/OCCT/pull/1318) (splits the token across lines instead), regression test `STEPWriterOversizedNameTests` (`OCCTIOTests`) via `Shape.writeSTEP(to:name:)` with a >72-char name.
+- `BOPAlgo_ArgumentAnalyzer`'s self-interference phase (backs `Shape.isSelfIntersecting(hardTimeout:)`) could run unboundedly past its `hardTimeout:` deadline on a pathological artifact — 619s+ CPU against a 30s deadline, never returning. Two compounding causes: (1) `Intf_Interference::Insert` called `Intf_TangentZone::GetPoint(Index)` inside a nested comparison loop; `GetPoint` is O(n) per call (the backing `NCollection_Sequence` has no O(1) indexed access), so every comparison paid that cost again — profiling attributed ~80% of runtime to `NCollection_BaseSequence::Find`. (2) the phase never polled its cooperative progress indicator below `BOPAlgo_CheckerSI::CheckFaceSelfIntersection`, so a caller's timeout could only fire between whole-face checks, not within one — exactly where the artifact got stuck. **Fixed in v1.15.1** (`Scripts/patches/0010-*`, xcframework rebuilt): `Intf_TangentZone::Points()` caches a true random-access array per zone (O(1) lookup); `Intf_Interference::SetBreaker` (thread-local, RAII-scoped via `Intf_InterferenceBreakerScope`) lets `Insert()` poll every 256 calls and abort by throwing `Standard_Failure`, wired up in `BOPAlgo_CheckerSI`'s self-intersect functor only when single-threaded (an exception from an `OSD_Parallel::For` worker thread would risk `std::terminate()`). Verified: a 0.5s deadline now returns in 0.547s and a 30s deadline in 30.1s, correct results throughout. Reproducer at [`Scripts/repro/319-selfintersection`](https://github.com/SecondMouseAU/OCCTSwift/tree/main/Scripts/repro/319-selfintersection); filed upstream as [OCCT#1385](https://github.com/Open-Cascade-SAS/OCCT/issues/1385) (repro) / [OCCT#1386](https://github.com/Open-Cascade-SAS/OCCT/pull/1386) (fix, CI green on all 3 platforms). #319.
 
 ### Carrying OCCT source patches
 

Package.swift

@@ -33,18 +33,19 @@ let occtTarget: Target = useLocalBinary
         name: "OCCT",
         path: "Libraries/OCCT.xcframework"
     )
-    // v1.12.9 rebuild: OCCT 8.0.0p1 + our carried patches — 0001 (ShapeFix_Face guard, #263),
+    // v1.15.1 rebuild: OCCT 8.0.0p1 + our carried patches — 0001 (ShapeFix_Face guard, #263),
     // 0002 (backport of upstream OCCT#1334, #280), 0003 (fillet TopOpeBRep thread_local, #298),
     // 0004 (ShapeAnalysis_FreeBounds owires init, #310), 0005 (ShapeFix_Face null-Context guard
     // in FixPeriodicDegenerated, #317), 0006 (BRepGProp_EdgeTool adaptor NbPoles, #318), 0007
     // (ShapeAnalysis_FreeBounds lwire reset, #323), 0008 (Geom_BSplineCurve O(1)
-    // PeriodicNormalization, #323), and 0009 (StepData_StepWriter split oversized string, #323).
+    // PeriodicNormalization, #323), 0009 (StepData_StepWriter split oversized string, #323), and
+    // 0010 (Intf_Interference O(1) tangent-zone lookup + checkpointed breaker, #319).
     // Bump BOTH url and checksum whenever the xcframework is rebuilt, or URL-resolving consumers
     // silently keep the previous kernel while local sibling builds get the new one.
     : .binaryTarget(
         name: "OCCT",
-        url: "https://github.com/SecondMouseAU/OCCTSwift/releases/download/v1.12.9/OCCT.xcframework.zip",
-        checksum: "52282c923321fbef5d6ff09ad2a3f4e664da2a37a236b0d1c4de48cbc73f00bb"
+        url: "https://github.com/SecondMouseAU/OCCTSwift/releases/download/v1.15.1/OCCT.xcframework.zip",
+        checksum: "b1c967f90ea45a74ba97893ce78297d8c2dee516024f9c79f80b16b388ece2c1"
     )
 
 let package = Package(