Skip to content

stl2step 1.4.1 + SolidOut 0.2.0 for macOS and Windows

Latest
Latest

Choose a tag to compare

View all tags
@BlinkingSun BlinkingSun released this 06 Sep 22:15
· 1 commit to main since this release
v1.4.1
cf8abc6

Downloads

SolidOut 0.2.0 — the desktop app (same UI on both platforms, engine 1.4.1 built in; no separate install)

Platform File Notes
macOS (Apple silicon) SolidOut-macOS-AppleSilicon.dmg Developer ID signed, notarized and stapled. Conversions need macOS 26 or newer (the bundled OCCT libraries).
Windows 10/11 (x64) SolidOut-Windows-x64-setup.exe (recommended) · SolidOut-Windows-x64.msi Unsigned (no Authenticode): SmartScreen will ask once on first run.
Checksums SHA256SUMS.txt All app files.

New in SolidOut 0.2.0: the engine updates itself. The Engine card in the right panel shows the engine in use and an Update engine button. On launch and on the button the app checks this repository's latest release; when a newer engine exists the line under the button turns amber and one press downloads, verifies (SHA-256 from the manifest; Developer ID Team ID on macOS) and installs it into the app's data folder. Nothing in the app bundle is modified, so signatures stay intact, and the next conversion uses the new engine without a restart.

Engine bundles — what the updater installs (you do not need these to use SolidOut)

Platform File Notes
macOS (Apple silicon) stl2step-engine-v1.4.1-macos-arm64.tar.gz stl2step + Frameworks/ (28 OCCT dylibs), Developer ID signed and notarized, macOS 26+.
Windows 10/11 (x64) stl2step-engine-v1.4.1-windows-x64.zip stl2step.exe + 33 DLLs beside it, unsigned.
Manifest engine-manifest.json · SHA256SUMS-engine.txt Schema 1: engineVersion, per-platform name, sha256, size, minOS/minWindows, teamId. A release without a manifest is not an engine candidate.

SolidOut is closed source; its binaries ship here as release assets. The engine it drives is this repository.

stl2step CLI 1.4.1 — build from source on macOS, Linux or Windows per the README; the CLI is what SolidOut bundles. 1.4.1 is 1.4.0 plus the version-string fix (the 1.4.0 binary reported stl2step 1.3.0) and the engine bundles above; there is no other engine change.

What 1.4 converts that 1.3 did not (measured on the training parts shipped in tests/corpus): shelf_bracket and Body244 no longer revert to facets (32 planes / 20 cylinders and 364 / 178, one valid closed solid each); plane-meets-cylinder ellipses bind as ELLIPSE edges (55 unhandled edges on the training set down to 1); the torus stage builds a certified torus face on the S04 fixture; the RESULT census counts exactly what ships. The same-surface union engine and mouth-round tori on the real parts are recorded for 1.5.

Engine changelog (1.4.1)

Fixed

  • Version string fix: the v1.4.0 binary reported stl2step 1.3.0 because project(stl2step VERSION …) in CMakeLists.txt and STL2STEP_VERSION_* in the public header were never bumped for the 1.4.0 release.

Added

  • Per-platform engine bundles (stl2step-engine-<tag>-macos-arm64.tar.gz, windows-x64.zip) and engine-manifest.json for the SolidOut in-app updater (published with the release).

Engine changelog (1.4.0)

Added

  • Torus engine — stage T transactional reclaim + U-3 face sharing the cylinder rim (D-140-6, D-140-11). Stage T admits torus candidates, but cylinder reclaim commits only when the U-3 Geom_ToroidalSurface face ships; a failed U-3 build restores the pre-stage-T topology and rebuilds with fresh vertices (Body28/Body11 byte-identical to the no-torus binary). The toroidal face shares the cylinder rim via explicit pcurves (Geom2d_Line((u0−f, v), (1,0)) on the reused rim TShape, both seam pcurves at u0 / u0+2π, and the c-rev wire). S04's boss-top round ships as a valid closed solid (TOROIDAL_SURFACE, solids=1, faceCount=9, I-checker PASS). S19 mouth rounds, the plate mouth band, and Handle pickup admit 0 tori — the certificate is honest on the real parts (smoothBuiltTori=0, expected-red S19.mouth-round-not-built). The torus closed-360 gate uses bandArc + asin(min(1, τ/hmin)) (θ_q) instead of the retired 1.5·bandArc hand constant; U-3 probe warnings are discarded on topology revert.
  • Torus math — the circle-on-torus class (src/refit_torus_math.{hpp,cpp}, torus_math_unit; D-140-6 §3(4), the D-130-3 hard gate applied verbatim; D-130-2). circleOnTorusMax returns, in closed form, the supremum distance from a circle COAXIAL with a ring torus to the untrimmed toroidal surface: every point of such a circle has the same (rho, z) = (r_c, h), so the distance is the constant |hypot(r_c − Rmaj, h) − Rmin| and that constant is the supremum — exactly 0 on every v-iso circle, in particular on the two tangency circles a mouth round is bounded by (S04: rho = 10 @ z = 22 and rho = 7 @ z = 25; S19: both rims of both tori). A non-coaxial circle, a spindle or horn torus (Rmaj − Rmin <= gp::Resolution(), OCCT's own guard) and a degenerate tube are refused with -1.0 / unhandled-circle-on-torus — tier 2, counted, never a widened tolerance. With pointTorusDist, torusVIsoCircle (the tier-1 bounds a blend patch carries) and torusVOfProfilePoint. Own translation unit for the reason the cone math has one (0d67eba): the P1 include allowlist is neither touched nor widened. No build path is edited; exactMaxAtBind still returns unhandled-surface for a torus until the build lane binds through this class.
  • The same-surface union census is readable with the door closed (D-140-8 U-R13; SPEC-union PART B). Under STL2STEP_LAWBAND_DIAG the DIAG_LAWUNION line no longer reports the STL2STEP_UNION door's behaviour: it takes a copy of every law band's claim, applies D-140-8 U-R1…U-R4 to the copy (the D-130-13(2) same-surface islands under max(bandResid, q), the joint-refit roll-back, the axial-extent overlap, the mutual-containment merge, in band order) and prints, per certified surface, nTri sigma punctures edgePieces domainFaces pinchVertices sizes=[..] pieces=[..] domainTris loc dir. sigma is U-R5's closed-form supremum max over the claim's welded edges of (R − dist(axis, midpoint)), punctures are U-R6's non-claim vertices within sigma of the surface, domainFaces counts U-R7's components of the claim plus the on-surface stars of its punctures, pinchVertices counts U-R8's vertices whose link inside a component is two or more fans. The copy is never written back: claimTris, the band fits and every emitted Region are untouched, so the shipped STEP is byte-identical with the diag on or off and with the door open or closed (the plate: R=10.000000 nTri=255 sigma=0.019733720 punctures=8 edgePieces=6 domainFaces=1 pinchVertices=0 domainTris=279; Handle pickup: R=3.999998 nTri=52 punctures=20 edgePieces=11 domainFaces=1 pinchVertices=0 domainTris=84). New TU src/refit_union_census.cpp (claimChordSagitta, unionCensus, declared in refit_internal.hpp; pure MeshView measurement, no Region, no build path) and unit target union_census_unit: sigma against the regular-ring closed form and a numerical supremum along every welded edge at three radii × three densities, world and placed frames, to Precision::Confusion(), plus the U-R6/U-R7/U-R8 counts on hand-built interrupted rings.
  • Ellipse math — the two closed forms behind the ELLIPSE bind class (src/refit_ellipse_math.{hpp,cpp}, ellipse_math_unit; D-140-9 §2, D-130-2, D-S3-111; the third instance of the 130-CONE-MATH / torus-math shape). Measured first (SPEC-ellipse PART 0): all 55 unhandled edges on the 9-fixture edge-class red line are plane∩cylinder ellipses whose plane-side pcurve is the exact 2-D ELLIPSE GeomAPI::To2d writes and whose cylinder-side pcurve is one degree-8 Bézier with affine u and a sinusoid-approximating v. ellipseOnPlaneMax is the EXACT supremum of |E(t) − S(pc(t))| = |W + A cos t + B sin t| over [f, l]: the tan(t/2) quartic's real roots (Ferrari, threshold-free shift, fixed Newton polish) plus the endpoints; W = 0 collapses to σ_max([A B]) (ellipseTrigSupZeroOffset), the shipped case to exactly 0.0, and for a = b it is ≤ circleOnPlaneMax (the D-S3-111 triangle bound) — the sanity relation between the two classes. ellipseOnCylMax is a certified UPPER BOUND and says so in its name: the section identity (ellipseIsCylinderSection: ρ ≡ R, θ = θ₀ ± t, z = z₀ + Λ cos(t − ψ), residual carried exactly), ε_u exact from the poles by the Bernstein hull, the axial residual per knot span by Chebyshev interpolation (L_n max_j |v(x_j) − z(x_j)| + Λ H^{n+1}/(2^{2n+1}(n+1)!), the refused midpoint-Taylor form is 2⁸ = 256× looser), composed through the exact two-point identity on a cylinder. The grazing predicate (D-S3-111's open predicate): ellipseCylGrazingSin = |N̂·d̂| = cos α = b/a and ellipseVertexBound = √(s² + h²) + s²/(2ρ_min) with |h| ≤ δ_pl, |s| ≤ (δ_cy + δ_pl|cos γ|)/sin γ; the class refuses (unhandled-ellipse-grazing) where that exceeds meshTolCap — never a hand angle. Every refusal is named (unhandled-ellipse-plane-{param,degenerate}, unhandled-ellipse-cyl-{oncyl,uaffine,span,degenerate}). No build path is edited; exactMaxAtBind still returns unhandled-other for an ellipse until the bind lane binds through these classes (SPEC-ellipse PART B), so edgeUnhandled is unchanged by construction and B0 holds trivially.
  • The ELLIPSE bind class — ellipse-on-plane (EXACT) and ellipse-on-cylinder / ellipse-on-cylinder-bound (a certified upper bound) in src/refit_ellipse_bind.{hpp,cpp} on top of src/refit_ellipse_math.{hpp,cpp}, reached by two branches in exactMaxAtBind (D-140-9 §1/§3). unhandled on the 9-fixture edge-class red line was 55, and every one of the 55 was an ellipse — an oblique plane∩cylinder cut shipping as tier-1 Geom_Ellipse with no closed-form supremum. The plane side is exact: both curves are evaluated at the same t, so E(t) − S(pc(t)) = W + A cos t + B sin t identically, and the supremum is the maximum of |·| over the real roots of the quartic in τ = tan(t/2) plus the two endpoints. The cylinder side is a bound, and says so in its name: under the section identity (centre on axis, a|Û⊥| = b|V̂⊥| = R, Û⊥·V̂⊥ = 0) the azimuth is affine — so |Δθ| ≤ ε_u exactly, from the poles, by the convex hull at the Greville abscissae — while the axial coordinate is z₀ + Λ cos(t−ψ), and no NURBS reproduces cos t in t, so the axial residual is bounded per span by L_n·max_j |v_pc(x_j) − z(x_j)| + Λ h^{n+1}/(2^{2n+1}(n+1)!) at the n+1 Chebyshev nodes, over an exact de Casteljau subdivision (every depth is a valid bound, so the minimum over a fixed depth set is taken). The two compose on the cylinder as √(4R² sin²(ε_u/2) + max|Δz|²). Refusal is by named clause, never by silence: unhandled-ellipse-{plane-param,plane-degenerate,cyl-oncyl,cyl-uaffine,cyl-span,cyl-degenerate,grazing}; the range clause carries D-130-16's seam sub-clause, so a full-turn ellipse edge is refused rather than certified. Measured: unhandled 55 → 1 (Body28 25→0, S11-b 24→0, linkage_bores_chamfer 4→0, handle-pickup 1→0; S16-R1-explode-success stays 1 under the seam clause, residue ellipse.s16-post-bind-tolerance), with overTol, overCap and polylineTier2 unchanged on every fixture and every corpus STEP byte-identical. No tolerance writer, no pcurve writer and no instrument was touched.
  • J6/free-edge containment, landed then fully sunset (D-140-4, D-140-5, D-140-12; SPEC-reverts-cycle2 §2). Cycle 2 landed three edits; none of them ships. Change 3 (the J6 if (j6UncollapsePass > 0) handoff plus smoothContainedRegions/smoothContainedTriangles) was reverted by D-140-5's sunset. Changes 1–2 — regionClosureHealEligible dropping builtAs == BuiltAs::Single in favour of regionBuiltAnalytic(...), and fallbackGuardPass == 0 widened to <= 2 — were kept as "dormant" by D-140-5, then reverted by D-140-12: on Linux they were live, not dormant, and let S16-R2-ChainUnstable complete the analytic path (spike-s16r2-linux: first build at 78f85b4). Restored to the 3bf0575 form: eligibility requires builtAs == Single; the heal stays behind fallbackGuardPass == 0.
  • Two RESULT keys, smoothContainedRegions and smoothContainedTriangles (D-130-10 declaration; RefitStats::containedRegions / containedTriangles, summed in Converter::run). Incremented at exactly one place — the J6-handoff demotion, after the admitted set shrank. Never on the seamed360-fail explode path, never on the D-S3-54 closure heal (which already fires at 3bf0575 and keeps reporting itself through DIAG_HEAL / DIAG_HEAL_SUM). The warning smooth: contained N region(s) to facets -- component built is emitted only when N > 0. DIAG_CONTAIN is emitted only from the J6 handoff, under diagP2Enabled(), with a real freeBefore > 0.
    The site never shipped live containment on any artefact this project owned (keys 0/0, DIAG_CONTAIN count 0). Sunset (D-140-4 §1(5)) fired in D-140-5 for the J6 handoff. D-140-12 then reverted changes 1–2 as well: they were not dormant on Linux — they let S16-R2-ChainUnstable build (planes/cylinders/tier2) where macOS and 3bf0575 Linux still revert. Eligibility and reachability do not stay.
  • S20_cross_bore_union, the same-surface union fixture, with its own negative control (D-140-8 U-R15/U-R16). An 80×60×40 block with a mid-span slot, one Ø20 bore on +X across both walls — so it ships as two coaxial faces at disjoint axial spans, [0,30] and [50,80] — and a Ø12 cross bore on +Z at x = 15 piercing the wall-A span at u = 90° and u = 270°, strictly inside it. Both bores are 48-gon prisms, and that is the mechanism: the two prisms' intersection polyline alternates between bore-edge × cross-facet vertices (exactly on R = 10) and cross-edge × bore-facet vertices, which lie on a chord of the bore and are therefore punctures in D-140-8 U-R6's sense — the plate's eight slot-window corners, reproduced by construction. gen_corpus asserts the topology on its own welded float32 mesh before any engine run and prints the S20-UNION census: union bore sigma=0.0214118609 mm = 3240.66178 q, edgePieces=38, punctures=92, domainFaces=1, innerWires=2, windingPerWire=[0,0], pinchVertices=40; control bore domainFaces=1, innerWires=0; axialOverlap=0. The predicate is never relaxed — a deflection-halving retry loop runs up to 8 times and the lane throws if it exhausts. exactVolume is a closed form on the generator's own parameters (inclusion–exclusion over slot ∪ bore ∪ cross bore; the two prisms' intersection integrand is piecewise quadratic, so Simpson on the breakpoint intervals is exact, not a quadrature) and is asserted against BRepGProp::VolumeProperties on the exact solid: closedForm=142038.9690342799 BRepGProp=142038.96903428048. The PRG multiset ceiling for R = 10 on this fixture is therefore exactly 2: over-segmentation fails (D-130-14) and over-merging fails (U-R3). src/ is untouched — the union itself is expected-red as S20.union-not-built.
  • Two real-export corpus fixtures, shelf_bracket (2340 tris; 32 planes + 20 cylinders + 735 facet faces in one valid closed solid, volumeDeltaPct = 0.000142) and Body244 (6268 tris; builtPlanes/builtCylinders 364/178 (D-140-3 record), volumeDeltaPct = 0.000099, rid=1 explodes to 148 facets under expected-red Body244.cyl360.rid1). shelf_bracket is the analytic-plus-facet-island proof on real geometry that S17_hole_vs_island was to assert synthetically; S17 is withdrawn this cycle under expected-red S17.cap-claimed-by-a2 (D-140-4 §4).
  • stl2step_grade, the mesh-oracle recovery grader (tests/grade/, targets stl2step_grade + stl2step_grade_selftest; D-140-1). Independent of stl2step::core (grafts only stl_quant); oracle classes plane / cylinder / cone / sphere / torus from the mesh at τ = 2q; seven training grades + ctest selftests. Cycle 6 torus/cone oracles certify on the training parts (D-140-1, D-140-6). Selftest core 68/68, synthetic 25/25.
  • Torus fixtures — S04 recoverable + S19_mouth_round (D-140-6 §4, SPEC-torus PART A). S04's mustRemainFaceted TorusNYI is cleared and recoverable gains torusRec(7.0, 3.0, {25,25,22}, +Z, 1, true) (STL bytes unchanged); S19_mouth_round is a 60×60×20 box, R=10 through hole, mouth rounds r=2 top / r=1 bottom. Generator-exact sidecars and *.exact.step; S04 now builds a toroidal face; expected-red S19.mouth-round-not-built stays; gates_full I-checker XFAIL on S19 registered (D-140-6 §4(e)). Fixtures only on the fixture commit; build landed on this branch.
  • Volume-attribution identity, default-off (STL2STEP_VOLID; D-140-10). Diag-only instrument beside dumpR2Probe: pcurve and contour readings of v(F)=(1/3)[kA+p·N], the §2.5 closes bracket, and the S01/S02/S03 unit (volid_identity 17/17). faceVolumeContribution and the caller probe are untouched; no product read of closes. Knob unset/0 is a no-op (OFF≡ON for STEP bytes). Measured closes 8/9 (plate volid.cone-bracket).

Removed

  • The J6 handoff containment round and its two RESULT keys (D-140-5): D-140-4 §1(5)'s sunset fired — 1 529 generated candidates found no constructible S18_uncontainable_rim (clause A, an orphaned collapsed chain within the site's 1.0 mm matchTol, fires 0 times post-heal) — so refit_build.cpp:15877-15925 reverts to its 3bf0575 form, dropping the containment round, smoothContainedRegions/smoothContainedTriangles, resetContained(), and the smooth: contained ... warning.
  • J6 eligibility and heal reachability restored to 3bf0575 (D-140-12, completing D-140-5): D-140-5 had left change 1 (regionClosureHealEligible → regionBuiltAnalytic) and change 2 (fallbackGuardPass <= 2) in place as widenings of a "live" heal. On Linux those two lines were the first-build at 78f85b4 for S16-R2-ChainUnstable (spike-s16r2-linux: smoothRevertedComponents 1→0, polylineTier2 0→25, overCap 0→1). Restored: builtAs == BuiltAs::Single and fallbackGuardPass == 0. MacOS B0 is unchanged because the path was dormant there; Linux S16-R2 returns to the revert contract.
  • The STL2STEP_UNION door (D-140-8 U-R14): cycle 1 did not close U-R9 (plate seamed cylinder with 2 inner window wires = UnorientableShape st=27; domain-boundary extraction yields window-sized caps and fragment loops, not a face-construction miss). The door is deleted; merge/emit stay off so no partial construction lands. S20 fixture, census, and expected-red rows stay. The union engine is deferred to 1.5.

Tests (1.4.0)

  • S19_mouth_round compose expectation (lane seeds-140c, D-130-24; D-140-6 §4(e)) — generator sidecar: six box planes in recoverable[]; recoverable6/7/8 (R=10 closed360 cylinder + two coaxial mouth-round tori) declared mustRemainFaceted with S19.mouth-round-not-built / TorusNYI wording; closed-form geometry kept in deferred comments. STL bytes unchanged; p1_compose_gate green.
  • 1.4.0 instrument seeds (lane seeds-140, D-130-24) — Body244 / shelf_bracket live volumeBudgetMM3 from volumeBudgetMM3(|meshVolume|) (D-140-3 §4(b)); edge-class new cells for Body244, shelf_bracket, S19_mouth_round, S20_cross_bore_union from edge_class_gate.py --measure; S11-b expected-red shrink; S20 live[] disposition ESCALATE (S20.union-not-built sunset). S04 / recoverable7 torus rows left unseeded. Body28 polylineTier2 left at 1442 (macOS still 1442). S16-R2 B0 UNCERTAINTY exception retired (D-140-13): canonical STEP identical to 3bf0575 on all platforms.
  • Per-run scratch dirs in corpus engine check — run_engine_check.sh writes fixture artefacts under mktemp -d instead of fixed /tmp/${id}_* paths, so concurrent ctest batteries no longer truncate each other's RESULT JSON.
  • Windows volid_identity portable launch (5ff406b; D-140-10): MSVC-hosted volid_identity uses _putenv/_popen instead of the POSIX setenv/popen path — no shell prefix — so the unit test launches on Windows CI.
  • gates_full ctest TIMEOUT 3600→6000 (75e4e95): budget from measured wall time on the shop PC (~3870 s Windows/MSVC; macOS 883 s, Linux 1174 s); Body244 dominates on Windows (~55 min vs ~3 min macOS) — Windows I/O pathology, 1.4.1 follow-up. Harness budget only; no gate logic change.

Fixed

  • J6 sunset completion — closure-heal eligibility and guard bound restored to 3bf0575 (D-140-12; D-140-5; cd6e86f): D-140-5's "dormant" widenings (change 1: regionClosureHealEligible → regionBuiltAnalytic; change 2: fallbackGuardPass <= 2) were live on Linux and let S16-R2-ChainUnstable take the analytic path where macOS still reverted. Restored builtAs == BuiltAs::Single and fallbackGuardPass == 0.
  • A component with refused-mint composites takes the facet fallback when its shell fails (D-140-13; D-140-2): buildWithInners still returns true after a failed prepare() on a mintRefused polyline-rim face — those faces can ship in a closing shell (shelf_bracket 32/20, Body244 builtPlanes/builtCylinders 364/178 (D-140-3 record)). When the component's shell then fails to close or validate, buildFaces takes the existing ChainUnstable → all-facet path instead of returning false, so S16-R2 ships as facets the way 3bf0575 did. Components without a mintRefused face keep the previous shell-failure semantics. Cap-mint refusal is unchanged.
  • RESULT census counts the shipped pass, not the torus probe (D-140-6, D-140-11): when stage T admits candidates, U-3 fails, and the pre-T snapshot rebuild ships, smoothBuiltPlanes (and the other construction counters) were left on the discarded probe's RegionSet — Body11 1603→1590, Body28 499→488, Body244 364→338 — while the STEP stayed byte-identical. The pre-T snapshot now carries its own RefitStats and the revert restores it (same leak class as the discarded probe warnings).
  • A component with refused-mint composites takes the facet fallback when its shell fails (D-140-13; D-140-2): buildWithInners still returns true after a failed prepare() on a mintRefused polyline-rim face — those faces can ship in a closing shell (shelf_bracket 32/20, Body244 364/178). When the component's shell then fails to close or validate, buildFaces takes the existing ChainUnstable → all-facet path instead of returning false, so S16-R2 ships as facets the way 3bf0575 did. Components without a mintRefused face keep the previous shell-failure semantics. Cap-mint refusal is unchanged.
  • Torus U-3 probe warnings discarded on topology revert (D-140-6, D-140-11): when stage T admits candidates but U-3 fails and the pre-T snapshot rebuild ships, warnings from the discarded probe build are no longer forwarded — Body244 returns to 2 warnings (IntAna + seamed360) with byte-identical STEP.
  • Transactional torus cylinder reclaim (D-140-6, D-140-11): stage T still admits candidates, but a failed U-3 torus face build restores the pre-stage-T topology and rebuilds with fresh vertices so Body28/Body11 match the no-torus binary; cylinder reclaim commits only when Geom_ToroidalSurface ships. The torus closed-360 gate uses bandArc + asin(min(1, τ/hmin)) (θ_q) instead of the retired 1.5·bandArc hand constant.
  • The OpenCASCADE toolkits the engine uses are named explicitly (github issue #1): stl2step_core linked only the finder's umbrella target, which omitted TKPrim (BRepPrimAPI_MakePrism) and TKBO (BRepAlgoAPI_Fuse, BOPAlgo_GlueEnum); they arrived transitively on Homebrew and on an OCCT that ships a CMake package config, and not at all under Debian/Ubuntu's ld --as-needed when OCCT is installed without one (undefined reference to BRepAlgoAPI_Fuse::~BRepAlgoAPI_Fuse() / libTKBO.so: DSO missing from command line). CMakeLists.txt now links the measured minimal cover of 15 toolkits, selected by if(TARGET ...) with a find_library fallback and the pre-7.8 TKSTL/TKSTEP* spellings as alternates.
  • takeFullCap may not mint against an unbuilt partner (D-140-2): a Seamed360 cap facing an island / eprimeFill / exploded / non-analytic region keeps the mesh polyline both sides already share, instead of inventing a full-2π CIRCLE TShape that the fill cannot carry. shelf_bracket's eight through-holes ship as cylinders; the shell closes with no J6: line. Body244 ships builtPlanes/builtCylinders 364/178 (D-140-3 record; volumeDeltaPct = 0.000099); the shell closes with no J6: line.
  • Corpus checker parses sidecars linearly (MSVC hosted CI): check_corpus.cpp parsed each sidecar with std::regex patterns chaining [\s\S]*? across the document; MSVC throws regex_error(error_complexity) on the plate sidecar (Exit code 0xc0000409, no stdout). Replaced with a single forward scan; matching semantics kept.

Changed

  • Grader selftest expected-red, and its first row retired in the same release (D-140-7): tests/gates/baseline/expected-red.json gains a grader section and grade_selftest the XFAIL/XPASS reporting that reads it — a listed check that fails is reported XFAIL and excluded from the exit code; a listed check that passes is reported XPASS, counted as a failure, and names the row to delete, so the list can only shrink. The section's first and only row, grade.reverse-seed.handle-pickup (case 9's cylinder-phase claim-order dependence), was registered when the grader landed and is gone again in this same 1.4.0 cycle. What closed it: the oracle's stage-A pass no longer consumes triangles or names a seed. Each class phase now answers one boolean per edge-connected component of the unclaimed set — does this component carry the class? — over a frozen ambient set, and §5.4's growth walk is then re-run over the union of the carrying components, entered at the lowest triangle index. That domain is a function of (mesh, class order) alone, so no permutation of the seed enumeration can move it; the three mechanisms that used to move it (band-offset chop, skipped-seed slack, two competing cylinders on the same quads) are each closed by it. The probe itself is untouched and still real — --reverse-seeds still reverses seedList(), which still drives the growth for every seed, and grade_selftest.cpp's assertions are unchanged. Selftest core 68/68 (was 67/68), synthetic 25/25, 0 XFAIL, 0 XPASS; the seven training grades are byte-identical under file order, --reverse-seeds and a third geometry-derived permutation (--seed-order n, triangle-centroid order), with no grade changed. The empty grader section stays: the machinery is what landed, the deferral was not.
  • Same-surface union engine → 1.5 (D-140-8): the STL2STEP_UNION door is deleted, not compiled-in-off. 1.4.0 ships the seed piece(s) per surface (plate cross bore 214/279 in 2 faces); S20 remains expected-red S20.union-not-built.
Assets 10
Loading