Plugin engines (Opus55, Qwen, Terra) each add two projects at the repo
root, and more are coming; at a dozen they would outnumber the core
projects. They are also a different kind of thing: out-of-solution,
runtime-loaded plugins. Grouping them under Engines/ keeps the root
readable.
Engines/Directory.Build.props now holds the shared TFM, nullable and
implicit-usings settings and the OpenNest.Engine reference, so a new
engine's csproj is nearly empty. The tests/ compile exclusion lives in
Directory.Build.targets because a removal in .props runs before the SDK
adds its default Compile glob and has no effect.
Build-Engines.ps1 replaces the per-README manual build-and-copy steps
for deploying engines into the benchmark's runtime Engines/ folder.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Spacing offsets move onto Clipper (ClipperBridge) for CPU preparation,
fixing spikes and inverted loops where features are narrower than the
spacing; Collision stays hand-rolled for the GPU path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Converts a PepApi year of PEP nests into .nest files that keep PEP's own
placements, so OpenNest.Benchmark can score PEP as its Baseline row.
Micro-joint tabs: PEP leaves tabs uncut by jumping them with a rapid, at a
contour's seam or partway along it (a cutout cut as two halves 0.02 apart).
The part still occupies that material, so open cut runs are chained end to
start across gaps up to 0.25 and bridged with a cut line, but only where
they close into a loop, so separate contours that lie close together (two
circles 0.25 apart) are never merged. Previously only seam tabs were closed,
leaving mid-contour tabbed cutouts open and the validator reading them as
garbage regions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conservative flattening circumscribes arcs, so at 0.01 the validator was up
to 0.01 too strict along curves: PEP's P260626-03 layout, exactly 0.25 apart
along an arc, failed with a 0.004 sliver. At 0.001 the worst error on either
side is 0.001. Validating the 26-job benchmark set with Opus55 went from
about 2 s to about 5 s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Clipper validator flagged valid Opus55 and PEP layouts (P260805-03,
P260626-03). Two causes:
- Arc.ToPoints(circumscribe) scales every vertex out by 1/cos(step/2),
endpoints included, so a 0.03125 corner fillet flattened at 0.01 poked
0.013 past the straight edges it meets. ClipperBridge now flattens itself:
circumscribed arcs keep their endpoints on the arc and put interior
vertices on tangent intersections, with the segment count chosen so the
outward error stays within the tolerance. Arc.ToPoints is unchanged for
its other callers.
- The conservative padding made a layout exactly at the spacing fail.
NestValidator now uses OffsetForValidation: the same conservative
flattening, round joins at a tenth of the tolerance, no padding. Its only
leniency is that join chord error at convex corners.
With both, Opus55 is valid on all 26 benchmark jobs (25 before the
Clipper migration; the old failure was a spike artifact).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Spacing offsets now go through Clipper for CPU preparation while the
per-pair Collision test stays hand-rolled for a future GPU port; record
that split, which offset path each caller uses, and the missing Clipper2
entry in the NuGet list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GetOffsetPerimeterEntities/GetOffsetPartEntities feed directional-distance
loops (FillLinear, Compactor, RotationSlideStrategy) that handle arcs
natively. Switching them to Clipper line output (plan option B) made
OpenNest.Tests run 48s -> 8m19s, Fill tests ~3x slower, and broke 20
exact-fit tests through tessellation and conservative padding, so they keep
the per-entity offset (option A), hardened:
- Arc, Circle and Line offsets are now side-symmetric. Right on a CCW arc
shrank instead of growing, Right on a CW circle grew, and Right on a line
offset to the left and reversed it. Only Left was used on hot paths, so
this was latent (SimplifierViewer drew both tolerance bands on one side).
- Shape.OffsetEntity closes every gap between consecutive offset pieces:
convex non-tangent line/arc corners get a round join about the original
corner, lines across a collapsed fillet are mitered, and any other gap
(concave arc corner, collapsed entity) is bridged with a line. Before,
only line-line corners were joined, so a vertex could slip through.
- Zero-area spikes are left in place and documented: they lie inside the
offset envelope, which is harmless for directional distance.
- OffsetOutward/OffsetInward become internal; PartGeometry is their only
caller.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Collision stays hand-rolled because it is the reference for a future GPU
kernel, but its inputs now come from Clipper region offsets. Pin down that
lines-only, round-join, 1e-4-precision polygons keep the contact and
part-in-part semantics: a neighbor inside a collapsed slot, a part inside
a hole that shrank by the spacing, and zero-spacing edge contact.
Document which steps are per-polygon preparation to cache and upload once,
and which are per-pair kernel-shaped work.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Its callers now take Clipper region offsets, which never produce the
self-intersections it patched over (and it only caught proper crossings,
so spikes survived it anyway). Polygon.OffsetEntity was its last caller;
the override is required by Entity but has no callers, so it becomes a
Clipper miter offset that keeps the Left/Right semantics and the input
winding. FindCrossing, SplitAtCrossing, SegmentsIntersect and the static
CalculateArea helper go with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Per-entity offsetting left spikes and inverted loops wherever a feature is
narrower than the spacing (1.nest), and RemoveSelfIntersections only
caught proper crossings. The callers that already flatten to polygons now
take a single Clipper region offset instead:
- PolygonHelper (BestFit) and PartBoundary use the conservative mode, which
keeps their never-under-estimate guarantee. PartBoundary also keeps holes
that appear when a perimeter curls back on itself.
- NestValidator offsets perimeter and cutouts in one region; Clipper drops
collapsed cutouts, so the collapsed-or-flipped heuristic goes away.
- CutOff.IntersectPerimeter offsets through the bridge. The old
OffsetEntity(Left) grew CW perimeters but shrank CCW ones, so with the
plate's perimeter cache a cut-off ran through the part; slots narrower
than twice the clearance now close up instead of leaving a gap.
- GetOffsetPartLines (3 overloads) and the AddOffset* helpers had no
callers and are removed, as is EntityView's never-defined DRAW_OFFSET
block.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NoFitPolygon.Compute, its triangulate-and-union MinkowskiSum branch and
UnionPolygons had no callers; only ComputeConvex (NfpSlideStrategy) is used.
The Clipper path helpers they relied on now live in ClipperBridge.
ConvexDecomposition.Triangulate stays because Collision uses it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Offsetting entity by entity leaves spikes and inverted loops wherever a
feature is narrower than the spacing, and RemoveSelfIntersections only
catches proper crossings. ClipperBridge flattens a ShapeProfile into one
region (perimeter positive, cutouts negative) and inflates it in a single
Clipper pass with round joins, so narrow features collapse and holes that
close up disappear.
Conservative mode circumscribes perimeter arcs, inscribes cutout arcs and
pads the inflation by the chord tolerance, so the result never
under-estimates the spacing. It replaces the circumscribed-polygon
guarantee the BestFit/PartBoundary callers rely on.
Clipper stays confined to CPU preparation whose output is cached; the
per-pair Collision path remains hand-rolled for GPU portability.
LayoutPart's display offset now goes through the bridge.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
PEP-exported programs carry arc centers that are not equidistant from the
start and end points (e.g. I0.03 on a 0.0598 chord). Building the arc from
the end radius left its start off the previous move's end, so contours
failed to chain. Project the center onto the chord's perpendicular bisector.
The Draw Offset display offset each entity separately, which left spikes and
inverted loops wherever a feature is narrower than the spacing (1.nest,
P260417-06). Inflate the flattened region with Clipper instead, which
collapses narrow features and drops holes that close up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rename the OpenNest.Engine.Sonnet5 scaffold to OpenNest.Engine.Opus55 and
implement an independent whole-job INestingEngine (no built-in engine,
registry, or best-fit internals are called or copied).
- PartCatalog: snapshot perimeter -> polygon per allowed orientation, with
adaptive chord tolerance and MBR-aligned rotations for Automatic parts.
- NoFitCache: spacing footprints and cached Clipper2 Minkowski NFPs (convex
fast path; concave sweep plus both containment terms).
- FrontierPacker: per-(type, orientation) free regions (inner-fit rectangle
minus NFPs), updated incrementally; gap-fill-largest, else least front
advance per area^beta.
- Engine: look-ahead stock choice by estimated whole-job net area, six
deterministic variants, tail re-plan of the last 1-3 sheets.
- Tests judged by OpenNest.Benchmark's NestValidator, including an NFP
containment regression guard.
P260805-10.nest (219 parts), all 9 stock sizes: 219/219 valid, 27 sheets,
91.7% utilization, ~7 s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Three standalone INestingEngine plugin projects (Solve() unimplemented,
throws NotImplementedException), each outside OpenNest.sln and discovered
at runtime via NestingEngineRegistry.LoadPlugins, same pattern as
OpenNest.Engine.Aurora. Each README spells out that Solve() must be an
independently designed placement algorithm and must never call/wrap/select
over the built-in engines (StockLadderNestingEngine, FixedStrategyNestingEngine,
PlateNesterFactory, NestingEngineRegistry).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The benchmark is about to be used as the objective for LLM-designed
engines, and several gaps would have rewarded the wrong behavior:
- Ranking was utilization-first, so dropping awkward parts raised the
score. Rank valid > fully placed > cost > plates, where cost is
salvage-credited sheet area plus a largest-sheet penalty per unplaced
part; placing a part is never scored worse than omitting it.
- Salvage rate was ignored in scoring; cost now uses EstimateNetArea,
recomputed from job geometry rather than trusted from the engine.
- Rotation constraints were never validated. Add RotationPolicy.Allows
(shared with NestJobPlacementValidator) and check every placement.
- Returned sheets were trusted, so an engine could loosen spacing or
invent a size. Sheets must now match offered stock.
- Part-in-part placements were flagged as overlaps; spacing now accounts
for cutouts, with an X-sorted sweep to prune distant pairs.
- Summary averaged per-job percentages; it now sums areas and cost.
- --spacing and sheet sizes parsed with the current culture.
- Warn when .nest jobs offer only their original sheet sizes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Part.Clone() rebuilt the clone from BaseDrawing and then re-applied the
part's absolute Rotation on top of it. Since BaseDrawing.Program.Rotation
is itself absolute (baked in), this double-counted it whenever the base
drawing already carried a nonzero rotation, silently corrupting the
clone's orientation while its Location stayed unchanged.
This only manifests for drawings needing canonical-frame axis correction
(nonzero Source.Angle), since DefaultPlateFiller wraps every drawing in a
rotated canonical copy before running any fill strategy. FillHelpers.
BuildRotatedPattern clones parts before tiling, so any strategy that
tiles interlocking pairs (Pairs, Strip/Remnant, Column/Row) could produce
overlapping placements for such drawings.
Fix: clone the already-composed Program directly instead of re-deriving
rotation from BaseDrawing.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Fixes BestFit overlap check to agree with the geometry BuildParts
actually places, resolving BestFitOverlapTests.KeptPairs_NoOverlap.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
PairEvaluator checked overlap on the raw candidate geometry before
applying the pair's OptimalRotation, but BestFitResult.BuildParts (and
everything downstream) rotates both parts by -OptimalRotation before
placing them. Re-tessellating a rotated arc at the overlap chord
tolerance samples different chord points than rotating an
already-tessellated polygon, so a few tangent-corner candidates came
out overlap-free in the raw frame but overlapping once actually
placed.
Move the landscape-normalization step before the overlap check and
rotate part1/part2 the same way BuildParts does before tessellating
and running Collision.HasOverlap, so Keep agrees with the geometry
that's actually placed. Fixes
OpenNest.Tests.BestFit.BestFitOverlapTests.KeptPairs_NoOverlap (was
failing 3/1082 candidates).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
_labelPoint is computed from BaseDrawing.Program's current geometry,
which already carries BaseDrawing.Program.Rotation (non-zero for
canonical-frame drawings). BasePart.Rotation is cumulative from that
same baseline, so it must be re-applied net of the baseline already
baked into _labelPoint, mirroring CanonicalFrame.RebindToOriginal.
DefaultPlateFiller runs its search in a canonical (MBR-axis-aligned)
copy of the drawing. Intermediate progress reports — the Nesting
Progress dialog, PlateView's active-parts overlay — were showing that
transient canonical orientation instead of the drawing's real one.
FillContext.OriginalDrawing carries the pre-canonicalization drawing
through the pipeline; ReportProgress rebinds reported parts to it via
CanonicalFrame.RebindToOriginal before they reach the UI. Uses a
shallow list copy rather than per-part Part.Clone() — Clone() re-derives
its target rotation from BaseDrawing.Program.Rotation + Rotation, which
would double-count the canonical drawing's baked source angle.
Splits PairCandidateSelector (strip-mode candidate ranking) and
PairRemnantFiller (leftover L-shaped area fill) out of PairFiller into
their own classes. Pure extraction — logic is unchanged, just relocated
and given dedicated unit boundaries so each piece can be tested and
reasoned about on its own.
Part.Intersects has been silently non-functional everywhere it's used.
Shape.ToPolygon()/ToPolygonWithTolerance() never called UpdateBounds(),
so every freshly-built polygon kept Entity's constructor-default
zero-size bounding box regardless of its actual vertices. Collision.Check's
first step is a bounding-box pre-filter, and a zero-size box can never
overlap anything, so it always short-circuited to "no overlap" no matter
what the real geometry looked like.
That masked a second bug: Part.Intersects built its polygons via the
unconditional 1000-segments-per-arc ToPolygon() default instead of an
adaptive tolerance. For parts with several small fillets/holes this
produced tens of thousands of vertices, making the now-correct bbox
check fall through into a triangulation/clip step too slow to return
in practice. Switched to ToPolygonWithTolerance at a named tolerance
matching PartGeometry's existing convention.
PairEvaluator's own Keep/overlap check had a third, independent bug:
it used Shape.Intersects (edge-crossing detection only) at a coarse
0.01 chord tolerance, which misses containment-style overlaps and can
polygonize rounded corners coarsely enough to hide a genuine sliver
overlap. Switched to Collision.HasOverlap (full polygon clip, handles
containment) at a tighter dedicated tolerance.
Verified against a real nest file: PairFiller was tiling a BestFit
pair that PairEvaluator had incorrectly marked Keep=true, producing
visibly overlapping parts on the plate that no downstream overlap
check ever caught.
Known follow-up: OpenNest.Tests.BestFit.BestFitOverlapTests.KeptPairs_NoOverlap
still fails on 3/1082 synthetic candidates that overlap by a sub-0.001
sliver right at a rounded-corner tangent point — a separate, much
smaller precision edge case in PairEvaluator's raw (pre-transform)
coordinate frame, not a regression from this change.
Replace the MainForm/AutoNestForm engine combo bindings and every
desktop NestEngineRegistry call: selection now lives in app-scoped
EngineSelection addressing jobs engines in NestingEngineRegistry, the
combo lists the four built-in strategies (StockLadder stays out per the
frozen UI decision), and Engines/ plug-ins load through
NestingEngineRegistry.LoadPlugins. Whole-job fill routes through the
public PlateFillService with the selected strategy and plate number;
interactive group fill (PlateView) and area fill (ActionFillArea) go
through PlateFillService with identical accept/cancel preview behavior.
Multi-plate and size-search orchestrators receive the explicit strategy.
No desktop code reads or writes process-global engine state.
Console --engine now names a jobs engine for --autonest (solved once
through NestingEngineRegistry.Create and committed onto the plate) or a
built-in fill strategy for single-plate fill through the public
PlateFillService; unknown names exit with the valid choices instead of
consulting the process-global legacy registry. MCP nesting tools take an
explicit engine argument per call with the session default, never read
process-global active-engine state, and reject whole-job engine names on
single-plate fill tools. NestingEngineRegistry gains an explicit Create
(name) resolution; PlateFillService gains a public ResolveStrategy and a
plate-number Nest overload used by interactive callers.
Pin committed fulfillment and exact placement poses for Default, Strip,
Vertical Remnant, and Horizontal Remnant through the production
PlateNesterFactory + NestJobRunner path, plus deterministic poses for
OrderedPlateNester via its StockLadder wiring (no legacy counterpart) and
authoritative progress-stage sequences for StockLadder and
FixedStrategyNestingEngine. Captured at 42bbde7 (post ShrinkFiller axis
fix); these fixtures are the regression net for the legacy-engine removal.
Note: Strip is pinned on a rectangle-variety job. On dense mixed-shape
jobs the iterative shrink path intermittently proposes overlapping
candidates (pre-existing scheduling nondeterminism), so its mixed-geometry
layout is deliberately not pinned.
The legacy engine reads RotationStart == RotationEnd == 0 as
unconstrained and its Pairs/RectBestFit strategies rotate freely, so it
returned poses a Fixed/BoundedSweep RotationPolicy forbids and the job
validator threw. DefaultPlateNester now delegates such requests to the
policy-aware OrderedPlateNester.
Fixes RunAsync_FiniteStockExhaustion_PreservesUnplacedRequirementAndLockedRotation.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
RemnantFiller removes the topmost placed part to keep a clean rectangular
obstacle for the next drawing, but the envelope then walls that slot off,
so the part was lost for nothing (4 squares on a 9x9 plate became 3).
Only remove it while another drawing still has demand.
Fixes the mixed-stock NestRunner tests.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Part.Rotation is cumulative, so rebinding canonical parts with
CreateAtOrigin(original, p.Rotation) double-counted the drawing's own
rotation, and FromCanonical rotated each part about its Location, which
moved it off its slot and out of the work area. Add
CanonicalFrame.RebindToOriginal (rotation = part - original program
rotation, footprint aligned to the canonical part) and use it in the
three places that duplicated the old logic.
The MBR only fixes the frame modulo 90 degrees and nest results are not
90-degree symmetric (an L gave 56/43/42/42 parts by orientation).
CanonicalAngle.Compute now picks one of the four orientations from the
centroid offset; symmetric shapes keep the MBR orientation.
Fixes the three NestInvarianceTests.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Min/max were seeded at 0, so any geometry not touching the origin got an
inflated box, and the first move only updated max (else-if). Rotated
canonical drawings are the common trigger: their origin ends up outside
the shape, which skewed Part bounds and bbox-based alignment.
Track the real extents and keep returning a zero box for empty programs.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
NestDxfJob imports a DXF folder plus a quantity workbook, runs a
registered whole-job engine, and only publishes a nest that passes
quantity, bounds, overlap/spacing, and cut-only checks both before and
after save/reload. Writes import and validation reports; partial or
invalid results exit nonzero. Standalone project (not in the solution);
documented in the README.
ajandClaude Opus 5.5