Layouts placed exactly at the part spacing can land ~1e-4 short once
rotated, rounded (e.g. PEP's 4-decimal exports) and snapped to the
Clipper grid, so both validators rejected layouts that were correct in
practice. NestTolerances.SpacingSlack (0.0005, far below anything a
cutting machine resolves) is now subtracted from the spacing by
NestLayoutCheck's inflation and NestJobPlacementValidator's edge-distance
check. The frozen LegacyNestValidator takes the same rule so the
equivalence tests keep comparing like with like.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The benchmark validator and every engine test re-triangulated both parts
for each nearby pair. NestLayoutCheck now uses TriangulatedRegion, with
Collision.HasOverlap as the fallback when it cannot decide. Verdicts are
unchanged (the frozen-validator equivalence tests still pass); validating
100 discs went from 1,254 ms to 94 ms.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The job runner already checked sheet bounds on material contours only,
but the benchmark validator and salvage scoring used Part.BoundingBox,
which includes scribe/etch moves. A PEP bend tick that ends a hair past
the part's edge passed the runner yet failed the benchmark when placed
flush to the sheet edge, and it could shrink the credited offcut. Marks
only mark the surface, so bounds and salvage now use material only.
Benchmark before/after (all five built-in engines, local fixtures,
salvage 0.5): no job changed validity or cost. Regression tests pin the
new rule: a protruding tick flush to the sheet edge is valid in all four
quadrants, and a tick past the parts envelope no longer shrinks salvage
(targeted fixture cost 130 -> 120).
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Engines had to reverse-engineer the benchmark validator: Opus55 assumed a
0.01 arc tolerance (the validator uses 0.001), Gpt6Astra added hand-tuned
paddings and copied the validator's check order, Qwen picked its chord
tolerance to stay under a constant it could not reference.
NestTolerances publishes the validator's arc tolerance, the Clipper grid
and SafeClearanceMargin (with its derivation). NestLayoutCheck moves the
benchmark NestValidator's checks into OpenNest.Engine as a public API
(Clears for a part pair, Violations for a whole result); NestValidator is
now a thin wrapper. Verdicts are unchanged: tests compare ordered
violation lists against a frozen copy of the old validator, and a
tangent-disc stress test covers 432 pairs at the safe margin.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
All three plugin engines turned a RotationPolicy into trial angles by hand
(fixed angle, stepped sweep, or right angles plus the minimum-bounding-
rectangle angle for Automatic), each with its own normalization, dedup and
sweep caps. EnumerateAngles gives one deterministic, Allows-checked list;
RotationCandidates.ForShape adds the MBR-aligning angles via the existing
Polygon.FindBestRotation, and DistinctOutlines drops angles where the part
looks identical. A cap of one returns the sweep start rather than throwing,
since engines request a single sample for small orientation budgets.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Engines optimized guesses at the benchmark cost: Opus55 re-implemented
salvage credit, Qwen used plate area per part area, Gpt6Astra ignored
salvage. NestJobCost moves StockLadder's EstimateNetArea into a public
home (net sheet area, unplaced-part penalty, whole-result Evaluate) and
the benchmark and StockLadder now call it. Scores are unchanged: tests pin
it against a frozen copy of the old computation and real benchmark runs.
Bounds still include marks, as before, so scores do not move.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every plugin engine rebuilt part material from a snapshot by hand and
filtered only rapids, so all three kept counting scribe/etch marks as
material after 1b5e1b1 fixed it in the host. JobPartGeometry is the
validator's own reader made public: SpecialLayers.IsMaterial, validated
closed contours, material area, and TryRead returning null for unreadable
parts. The job validators now use it, so engines and validation agree.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Engines assembled NestJobResult by hand - instance indices, fulfillment,
stock usage, status and PlateCommitted progress - and Qwen38FlashNext got
PlateIndex wrong (stock index instead of sheet order). The builder assigns
plate and instance indices itself and rejects overproduction and exhausted
stock, so engines only decide placements.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every plugin engine re-derived the quadrant/edge-spacing work area by hand
(Gpt6Astra, Opus55 and Qwen each had a copy, as did the placement
validator). One definition on the stock removes that duplication and the
chance of an engine disagreeing with the validator's bounds.
Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every nesting-geometry consumer filtered only rapids, so scribe/etch
moves counted as part material. An etch tick that ends a hair outside
the outline (PEP bend ticks start on the notch edge) made the part
"open geometry leaving the material region": the job validator threw
and every built-in engine plus Gpt6Astra crashed on real PEP jobs
(PT75, drawing 4980 A01 PT77). Marks are only on the surface, so they
should never affect placement, collision, area, or validation.
- SpecialLayers.IsMaterial excludes Rapid and Scribe; used by drawing
area, canonical angle, part collision, PartGeometry, plate perimeter,
best-fit/pair evaluation, rotation analysis, GPU evaluators, and both
validators. Timing, display, splitting and posts still see marks.
- ConvertGeometry also maps the saved SCRIBE layer name to Scribe, so
programs rebuilt from stored entities keep their marks.
- NestReader repairs older files (e.g. PepNestExport output) whose
programs saved etch as cut moves while source entities kept SCRIBE.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every DefaultPlateFiller.Fill makes a fresh canonical copy of the drawing,
and BestFitCache/FillResultCache keyed by drawing reference, so fills never
shared results and the static caches grew without bound.
CanonicalFrame now records which drawing each canonical copy came from.
Both caches key weakly on that source drawing, so every canonical copy
shares one entry and released drawings can be collected. An entry is
dropped when the drawing's Program instance or canonical angle changes.
Best-fit candidates are computed once per (drawing, spacing) through
BestFitFinder.FindCandidates and filtered per plate size with the same
filter FindBestFits uses. FillResultCache keeps canonical and
non-canonical callers apart.
Adds Debug-only PerfCounters for best-fit runs, offset perimeter builds
and Part.Intersects calls.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The candidate validator flattened every arc into 1000 segments and rebuilt
both parts' polygons and edge lists for every pair it compared, so spacing
checks on filleted parts cost millions of edge pairs each. Validation, not
the fill pipeline, was nearly all of a solve's wall time.
Placed contours are now flattened once with ToPolygonWithTolerance(0.001),
the tolerance the benchmark NestValidator and Part.Intersects already use,
and each part's shape is built once per candidate. Arcs stay inscribed, so
a layout placed exactly at the spacing still passes.
12-nest PEP corpus, Default + StockLadder, --parallel 1: 2820 s -> 227 s.
Every run that finished before gives the same validity, count, plates and
cost. Three StockLadder runs that used to hit the 5-minute timeout now
finish.
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>
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>
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>
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.
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.
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>
Bounding-box distance is a conservative lower bound on true contour
clearance, so pairs far apart can skip the polygon Overlaps/Distance
checks without letting an overlap or spacing violation through. Also
gate the per-contour-pair BoundaryDistance work on a running minimum.
Validation is on the hot path for every candidate placement.
ShrinkAxis.Length must shrink the Y extent and ShrinkAxis.Width the X
extent to agree with MeasureDimension/TrimToCount; Box's constructor
takes the X extent first. The estimate previously shrank the wrong
dimension and returned a mis-ordered box. Add a parameterized regression
test over both axes and both translated-remnant orientations.
Delete OpenNest.Engine/Nfp (AutoNester, BottomLeftFill, NfpCache,
SimulatedAnnealing, INestOptimizer, PlacedPart, SequenceEntry), the Core
InnerFitPolygon, and the NestPhase.Nfp member. None had callers outside the
folder: console --autonest and MCP autonest_plate call engine.Nest(), not
AutoNester.
Drop the Nfp cases from NestPhaseExtensionsTests, fix the --autonest help
text, and update CLAUDE.md. NoFitPolygon stays; BestFit pair evaluation
still uses it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
PairFiller.EvaluateCandidates wrapped its candidate loop in
FillStrategyRegistry.SetEnabled(...) and reset it with SetEnabled(null)
afterwards. Nothing inside that window re-enters the strategy pipeline
(EvaluateCandidate and FillRemnantBox only use FillLinear), so the
restriction had no effect on the running solve. It did mutate process-global
state: concurrent solves (e.g. benchmark --parallel) could observe the
narrowed strategy list, and the reset cleared any filter set elsewhere.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Repo-wide sweep with the pinned CSharpier 1.3.0 tool. Whitespace and
line-wrapping only; OpenNest.Engine.Tests (109) and OpenNest.IO.Tests
pass after reformat, full solution builds 0 errors.
Added .csharpierignore so csproj/config XML keeps its existing layout
(CSharpier's XML wrapping churns attributes with zero benefit).
Formatting is now enforceable: dotnet csharpier check . passes.
The witness-probe overlap test missed small corner intersections: its
candidate points (crossing-edge midpoints and vertex-centroid midpoints)
can all land on a part boundary or outside the intersection, so two 10x10
parts at (0,0) and (9,9) with zero spacing were accepted despite sharing
a 1x1 unit of material.
Route the overlap decision through Collision, which clips triangulated
polygons and keeps only positive-area regions, catching corner overlaps,
containment, and coincident poses while legal edge/corner contact stays
legal. Collision's hole subtraction was conservative (partially-clipped
triangles were kept whole), so a part inside another part's cutout could
false-positive depending on triangulation alignment; subtract holes
exactly instead: a piece outside a convex hole triangle is the union of
its clips against each edge's outside half-space.
aj
