Three fixes to TryUpgradeOrNewPlate and a new post-pass:
1. Change ShouldUpgrade from < to <= so upgrade wins when costs are
tied (e.g., all zero) — previously 0 < 0 was always false
2. Guard against "upgrades" that shrink a dimension — when options are
sorted by cost and costs are equal, the next option may have a
smaller length despite higher width (e.g., 72x96 after 60x144)
3. Revert plate size when upgrade fill fails — the plate was being
resized before confirming parts fit, leaving it at the wrong size
4. Add TryConsolidateTailPlates post-pass: after all nesting, find the
lowest-utilization new plate and try to absorb its parts into
another plate via upgrade. Eliminates wasteful tail plates (e.g.,
a 48x96 plate at 21% util for 2 parts that fit in upgraded space).
Real nest file: 6 plates → 5 plates, all 43 parts placed.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Convert static class to instance with private constructor; shared
parameters (template, plateOptions, salvageRate, minRemnantSize,
progress, token) become fields, eliminating parameter threading
across all private methods (10→3 params on Nest entry point stays
unchanged; private methods drop from 7-9 params to 1-2)
- Extract FillAndPlace helper consolidating the repeated
clone→fill→add-to-plate→deduct-quantity pattern (was duplicated
in 4 call sites)
- Merge FindScrapZones/FindViableRemnants (98% duplicate) into single
FindRemnants(plate, minRemnantSize, scrapOnly) method
- Extract ScoreZone helper and collapse duplicate normal/rotated
orientation checks into single conditional
- Extract CreateNewPlateResult helper for repeated PlateResult
construction + PlateOption lookup pattern
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The consolidation pass was iterating stale remnant lists after placing
parts, causing overlapping placements. Now recalculates remnants from
the plate after each fill operation. Also added plate options to the
real nest file integration test.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Small parts no longer create their own plates during the main pass.
Instead they're deferred to the consolidation pass which fills them
into remaining space on existing plates, packing multiple drawing
types together. Drops from 9 plates to 4 on the test nest file.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
After the main single-pass placement, leftover items are now packed
together using the engine's multi-item Nest()/PackArea() methods
instead of creating one plate per drawing. First tries packing into
remaining space on existing plates, then creates shared plates for
anything still remaining.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Previously, parts classified as "Large" skipped all existing plates
and always created new ones. This caused one-unique-part-per-plate
behavior since most parts exceed half the plate dimension. Now large
parts search viable remnants on existing plates before creating new
ones, matching the intended part-first behavior.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Small parts must only go into scrap zones (both dims < minRemnantSize)
to preserve viable remnants. The implementer had inverted this, giving
small parts access to all remnants. Also fixed the test to verify
remnant preservation behavior and removed unused FindAllRemnants helper.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Implements the main Nest() method that ties together sorting,
classification, and placement across multiple plates. The method
processes items largest-first, placing medium/small parts into
remnant zones on existing plates before creating new ones. Includes
private helpers: TryPlaceOnExistingPlates, PlaceOnNewPlates,
TryUpgradeOrNewPlate, FindAllRemnants, and CloneItem.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds IsScrapRemnant(), FindScrapZones(), and FindViableRemnants() to
MultiPlateNester. A remnant is scrap only when both dimensions fall
below the minimum remnant size threshold (AND logic, not OR).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Introduces PartClass enum and Classify() static method that categorizes
parts as Large (exceeds half work area in either dimension), Medium
(area > 1/9 work area), or Small.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements static MultiPlateNester.SortItems with BoundingBoxArea and Size sort orders, covered by two passing xUnit tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When an item was too large for every plate option, its dimensions dominated
the global min-dimension filter, causing all candidate plates to be rejected.
This made auto-nesting exit immediately with no results even when the other
items could fit. Oversized items are now excluded from the filter so the
remaining items nest normally.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Try all valid best fit pairs instead of only the first when qty=2,
picking the best via IsBetterFill comparer (fixes suboptimal plate
selection during auto-nesting)
- Pre-compute best fits across all plate sizes once via
BestFitCache.ComputeForSizes instead of per-size GPU evaluation
- Early exit plate optimizer when all items fit (salvage < 100%)
- Trim slide offset sweep range to 50% overlap to reduce candidates
- Use actual geometry (ray-arc/ray-circle intersection) instead of
tessellated polygons for slide distance computation — eliminates
the massive line count from circle/arc tessellation
- Add RayArcDistance and RayCircleDistance to SpatialQuery
- Add PartGeometry.GetOffsetPerimeterEntities for non-tessellated
perimeter extraction
- Disable GPU slide computer (slower than CPU currently)
- Remove dead SelectBestFitPair virtual method and overrides
Reduces best fit computation from 7+ minutes to ~4 seconds for a
73x25" part with 30+ holes on a 48x96 plate.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When plate costs are equal (e.g. all zero), the optimizer now picks the
plate size with the tightest density instead of the smallest plate.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Strategies now promote results to IsOverallBest when they beat the
pipeline best, so the UI updates immediately on improvement rather
than waiting for each phase to complete
- PlateView only updates the main view on overall-best results, fixing
intermediate angle-sweep layouts leaking to the plate display
- Skip Row/Column strategies for rectangle parts (redundant with Linear)
- Intercept Escape key at MainForm level via ProcessCmdKey so it always
reaches the active PlateView regardless of focus state
- Restore keyboard focus to PlateView after fill progress form closes
- Remnant engines use SelectBestFitPair for orientation-aware pair
selection; DefaultNestEngine tries both landscape and portrait pairs
- RemnantFiller preserves more parts during topmost-part removal
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Tries each candidate plate size via the nesting engine, compares results
by part count then net cost (accounting for salvage credit on remnant
material), and returns the best option.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Strategies and fillers previously called NestEngineBase.ReportProgress
directly, each constructing ProgressReport structs with phase, plate
number, and work area manually. Some strategies (RectBestFit) reported
nothing at all. This made progress updates inconsistent and flakey.
Add FillContext.ReportProgress(parts, description) as the single
standard method for intermediate progress. RunPipeline sets ActivePhase
before each strategy, and the context handles common fields. Lower-level
fillers (PairFiller, FillExtents, StripeFiller) now accept an
Action<List<Part>, string> callback instead of raw IProgress, removing
their coupling to NestEngineBase and ProgressReport.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
BuildOffsets had Width and Length swapped after the Box axis correction
in c5943e2. Horizontal pushes used Length (X) for perpendicular sweep
and Width (Y) for push start — backwards. This caused part2 to start
inside part1's footprint, producing overlapping best-fit pairs.
Added regression test that verifies no kept best-fit pairs overlap.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Box constructor and derived properties (Right, Top, Center, Translate, Offset)
had Width and Length swapped — Length is X axis, Width is Y axis. Corrected
across Core geometry, plate bounding box, rectangle packing, fill algorithms,
tests, and UI renderers.
Added FillSpiral with center remnant detection and recursive FillBest on
the gap between the 4 spiral quadrants. RectFill.FillBest now compares
spiral+center vs full best-fit fairly. BestCombination returns a
CombinationResult record instead of out params.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The approach rapid from sheet origin was drawing to part.Location (the
program coordinate origin) then a second rapid to the actual first
pierce point. This created a dog-leg through the part origin instead
of a single straight rapid to the lead-in. Also fixed PlateProcessor
using the original program's start point instead of the processed one
when the cutting strategy is applied on-the-fly.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Change Nest() to decide fill vs pack based on total area coverage
instead of qty != 1. Items covering < 10% of the plate are packed,
so large parts get prime position and small low-qty parts fill gaps.
Qty=2 items are placed as interlocking best-fit pairs in remnant
spaces after the main pack phase, rather than as separate rectangles.
- Add ShouldFill() capacity-based heuristic
- Split pack phase: regular items pack first, then pairs
- Add PlaceBestFitPairs() for Phase 3 remnant pair placement
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
For qty 1-2, skip the full 6-strategy pipeline: place a single part
or a best-fit pair directly. For larger low quantities, shrink the
work area in both dimensions (sqrt scaling with 2x margin) before
running strategies, with fallback to full area if insufficient.
- Add TryFillSmallQuantity fast path (qty=1 single, qty=2 best-fit pair)
- Add ShrinkWorkArea with proportional dual-axis reduction
- Extract RunFillPipeline helper from Fill()
- Make ShrinkFiller.EstimateStartBox internal with margin parameter
- Add MaxQuantity to FillContext for strategy-level access
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Skip ExtentsFillStrategy for rectangle/circle parts
- Skip PairsFillStrategy for circle parts
- PackBottomLeft now tries rotated orientation when items don't fit
- PackBottomLeft tries both area-descending and length-descending sort
orders, keeping whichever places more parts (tighter bbox on tie)
- Add user constraint override tests for AngleCandidateBuilder
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace RotationAnalysis.FindBestRotation with PartClassifier.Classify in
RunPipeline, propagate ClassificationResult through BuildAngles signatures and
FillContext.PartType, and rewrite AngleCandidateBuilder to dispatch on part type
(Circle=1 angle, Rectangle=2, Irregular=full sweep).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace verbose value tuple with named PartPair record struct, extract
AnchorToWorkArea/PairBbox helpers to eliminate duplication, and delegate
RepeatColumns to FillLinear.Fill which already handles geometry-aware
column tiling with overlap fallback.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
FillGrid had no overlap check after perpendicular tiling of the row
pattern (Step 2), unlike Step 1 which had one. When geometry-aware
FindPatternCopyDistance underestimated row spacing, overlapping parts
were returned unchecked.
Changes:
- Make FillLinear.HasOverlappingParts shape-aware (bbox pre-filter +
Part.Intersects) instead of bbox-only, preventing false positives on
interlocking pairs while catching real overlaps
- Add missing overlap safety check after Step 2 perpendicular tiling
with bbox fallback
- Add diagnostic Debug.WriteLine logging when overlap fallback triggers,
including engine label, step, direction, work area, spacing, pattern
details, and overlapping part locations/rotations for reproduction
- Add FillLinear.Label property set at all callsites for log traceability
- Refactor LinearFillStrategy and ExtentsFillStrategy to use shared
FillHelpers.BestOverAngles helper for angle-sweep logic
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Extract shared BestFitAxis helper parameterized by orientation,
eliminating 23-line duplicate in rectangle packing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
StripeFiller and FillExtents had identical 24-line overlap detection
methods; move to FillHelpers and delegate from both callers.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Extract shared FillAxis helper parameterized by orientation,
eliminating 34-line duplicate between horizontal and vertical fills.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Extract shared SortStrips helper parameterized by axis selectors,
eliminating 61-line near-duplicate between column and row sorting.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Part.Intersects: filter intersection points at a vertex of either
shape (was both), so edge-touching parts are not flagged as overlapping
- NestEngineBase.HasOverlaps: use epsilon-based bounding box pre-filter
consistent with FillExtents and Plate.HasOverlappingParts
- PartGeometry.GetOffsetPartLines: remove extra chordTolerance added to
spacing offset — was causing 0.002" gap beyond the intended part spacing
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Etch marks for up bends are now real geometry entities on an ETCH layer
instead of being drawn dynamically. They flow through the full pipeline:
entities → FilterPanel layers → ConvertGeometry (tagged as Scribe) →
post-processor sequencing before cut geometry.
Also includes ShapeProfile normalization (CW perimeter, CCW cutouts)
applied consistently across all import paths, and inward offset support
for cutout shapes in overlap/offset polygon calculations.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Shape.OffsetOutward produces inward offsets for certain rotated polygons,
causing geometry-aware copy distances to be too small and placing
overlapping parts. Root cause is in the offset winding direction
detection — this commit adds safety nets while that is investigated.
- FillLinear.FillGrid: detect bbox overlaps after geometry-aware tiling,
fall back to bbox-based spacing when overlaps found
- FillExtents.RepeatColumns: detect overlaps after Compactor computes
copy distance, fall back to columnWidth + spacing
- PairFiller/StripeFiller remnant fills: use FillLinear directly instead
of spawning full engine pipeline (avoids strategies with the bug)
- Add PairOverlapDiagnosticTests reproducing the issue
- MCP config: use shadow-copy wrapper for dev hot-reload
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
High-utilization pairs (>=75%) are no longer discarded for exceeding
the aspect ratio limit, since the material isn't being wasted.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
OptionsForm now shows checkboxes for each fill strategy, persisted via
the new DisabledStrategies user setting. FillStrategyRegistry exposes
AllStrategies and DisabledNames for the UI. MainForm applies disabled
strategies on startup via OptionsForm.ApplyDisabledStrategies().
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
FillExtents falls back to the unadjusted column when iterative pair
adjustment shifts parts enough to cause genuine overlap. StripeFiller
rejects grid results where bounding boxes overlap, which can occur when
angle convergence produces slightly off-axis rotations.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
PushBoundingBox left exactly partSpacing between bounding boxes, but the
geometry push inflates offset lines by halfSpacing + chordTolerance per
side (totaling partSpacing + 2*chordTolerance). The 0.002 overlap caused
RayEdgeDistance to return MaxValue for negative t, making the nearest
obstacle invisible and allowing the part to push through it.
Subtract 2*ChordTolerance from the BB push distance so the gap is wide
enough for the offset geometry lines to remain non-overlapping.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Replace raw Panel with EntityView (via SplitPreview subclass) for proper
zoom-to-point, middle-button pan, and double-buffered rendering
- Add draggable handles for tab/spike positions along split lines; positions
flow through to WeldGapTabSplit and SpikeGrooveSplit via SplitLine.FeaturePositions
- Fix OK/Cancel buttons hidden off-screen by putting them in a bottom-docked panel
- Fix DrawControl not invalidating on resize
- Swap plate Width/Length label order, default edge spacing to 0.5
- Rename tab labels: Tab Width→Tab Length, Tab Height→Weld Gap, default count 2
- Spike depth now calculated (read-only), groove depth means positioning depth
beyond spike tip (default 0.125), converted to total depth internally
- Set entity layers visible so EntityView renders them
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
PolygonHelper.ExtractPerimeterPolygon always used OffsetSide.Right
assuming CCW winding, but DXF imports can produce CW winding. This
caused the spacing polygon to shrink inward instead of expanding
outward, making parts overlap during nesting.
Now detects winding direction via polygon signed area and selects
the correct OffsetSide accordingly.
Also adds save_nest MCP tool and a BOM-to-nest builder utility
(tools/NestBuilder) for batch-creating nest files from Excel BOMs.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The NFP engine was a thin wrapper that delegated all single-drawing
operations to DefaultNestEngine. Remove the engine and its registry
entry to reduce dead code.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
RemnantFiller: add placed parts as a single envelope obstacle instead
of individual bounding boxes to prevent the next drawing from filling
into inter-row gaps. Remove the topmost bounding-box part to create a
clean rectangular boundary.
PairsFillStrategy: guard against recursive invocation — remnant fills
within PairFiller create a new engine that runs the full pipeline,
which would invoke PairsFillStrategy again causing deep recursion.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Make quantity trimming direction-aware: DefaultNestEngine uses TrimAxis
(virtual property on NestEngineBase) so HorizontalRemnantEngine removes
topmost parts instead of rightmost. Rename ShrinkAxis.Height → Length
for consistency with Width/Length naming used throughout the codebase.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Height shrink now uses HorizontalRemnantEngine (minimizes Y-extent)
and width shrink uses VerticalRemnantEngine (minimizes X-extent).
IterativeShrinkFiller accepts an optional widthFillFunc so each
shrink axis can use a different fill engine.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Evaluate pair candidates in parallel batches instead of sequentially
- Add GridDedup to skip duplicate pattern/direction/workArea combos
across PairFiller and StripeFiller strategies
- Replace crude 30% remnant area estimate with L-shaped geometry
calculation using actual grid extents and max utilization
- Move FillStrategyRegistry.SetEnabled to outer evaluation loop
to avoid repeated enable/disable per remnant fill
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace stored property setters (BestPartCount, BestDensity, NestedWidth,
NestedLength, NestedArea) with computed properties that derive values from
BestParts, with a lazy cache invalidated on setter. Add internal
ProgressReport struct to replace the 7-parameter ReportProgress signature.
Update all 13 callsites and AccumulatingProgress. Delete FormatPhaseName
in favor of NestPhase.ShortName() extension.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
