The exit point should be the corner farthest from the origin so the
perimeter (cut last) ends near the machine home. The mapping was
backwards — Q1 (origin bottom-left) was returning (0,0) instead of
(w,l).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Drop Circle.ToArcFrom (zero-sweep problem), keep Circle in shape
and handle in ConvertShapeToMoves with full-circle ArcMove
- Use point-distance tolerance for Arc.SplitAt instead of angle
comparison to avoid wrap-around issues at 0/2pi
- Simplify SplitAt return types to non-nullable tuple
- Add ArgumentException guard in ReindexAt
- Add throw for unexpected entity types in ConvertShapeToMoves
- Document absolute coordinate convention and shared references
- Clarify variable names for both replacement sites
The strategy output (lead-ins, start points, contour ordering) must be
saved in the nest file, so Apply() runs when parts are placed — not
during post-processing.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Exit point from plate quadrant, nearest-neighbor cutout
sequencing via ShapeProfile + ClosestPointTo, contour type
detection, and normal angle computation.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Rename Angle properties to ApproachAngle (avoid shadowing Math.Angle)
- Arc rotation from contour winding, not hardcoded CW
- Add winding parameter to LeadIn/LeadOut Generate methods
- Add exit point derivation from Plate quadrant
- Add contour re-indexing section (split/reorder at closest point)
- Add ContourType.cs and AssignmentParameters.cs to file structure
- Clarify normal direction convention
- Note SequenceMethod value 6 intentionally skipped (PEP numbering)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Lead-in, lead-out, and tab class hierarchy for CNC cutting
approach/exit geometry, using ShapeProfile + ClosestPointTo
for contour sequencing.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
"Length" is more natural than "height" for flat plate materials.
Renames the field on OpenNest.Geometry.Size, Box.Height property,
and all references across 38 files.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rename SizeDto.Height to SizeDto.Length so the serialized JSON
uses "width"/"length" which is more natural for plate materials.
The core Size struct still uses Height internally.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace three separate XML metadata files (info, drawing-info,
plate-info) and per-plate G-code placement files with a single
nest.json inside the ZIP archive. Programs remain as G-code text
under a programs/ folder.
This eliminates ~400 lines of hand-written XML read/write code
and fragile ID-based dictionary linking. Now uses System.Text.Json
with DTO records for clean serialization. Also adds Priority and
Constraints fields to drawing serialization (previously omitted).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
DXF files can have endpoint gaps at entity junctions that fall right at
the floating-point boundary of Tolerance.Epsilon (0.00001). This caused
shapes to not close, resulting in 0 area and 0% utilization in Best-Fit.
Added ChainTolerance (0.0001) for endpoint chaining in GetConnected and
Shape.IsClosed, keeping the tighter Epsilon for geometric precision.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Console app for running the nesting engine from the command line.
Supports plate size/spacing overrides, quantity limits, overlap
checking with exit codes, and benchmark-friendly flags (--no-save,
--no-log). The MCP test_engine tool shells out to this project.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The push algorithm's copy distance formula (bboxDim - slideDistance)
produced distances smaller than the part width when inflated boundary
arc vertices interacted spuriously, causing ~0.05 unit overlaps between
all adjacent grid parts.
Two fixes applied:
- Clamp ComputeCopyDistance to bboxDim + PartSpacing minimum
- Use circumscribed polygons (R/cos(halfStep)) for PartBoundary arc
discretization so chord segments never cut inside the true arc,
eliminating the ChordTolerance offset workaround
Also parallelized three sequential fill loops using Parallel.ForEach:
- FindBestFill angle sweep (up to 38 angles x 2 directions)
- FillPattern angle sweep for group/pair fills
- FillRemainingStrip rotation loop
Added diagnostic logging to HasOverlaps, FindCopyDistance, and
FillRecursive for debugging fill issues.
Test result: 45 parts @ 79.6% -> 47 parts @ 83.1%, zero overlaps.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When the greedy maximum-row tiling leaves a thin remainder, tries removing
the last row and re-filling the larger strip. Picks whichever total is
higher. Fixes cases where e.g. 4 rows + 11 remainder = 47 beats 5 rows = 45.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The GPU pair evaluator reported false-positive overlaps for all
candidates because the bitmap coordinate system didn't account for
Part.CreateAtOrigin's Location offset. When rotation produced negative
coordinates, CreateAtOrigin sets Location = -bbox.Location (non-zero),
but the offset formula assumed Location was always (0,0).
Two fixes:
- Rasterize bitmaps from Part.CreateAtOrigin directly (new FromPart
method) instead of separately rotating polygons and computing bbox,
eliminating any Polygon.Rotate vs Program.Rotate mismatch
- Correct offset formula to include the Location shift:
(Part2Offset - partB.Location) instead of raw Part2Offset
Also optimized post-kernel bounding computation: pre-compute vertices
once per rotation group and process results with Parallel.For, matching
the CPU evaluator's concurrency.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Pre-size TilePattern result list from estimated copy count to avoid
resizing. Remove redundant Program.BoundingBox() call and UpdateBounds()
in MakeSeedPattern — the Part's BoundingBox is already correct.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
max(aUpper_i - bLower_j) across all pairs simplifies to
max(aUpper) - min(bLower), computed in a single pass.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
FindCopyDistance and FindPatternCopyDistance were cloning entire Parts
(including deep Program copies) just to get offset locations for
GetLines. Compute offset locations directly instead. Also skip the
Pattern wrapper in TilePattern — clone parts directly.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
RotationDirection and direction filtering were recomputed on every
GetLines call. Pre-compute edges per PushDirection once in the
constructor so GetLines only translates cached edges.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Part.Clone() re-clones from the drawing's unrotated program, re-rotates,
and walks all CNC codes twice for bounding box — 4 O(c) passes per clone.
CloneAtOffset clones from the already-rotated program and computes the
bounding box arithmetically, reducing to 1 O(c) pass per clone.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Opposite hull edges (180° apart) produce identical bounding rectangles
(width/height swapped), so only half the edges need to be checked to
find the minimum area. Applied to the unconstrained overload only.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add bounding box rejection in HasOverlaps to skip expensive
Part.Intersects (CNC→geometry conversion) for non-adjacent parts.
Eliminates ~35% CPU in IsBetterValidFill for grid layouts.
- Optimize RayEdgeDistance: access Line fields directly instead of
property getters (avoids Vector struct copies), inline IsEqualTo
with direct range comparison (avoids Math.Abs), and precompute
deltas for reuse in interpolation.
- Cache line endpoints in DirectionalDistance outer loop to avoid
repeated struct copies in the inner loop.
- Add fill timer to ActionClone.Fill, displayed in PlateView status
bar as "Fill: N parts in M ms".
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Test code and fixture data now live together in the OpenNest.Test repo
at git.thecozycat.net/aj/OpenNest.Test.git.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
After the main fill, detect if the last column/row is an "oddball"
with fewer parts than the main grid. If so, remove those parts and
re-fill the remainder strip independently using all strategies
(linear, rect best-fit, pairs). Improves 30→32 parts on the test
case (96x48 plate with 30x7.5 interlocking parts).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
