GitHub (ajisaacs/OpenNest) now owns branches, tags and pull requests.
Gitea keeps a read-only pull mirror that syncs hourly and refuses pushes,
so PRs merged on GitHub are no longer closed by the old Gitea push mirror.
MaximalRectangles.InRegion finds the largest axis-aligned rectangles that fit
wholly inside a Clipper region, such as a cutout shrunk by the part spacing.
It grids the region on its vertex coordinates plus even divisions, keeps a cell
only when no edge enters it and an even-odd row scan puts it inside, then runs
the shared histogram search.
Exact for regions with only horizontal and vertical edges; slanted and curved
edges are followed as a staircase that never crosses the boundary, short of the
true maximum by up to about one cell per side. Tests cover a rectangle, an
L shape, a frame with a hole, a round hole (inscribed square), a diamond and a
star; disabling the edge-crossing check fails the three slanted-edge tests.
The histogram search over a free/blocked cell grid never depended on the
obstacles being boxes. Move it to OpenNest.Core as MaximalRectangles.FromGrid
so other callers can build their own grids; RemnantFinder keeps building its
obstacle grid and calls it. No behavior change.
Add DrawingAligner: bounded multi-start rigid ICP that maps a revised
drawing's geometry into the old drawing-local frame as an operator-review
seed. Material-filtered programs flatten through the shared chord-error
machinery, resample via the shared ContourSampler ring scheduler, and fit
closest bounded target segments with trimmed weighted least squares
(reflection never solved, never applied). Seeds come from outer-centroid
translation and MBR angle deltas with the 90-degree family plus identity.
The stable outer boundary is fitted first; retained hole evidence only
reorders candidates inside the outer tie band, so a moved hole cannot
drag the perimeter. Structured AlignmentResult reports convergence,
residual quantiles, bidirectional coverage, symmetry ambiguity,
reflection uncertainty, and a bounded diagnostic IoU. Refusals
(invalid input, insufficient support, unsupported topology, sample
budget, cancellation) are reported as reasons, not thrown.
Tests verify recovered poses against exact fixture geometry (true
segments/arcs/circles, not sample clouds) for known transforms,
symmetric rectangle/circle ambiguity, mirror non-selection, moved-hole
perimeter preservation, and bounded region arithmetic (identical=1,
disjoint=0).
Extract the pure world-space position/tangent math from the WinForms
cut-direction arrow renderer into Core (ContourSampler), keeping screen
conversion, arrowheads, and the per-move display policy in the view.
Add a contour-wide arclength scheduler (RingMoves) for measurement use:
distance carries across segment boundaries, short segments are never
omitted, and the closing vertex is not duplicated. Characterization
tests pin the arrow policy (counts, short-move skipping, CW/CCW and
full-circle sweeps, subprogram offsets, suppressed/rapid moves,
incremental mode) and the scheduler's start-vertex invariance.
The plates ListView showed stale part-count and utilization columns
after interactive placement, clone, or fill, because only the drawing
list was refreshed on the debounced PartAdded/PartRemoved timer. Also
refresh the plate rows when a drawing is deleted from every plate.
Parts drawn a few millionths over their sheet's work area (for example a
36.125006 panel on a 36.125 sheet with no edge spacing) pass NestLayoutCheck,
which allows Tolerance.Epsilon of overhang, but the Rectangles engine refused
them and left them unplaced. The engine now allows 90% of that slack: a box
that exceeds the sheet by no more than the allowance packs as exactly the
sheet's size, so it spans the axis and the overhang lands only past the edge.
The check's bounds slack is named NestTolerances.WorkAreaSlack (same value,
no behavior change) so the engine and the test layout assertion share it.
Rectangle-lane benchmark (91 strict + 77 boxable jobs): every layout valid;
4 jobs that left panels unplaced now place them (strict complete 86 -> 88,
boxable complete 74 -> 75, one more boxable job places 2 more parts); no other
job changed.
DefaultPlateFiller.PackArea (Default, Strip and both Remnant strategies)
now packs leftover parts with the shared maximal-rectangles packer instead
of bottom-left corner points. Every fit rule and pick mode is tried; the
layout placing the most parts, then box area, per priority tier wins, and
the strategy's own fill comparer breaks ties so remnant strategies keep
their clear side. PackBottomLeft and PackEngine are removed.
Rectangle-lane benchmark (91 strict + 77 boxable jobs) against the old
packer: no strategy lost a valid layout; Default, Vertical Remnant,
Horizontal Remnant and Strip complete 2-3 more strict jobs, and Strip
gains a valid one. Cost on jobs complete in both runs falls for every
strategy (Default -0.6% strict, -2.7% boxable). Default is about 7-10%
slower on the changed jobs.
Golden layouts for Default and both Remnant strategies are re-captured;
each is complete, passes NestLayoutCheck and repeats exactly.
Move the Rectangles engine's global and ordered pick loops into
RectanglePacking.MaxRectsPacker, which works on plain box sizes so other
fill paths can reuse it. SheetPacker maps part orientations onto it.
No behavior change: rectangle-lane benchmark results are identical job for
job (91 strict and 77 boxable jobs).
MaxRectsSheet only tracks free rectangles and has no job dependencies, so it
moves from the Rectangles engine into OpenNest.Engine.RectanglePacking where
the interactive fill packer can use it. No behavior change.
DxfImport_ArcBoundingBoxes_Diagnostic read a drawing from one user's
desktop, returned early elsewhere and asserted nothing, so the near-zero
sweep bounding-box fix (4053f1f) had no regression test that could fail.
Replace it with ArcBoundingBoxTests: a forward and a reversed arc with a
sweep below Tolerance.Epsilon must be bounded by its endpoints, and an
ordinary arc crossing 90 degrees must still reach the top of its circle.
With the 4053f1f guard reverted, both near-zero cases fail (the box grows
to the full circle); with it in place all three pass.
GeometrySimplifierTests hardcoded a customer DXF on one user's desktop and
returned early when it was missing, so the test passed without running
anywhere else. Read the path from the optional "SimplifierGapDxfPath"
entry in test-config.json and report a skip when it is not configured,
matching the other external-fixture tests.
Replace the customer part name in a RemnantFinderTests comment with a
neutral description. Customer drawings stay outside the repository.
EllipseConverter, SplineConverter and GeometrySimplifier each carried an
identical private SumSignedAngles. Move the unchanged body to
ArcFit.SumSignedAngles and call it from all three, keeping the ordered
accumulation, strict half-turn comparisons and empty/single-point result.
ArcFitTests compares the shared method bit-for-bit with a separate
test-local accumulator across half turns, the atan2 seam, multiple
turns, translated centres and NaN inputs, and checks that inputs are not
mutated. The converter winding characterization from 8664656 still
passes unchanged.
Desktop Auto Nest, Console --autonest, MCP autonest_plate and the API
NestRunner all run through NestPipeline, so nothing calls the old
orchestration any more.
Delete MultiPlateNester (with MultiPlateNestOptions, MultiPlateResult,
PlateResult, PartClass and PartSortOrder), PlateOptimizer and
PlateOptimizerResult, plus their tests. The explicit-strategy contract
those tests checked now lives at PlateFillService.ResolveStrategy, which
keeps its null-means-Default, canonical-name and unknown-name tests.
CreateFiller loses its internal visibility, which only the deleted
orchestrators used.
This breaks source and binary compatibility for external callers of the
removed types; whole-job callers use NestPipeline.Run or INestingEngine.
Skip FillGrid's Step 2 overlap check only when Step 1 checked this exact
row clean and the perpendicular tiling appended zero parts, so gridResult
still holds the same Part objects in the same order and poses. The local
rowIsVerified flag is cleared by every Step 1 skip or bbox fallback, and
any nonzero append (including partial-copy parts) keeps the eager check.
Debug work assertions pin the reduced overlap work (horizontal stripe
8->4 exact calls, 4->2 preparations, 16->8 triangulations; vertical
stripe 36->18, 4->2, 38->19) and keep single-seed, Horizontal full-grid,
partial-only and PerpOnly work at base; the invalid overlapping-seed
control pins both fallback stages plus the retained exact-call total.
Three mutants (forced eager check, dropped flag, complete-rows-only)
are caught by their designated controls. Serial whole-job layout is
byte-identical to the frozen 62b5a8d0 oracle before and after.
The reorder event snapshot could previously show the reverse pass's
stale timeout label. Set a sentinel before each click so the
ProgramChanged handler can only observe the label applied by the
current operation's highlighting pass.
Quality review of 50884da found the unshown RichTextBox oracle compared
cached CRLF text against the highlighted editor's native LF-normalized
text, which would fail on first Windows execution. Compare generated
text with normalized line endings on both sides instead, capture the
status label and preview state inside the ProgramChanged handler to
prove fallback-before-notify ordering, document the HighlightSpan
UTF-16 and rule-index contract, and pin the null-text argument check.
Address Task 3b quality review:
- FormatSize now emits invariant decimal notation so a comma-decimal
locale cannot produce stock text its own invariant parser rejects.
- Validation notices embed a bounded single-line preview instead of the
full cell value.
- Documented GetPlateOptions' all-or-nothing throwing contract.
- STA test helpers run the worker as a background thread with a
diagnostic timeout message; added a de-DE fractional stock round-trip
regression.
Adds docs/nesting-engines.md: which engine suits which jobs, the old plug-in
names the registry maps, retired engines, and the rule that an engine change
lands only when it beats that engine's current benchmark result with every
layout valid. README, release and automatic-nesting docs drop the bundled
plug-in wording; AGENTS.md now places built-in engines in
OpenNest.Engine/NestingEngines and keeps only external plug-ins out of the
solution.
Rectangles and Irregular now ship inside OpenNest.Engine.dll, so the Windows
package no longer fetches, tests and bundles OpenNest-Engines at a pinned
commit. Removes scripts/external-engines.json and the Engines/ folder,
manifest and license from the package; build-info.json drops enginesCommit.
ReleaseSmoke now checks the packaged registry: every built-in engine must
instantiate from the packaged OpenNest.Engine.dll, the Opus55NestingEngine
name must resolve to Irregular, and an unknown name must be rejected. Linux
check against a published OpenNest.Engine: pass case exit 0; missing engine
DLL exit 1. Full Windows packaging still needs the Windows runner.
Moves the two production plug-in engines into OpenNest.Engine under names
that describe the jobs they suit:
- Rectangles: plain and near-rectangular plates, maximal-rectangles box
packing (was the RectanglesNestingEngine plug-in)
- Irregular: irregular profiles, no-fit-polygon frontier packing (was the
Opus55NestingEngine plug-in)
Their tests and the shared engine contract/layout test kit move into
OpenNest.Engine.Tests/NestingEngines.
The registry maps the old plug-in names to the new engines, so saved desktop
selections, scripts and API requests keep working, and a leftover plug-in DLL
under an old name cannot shadow its replacement. Desktop startup passes the
registry's lookup when restoring the saved Auto Nest engine.
aj