DefaultPlateNester, StripPlateNester and RemnantPlateNester differed only
in the filler they built. They are replaced by one internal PlateNester
over a Func<Plate, PlateFiller>, which keeps the old Place exactly:
restricted rotation goes to OrderedPlateNester, and each trial gets a fresh
private plate and filler.
PlacementStrategies maps Fill (and its legacy "Default" alias), Strip and
both Remnants to a filler policy and layout. PlateNesterFactory reads it
case-sensitively and PlateFillService ignores case, as before; each keeps
its own error type and message. Nothing outside the Engine and its tests
used the three public nester types, so they are removed.
No layout changes: all 29 fill-stack pins pass with the pin file untouched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
DefaultPlateFiller becomes PlateFiller(plate, policy, layout); PlateFillerBase
and StripPlateFiller are gone. PlateLayout.FillThenPack is the old
orchestrator path (Fill and both Remnants); PlateLayout.Strips is the old
StripPlateFiller.Nest, and Strip is Fill policy plus Strips layout.
Lifetimes are unchanged: a strip-layout filler runs every Fill, group Fill
and PackArea on a fresh inner filler and gives each strip fresh remnant
fillers, so learned angles and PlateNumber never leak into them. The
comparer, preferred direction, trim axis and angle recording now read the
policy directly. The base class's unsorted two-angle BuildAngles is dropped:
only DefaultPlateFiller's pipeline called BuildAngles, and Strip never ran it.
Layout-neutral: all 29 FillStackCharacterizationTests pins pass unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fill, Vertical Remnant and Horizontal Remnant are now three values of an
internal PlateFillPolicy that DefaultPlateFiller takes (Fill by default),
replacing RemnantPlateFiller and RemnantFillPolicy. The policy holds the
comparer factory, preferred direction, trim axis and angle mode; the
remnant angle mode still replaces the angle builder outright, and the
public FillPolicy that strategies read is unchanged. Layout-neutral: the
fill-stack characterization pins pass unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NestLayoutCheck rejected Fill layouts whose interlocked arcs sat exactly
at the part spacing: its SpacingOutline circumscribes arcs by up to 0.001
each, more than the 0.0005 slack, so facing arcs overlapped. The runner's
placement validator measured inscribed 0.001 polygons instead and so
accepted gaps up to ~0.002 under the slack.
Both now find candidates with SpacingOutline and confirm them on
NativeContours, the native contours flattened at a twentieth of the slack
(moved out of PartSpacingReview, which keeps using them). The runner calls
NestLayoutCheck.KeepsSpacing, the same outline, tie rule and
confirmation, so the two validators accept exactly the same pairs.
Clears stays unconfirmed and stricter for engine certification. No
tolerance changed.
FixedStrategyEngine_IrregularJob pins now assert LayoutAssert.Valid with
unchanged digests; the NativeU 70-part grid's 28 reports were the same
cause and its test now requires none.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
BestFitFinder gathered each rotation strategy's candidates, and PairEvaluator
gathered evaluated results, in ConcurrentBags, so their order depended on which
worker finished first. The stable area sort then broke equal-area ties in that
arbitrary order, and fills, which try candidates in order and keep the first of
equal layouts, could nest the same parts differently between runs.
Both now fill index-addressed slots: candidates keep rotation-angle and
slide-offset order, results keep candidate order, and ties break the same way
every run. The candidates produced and their scores are unchanged.
The Strip whole-job pin in FillStackCharacterizationTests is re-pinned
(A66602F707290A36 -> FD03A4529D37E32C). That job ties between best-fit
pairs; the old pin recorded whichever tied pair the bag happened to return
first. The new layout places the same 26 parts, passes LayoutAssert.Valid
and gave the same digest in five separate test processes. The other 28
pins are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
StripPlateNester sent every request to the strip fills, which try two
classification angles and ignore rotation bounds. With a fixed or bounded
rotation at quantity above one they proposed forbidden poses, the job
validator threw "Candidate rotation is not allowed for the requirement."
and the whole Strip job failed. Strip now uses the same guard as
DefaultPlateNester and hands such plates to OrderedPlateNester.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Characterization digests for the interactive PlateFillService surface and
the fixed-strategy whole-job engines on L-shaped, notched, arc, bar and
chevron parts: single-item fills, whole-plate nests (including one where
learned angles change the layout, guarded by a full-sweep comparison),
fills beside clamp zones, and one job per Fill/Strip/remnant engine.
They let the planned filler/nester consolidation be shown layout-neutral.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every whole-job engine chooses among the offered stock sizes, and on the
synthetic benchmark nests Default matched or beat StockLadder on all 18
runs (saved sizes, a six-size pool, with and without salvage credit),
costing 30-40% less in total with mixed sizes. Its sheet-swap step never
produced a win over Irregular's tail re-plan.
The registry maps "StockLadder" to Default so saved selections, scripts
and API requests keep working. OrderedPlateNester stays: Fill and the
remnant strategies use it for restricted rotations, and its golden layout
is now pinned through NestJobRunner with the parts requested largest
first, as StockLadder requested them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Auto Nest's Stock Options gets a Load from ledger button that reads the nest
material's free sheets and drops from MaterialLedger's API-key-protected
GET /api/stock/opennest. Ledger rows are request-local: the engine sees them as
ordinary NestPlateStock rows with ledger:{n} ids and finite quantities, and they
are never written to Nest.PlateOptions.
- OpenNest.Data/MaterialLedger: settings file, read-only client (X-Api-Key,
5 s timeout, closed LedgerStockResult set) and LedgerStockMapper (bundles and
sheets one row each, same-size drops grouped with ordered item codes, unknown
shapes and too-small items skipped, inch/mm conversion).
- Options: MaterialLedger URL and API key (DPAPI, CurrentUser) with Test.
- Auto Nest grid: Use, Qty and Source columns for every row; PlateOption gains
request-local Quantity and Id, which NestStockBuilder passes to the engine.
Ledger rows score by area; hand-typed costs stay saved.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Without salvage credit (desktop Auto Nest on a single plate size) every
complete one-sheet layout costs the same, so engines kept whichever layout
came first. N26-23S nested 27.86" long when turning two parts fits it in
21.3".
NestJobCost.CompactnessOffcut measures the largest full-width/full-length
edge offcut as salvage does, whatever the salvage rate. Default and Circles
(CandidateScore), Rectangles and Circles sheet choice (SheetByStock) and
Irregular sheet and plan ranking use it only after cost, so no cost changes.
Sheet-level keys round to 1e-6 to keep ties platform-independent.
Rectangles layout digests are re-captured: each changed fixture keeps its
sheets, parts and cost and leaves a larger offcut.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Roslyn duplicate scans found the same code copied across layers:
- Data: the three browse sessions share BrowseSession<TPage, TSort>;
the remote clients share NestServerHttp (JSON options, If-Match,
error text).
- WinForms: five paged lists share BrowsePager for loading text,
navigation, sort glyphs and failure messages. The Add Drawings to
Release, Drawing Library and set browser dialogs now dispose their
browse sessions.
- Server: NestQueryRequest reuses DrawingLibraryRequest's helpers;
revision history queries share count and paging binds.
- Engine: RemnantPlateNester delegates to DefaultPlateNester;
CuttingPlanService reuses JointCuttingPlanSearch.Map.
- Core: LineArc and CleanHole lead-ins share LineArcApproach with
unchanged arithmetic order, so output is bit-identical.
- IO: Dxf.GetGeometry overloads share one reader; .nest JSON test
helpers move to NestArchiveJson.
Also fixes the production compiler warnings (CS8632, CS8603, CS0108)
and moves a test off an obsolete forwarding API.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fill only tried 0 and 90 degrees for parts classified as rectangles, so a long
part filling a 60 x 120 plate four high at 90 degrees left a 55 x 59 strip empty
although two more fit tilted about 33 degrees (a reported nest: 4 placed, 6 fit).
The second tilted part only fits with its corner tucked beside the grid, in the
strip the grid's slack leaves, so a plain rectangular remnant fill finds one.
TiltedRemnant runs after the other strategies. It moves the best layout flush to
one end so leftover room becomes one remnant, and to either end of the other axis
so its slack forms one strip; then tiles the part at the most upright angles that
fit the remnant's height and keeps only copies at least the part spacing from the
layout. It is skipped when the part fits the remnant upright, for circles, for an
explicit rotation range and once the requested quantity is placed.
Fill benchmark (--parallel 1): the synthetic set is unchanged (all valid, same
cost and plates). The reported part needs 1 sheet instead of 2 for 6 parts and
2 instead of 3 for 12.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
TryImprove kept a swap when the trial run looked fuller than before, but
the trial charges each type's self-nesting gain again even when the run
already holds that type. Trading a placed copy for an identical unplaced
one therefore looked fuller by the gain and repeated forever: FlatBar ran
into the benchmark's 5-minute timeout on tabbed_cog_layout once the
all-edges MBR fix gave one part new rotations (it finished in 7 s before).
The swap is now judged on the merged run, so every kept swap really
leaves the run fuller and the loop terminates. Swaps the old test kept
that do fill the run are still kept.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Desktop Auto Nest's "Best plan so far" strip now fills in for every built-in
whole-job engine except FlatBar, not only Irregular.
- Rectangles makes one sheet-by-sheet run, so it reports the finished job as
its one plan, then replays its plate commits (token checked as before).
- Circles reports when a stock-preference run or a tail re-plan beats its
best result so far.
- Default scores each candidate plan as it ranks results (valid, fewest
unplaced, lowest cost, candidate order) and reports it only when it beats
every plan already shown, so a later candidate never shows a worse plan.
The chosen result is always the last plan reported.
- Shared ReportPlan/SameLayout helpers on CandidateProgress; Irregular uses
them too. Reporting never changes the search.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Auto Nest no longer draws parts on the plate view before they are kept:
its result is a set of sheets, possibly of several sizes, not parts on the
current plate. Ghost parts stay with interactive Fill and remnant tools.
- NestJobStage.PlanImproved carries an engine's best whole-job plan so far;
Irregular reports one whenever a variant or the tail re-plan beats it.
- NestPipeline adds PlanSheets: the plan on detached drawings built from
the job geometry (with the caller's names and colors), so previews never
touch caller drawings or quantities.
- SheetStripControl draws every sheet side by side at one scale, repeats
once with their count, captioned with size, parts and utilization.
- The progress window shows "Best plan so far"; a review window shows the
final sheets, totals, cost summary and unplaced parts before commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NestProgress gains PreviewSheet, a detached plate for the stock the preview
parts lie on; Irregular builds one per stock. PlateView draws it (size, edge
spacing, clamps) in place of the plate while the preview is shown and drops
it whenever the preview is cleared or the plate changes. Auto Nest zooms out
when a trial sheet falls outside what it has already shown.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The frontier packer reports its placements after each choice; the engine
turns them into preview parts at most four times a second and sends them as
NestProgress.BestParts with the stock work area, which desktop Auto Nest
already draws on the plate view. Reporting only reads finished placements,
so layouts are unchanged with or without a progress sink.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every Auto Nest run, and every Default-engine run, recomputed each part's
best-fit pairs: job snapshots are fresh objects and BestFitCache keyed
its entries on Drawing reference, so the Irregular engine paid
PairCatalog.Build (about half of a 118-part solve) on every run of the
same drawings.
BestFitCache now keeps a process-wide, bounded (512 sets, least recently
used first) pool of unfiltered candidate sets keyed by the exact content
of a PartGeometrySnapshot (mode and every motion value, bit for bit;
PartGeometrySnapshot.ContentComparer) and the spacing. Every drawing
DrawingJobMapper.CreateDrawing reconstructs from a snapshot declares
that geometry (BestFitCache.ShareByGeometry), so the Irregular pairs and
blocks, the cutout lattice and the Fill/Default job paths all read and
fill the pool. Candidates depend on the program geometry and spacing
alone, so a pooled set is the set the search would produce: benchmark
layouts, costs and placements are unchanged on the synthetic nests (all
engines) and on the 67-drawing, 118-part job. Failed searches are not
pooled, Invalidate leaves the pool alone, Clear empties it, and a
drawing whose Program is replaced stops vouching for its geometry.
Same 118-part job solved twice in one benchmark process (--parallel 1):
the second solve now takes 13.1 s after a 33.0 s first; before, 35.6 s
after 41.8 s.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A 67-drawing, 118-part job took about six minutes through the desktop
Auto Nest path and was cancelled before any layout appeared. Almost all
of it was the cutout pre-pass, the rest the Irregular engine.
Cutout pre-pass (317 s -> 8 s on that job, same bundles and poses):
- The lattice and NFP searches prepare the frame's and the occupants'
validation outlines once per search (NestLayoutCheck.Prepare) instead
of re-outlining and re-triangulating the frame at every candidate, and
test candidates against the small occupants before the frame. Answers
match the pairwise check exactly, including its leftmost-inflated rule.
- A hole's proposal is replayed for the same frame, hole and offered
inserts: automatic-rotation inserts ask the same question at every
frame angle.
Irregular engine (40 s -> 15 s, benchmark 42 s -> 19 s, same layouts):
- Concave no-fit polygons are built as the union of pairwise convex sums
of each footprint's convex pieces (ear clipping plus Hertel-Mehlhorn
merging on the Clipper lattice) instead of n*m edge-pair quads; a
footprint that cannot be decomposed keeps the old sweep.
- Pair and block catalogs share one private drawing per part geometry,
owned by the engine instance, so each part's best-fit search runs once
per instance and across a pipeline's trial and ordinary solves.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Dropping the parts that land on a clamp suits small parts, which fill the
gaps between clamps. Large parts often do better if the fill area stops
short of the clamps. The rows still fit, and no part is lost on a clamp.
When every clamp in the fill area lies on one edge, interactive Fill now
also fills the area with that edge moved just past the clamp keep-outs,
as a larger edge spacing would. It keeps whichever layout the strategy's
comparer ranks better, and a tie keeps the dropped-parts layout. Clamps
on more than one edge skip the comparison. The second fill reuses the
cached best-fit pairs. While it runs, a preview counts as the new best
only when it beats the first layout, so the preview never gets worse.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Interactive Fill on a plate with clamps used to split the plate into
clamp-free rectangles and fill each one separately. It showed no progress,
so Accept did nothing until the end, and Stop discarded the result. Packing
around the clamps also tends to give a worse layout than losing the few
parts that land on them.
FillItem and FillGroup now fill the area as if there were no clamps. They
then drop each part that enters a clamp keep-out or sits too close to an
existing part, using the same check as before. For a multi-part group, the
whole copy is dropped when any member fails. Progress previews are filtered
by the same check, so a layout accepted early is clamp-clear. A quantity
request that loses copies asks for that many more, up to two more times.
Pack and whole-plate Nest still pack only into the clamp-free areas.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Circles packs every part as its minimum enclosing circle at centers tangent
to two of the work-area walls, placed discs and clamp keep-outs, bottom-left
and left-bottom, with hexagonal and square lattice seeds. It bids only when
every part fills at least 0.8 of its enclosing circle and otherwise returns
everything unplaced at once. Default runs it as a third candidate.
Rectangles' sheet-by-sheet stock loop moves to NestingEngines/SheetByStock
(Rectangles output unchanged) and Default's candidate scoring and commit
forwarding become shared helpers. Circles also runs the whole job preferring
each offered stock and keeps the best result. Contract tests take shapes an
engine can override.
Synthetic round jobs, salvage-credited: Circles wins or ties all six with
valid layouts; Default cost falls about 4% versus Irregular alone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The cutout prepass now bundles frames smallest first, so a larger frame's
round cutout can take a ring part or an earlier bundle as one insert through
the circle packer: a flange holds a washer that holds a smaller washer that
holds a slug. Chains stop at four parts deep. Lattice and NFP proposals never
take these inserts, since they could seat a part in a ring's bore.
Expansion and released demand recurse, so placed order cuts innermost first
at every level and an unplaced outer frame returns its whole chain to demand.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add CutoutCirclePacker, a circle-in-circle proposal for the cutout
router. When a cutout is one circle and every insert fills at least 0.8
of its enclosing circle, it packs the inserts largest first at centers
tangent to the wall or placed discs, trying a center-out pass (a single
insert sits concentric) and a wall-first pass (optimal counts for up to
nine equal discs). Every pose passes NestLayoutCheck.
The router runs it first and skips the lattice and NFP searches when it
places all demand; otherwise the higher count wins and existing
proposals keep ties. On synthetic disc jobs the prepass drops from
minutes to seconds and places as many or more inserts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Item.Create builds the circle packer's view of a part: the minimum
enclosing circle of its outward-flattened perimeter (padded by the
layout check's flattening allowance), its fill ratio, the offset back
to the part origin, and the largest disc clear of material inside its
cutouts. Gears and sprockets pack by their tip circle around their
bore; the fill ratio lets a future engine decide which jobs to bid on
instead of a yes/no round test.
Adds MinimumEnclosingCircle (Welzl, repeatable order) to Core geometry.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Staggered stripes were spaced only so adjacent stripes touched
diagonally. When the in-stripe pitch stretched past sqrt(3)*D, stripes
two apart, which sit directly over each other, came closer than a
diameter. Clamp the stripe spacing to at least a radius in both fills.
FillEndOdd also divided by zero when a single circle fit per stripe;
it now stacks those stripes a diameter apart. FillEndEven now clones
the template item, so placed circles keep its radius instead of zero.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Reads each stock as a bar and cuts parts in one row along its length, packing
part lengths with a port of the CutList saw cut-list packer (first-fit
decreasing with swap optimization, exhaustive search for small jobs) plus a
best-fill subset-sum candidate. Each part is packed at its length less its
measured self-nesting gain, and a slide-left compaction pass nests shaped ends
(turning alternate mitred parts) and refills the freed length. Every slid pose
is confirmed with NestLayoutCheck.Clears.
On five synthetic flat-bar jobs every layout is valid and complete; total cost
33696 vs Rectangles 36816 and Default 34080.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A tabbed cutout's slug stays attached while the sheet is cut, but nesting
saw a closed hole and could place parts inside it. Nesting and layout paths
now read material through NestingMaterial.ForNesting, which drops closed
contours with tab spans so the slug counts as part. That covers part
outlines and collision lines, the cutout router, fill expansion, best-fit,
NestLayoutCheck and the GPU bitmaps. Area, overlap diagnostics and cutting
keep the hole, and programs without tabs read exactly as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NestPipelineResult.Unplaced lists each requirement's unplaced copies with a
typed NestUnplacedReason, counted from the returned placements: no stock fits
at the sampled rotations, reserved in a cutout of an unplaced frame, stock
exhausted, plate limit, no position found, or proposal not keepable. Desktop
Auto Nest, Console --autonest and MCP autonest_plate/autonest_job report them;
the API returns NestResponse.Unplaced, persisted from response schema 5.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NestPipeline.Run now bundles parts that fit other parts' enclosed cutouts
into their frames before the selected engine, for desktop, Console, MCP and
API alike. A valid complete bundled trial is kept; a refused trial falls back
to the ordinary solve (reason on RefusedCutoutTrial), and an incomplete trial
is compared with it, keeping whichever valid proposal places more. An
incomplete but valid trial is now keepable like any incomplete result.
Expanded sheets list each frame's inserts before the frame, so the committed
order cuts every insert before the cutout that releases its slug.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Console and Mcp each re-derived whether a name was a fill strategy, a
whole-job engine or unknown. EngineNameResolver now owns that decision,
trims consistently and accepts renamed legacy engine names; front ends
keep their own message wording.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ZeroClampMixedShapeLayoutIsBitIdenticalToBaseline has failed on Windows since
it was added in de9a087 while passing on Linux CI (bisect pointing at
c374e02 was an artifact: the assumed-good start commit fails too). No
commit changed the layout; the two platforms always chose differently.
In that fixture two candidate sheets (ContactPoint/Global and
BestShortSide/Ordered) both place all 30 parts at the same cost, so Choose
falls to its "more material placed" tie-break. SheetPacker summed material
area in placement order, and the disc's area differs in the last bit
between the Windows CRT and glibc math libraries. On Linux the two orders
rounded to ...253 vs ...257, so the later candidate won on rounding noise;
on Windows both rounded to ...257 and the tie fell to enumeration order as
Choose documents.
Sum a plan's area per type in type order instead (the same form as the
engine's demand area), so plans placing the same parts get bit-identical
areas and exact ties keep enumeration order everywhere. No tolerance is
loosened. The Windows layout is unchanged; Linux now matches it, so the
frozen hash moves to the Windows value, verified identical on both.
Verified: full Engine.Tests pass on Windows and Linux (1105 passed, 2
skipped); Rectangles and Default benchmark results on the six synthetic
nests are identical before and after (validity, placed parts, cost,
utilization, plates).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Circumscribing every arc holds all material along convex arcs but drops a
sliver along concave ones (a notch), so a tiny clamp 6e-7 into the material
beside a notch read as clear. Tiny-clamp checks now test the part flattened
both ways (all arcs inscribed, all arcs circumscribed): along any arc one of
the two holds the real material, so together they find every real overlap.
PM 407f5eb7-329f-41e4-947e-64ad52aa4034 (delta review s6-delta2-review.md)
Fine inscribed flattening still let a native arc cut 6.2e-7 into a 1e-5
clamp unseen: less than one chord's sag, so the layout check, interactive
fill and Check Overlaps all called it clear. Tiny-clamp checks now
circumscribe perimeter arcs and inscribe cutout arcs (lines stay exact), so
the polygons hold every bit of real material. Any real overlap is reported;
the cost is a possible report within the tolerance of an arc, never a false
clear. The analyzer builds the same polygons instead of a second
OverlapMaterial read.
PM 407f5eb7-329f-41e4-947e-64ad52aa4034 (delta review s6-delta-review.md)
Recovery only tried contiguous runs of clamps, so a part with a notch over
a middle clamp had all three clamps removed although the two outer ones
were enough (D8: 'the fewest clamps that let it fit'). One probe per clamp
now finds the clamps every removal must include (a part that cannot fit with
only that clamp left needs it off; removal is monotone), and the search
tries every set by size, mandatory clamps plus the optional ones nearest the
origin first. Full-sheet parts resolve with one probe per clamp.
PM 3df03af2-35a6-49c2-9082-fc6208f5b708
The tolerance-free rectangle test was fed approximated material: the layout
check passed it the 1e-4-grid validation outline (a strip of real material
that wholly contains a 1e-5 clamp could vanish), and interactive fill and
the overlap check passed 0.001 inscribed arc chords (a clamp inside a disk
but outside its chords read as clear). All three now flatten the native
contours at a quarter of the clamp's smaller side, lines kept exact, so the
arc-chord sliver can never hold the clamp. Ordinary clamps are unchanged.
PM 407f5eb7-329f-41e4-947e-64ad52aa4034
Offering several identical sheets (same size, spacing and clamps) multiplied
every feasibility probe, so eight full-sheet parts on four 48x120 options
with the eight-clamp rail spent the 1000-probe budget and recovered nothing
although each part fits with its clamps removed. A probe's answer is now
reused for the same part, engine and physically identical stock.
Delta review (s3b-delta-review.md) policy gate 3; PM 0e655fdb-a99a-40d8-968a-1863d8068c3e
Each blocked requirement was tied to the one stock where it needed the
fewest removals; once that stock's finite pool ran out the part stayed
unplaced even with another offered stock untouched. Recovery now runs in
rounds: a blocked part with demand left takes its best removal among the
stocks it has not tried that still have sheets (that stock's own minimal
run, C11). A variant used in several rounds stays one stock entry.
PM bf4438ac-929a-405f-ba26-fcfaa9b1c55e
ajandClaude Opus 5.5