The kernel's Continuity moves out of bridge.rs into continuity_order.rs and gains G4. One
table holds each order's command word (g0-g4), its code (G0-G4, for tips and messages) and
the message ids of its label and hover text. Construction says which orders the patch,
bridge, network, match and sketch blend curve build today; Continuity::built_by is the one
refusal every command and kernel entry point uses. A design may now save g4 (format note).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LUXeKMoxpM7Vuv8Rb6E3K2
The analysis sliced the display mesh with planes: faceted, approximate, and
broken at triangle seams and trims; the window drew it as a 2D overlay without
depth while the tool was active. Now one kernel routine (kernel::iso) clips a
surface's u = const or v = const line to a face's pcurve loops; the curve through
a point (sketch projection) and the analysis grid (count each way, evenly across
the face's extent) both use it. The loop polygon follows the loop's order (a
pcurve the loop runs against is walked backwards), which the inside test got
wrong before, and closes over vertex-loop poles.
ModelBody caches the lines per face, count and display quality. The session
keeps the picked faces as view state (inspect.isocurve add/clear, hostile counts
refused); the viewport, ui.render and snapshots draw them depth tested in the
accent colour. Curvature combs on faces follow the same exact lines.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A corner where three or more rounded edges meet, every edge there rounded, is
set back (1.5 r unless a setback names it) and left open by SMLib
(blend_smlib::fillet_setback_open: the sheet and each corner's hole, in the
sheet's orders); each hole is filled by fill::matched_patch, the height field
over its plane fitted to the position, tangent plane and normal curvature
across the boundary of every face beside it (exact second derivatives), rounds
and planes alike, each edge's pcurve its exact projection; SMLib sews the
sheet and the patches into one solid, which must close. The history is the
sewing's carried through the sheet's, each patch face made from the rounded
edges its neighbours come from.
corner_g2 is that construction now. The polar patch over three sections it
held (never sewn in) could not be: its boundary is a closed cubic through
stations, which rounds off a corner of the loop, and the setback hole has six.
The 60 x 40 x 20 box corner of r5 rounds (07/015): 10 faces, a valid closed
solid (record and SMLib's check), curvature jump across the patch's seams
6.3e-4 per mm clear of the hole's corners against the ball's 0.2; set back 8
between the hull and prism bounds; set back 11 to 15 inside the box but by
under 7e-4; all eight corners of a box filled; refusals named.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The branch rebased onto main keeps each of its commits; where both sides
changed the same code, these are the pieces neither side's commit had.
- Cargo: one [features] table in the CLI's manifest (two met in the merge).
- geom: a NURBS surface's `closed` set once in a placed copy (both sides
added it).
- kernel: the whole-face extrude tools (#2040) work around truck's boolean
on truck types, which left the workspace with the SMLib kernel; the
module, its tests and the doc crate's tool and tests go. The display
planner's flat fill (tess/fallback.rs) has no caller with the schema 4
mesher, and the tessellator's seam and uv modules are the branch's. A
body's boundary tolerance from a STEP file belongs to the truck body the
schema 4 record replaced: `with_boundary_tol` goes. `unmeshed_faces`
counts the coarse mesh as the branch does. A loft's section re-split
passes the lead's perimeter shares to main's `resample` (a circle breaks
there when the matched corners do not go round in order). Main's tests of
the mesh-based measure (the SMLib kernel measures exactly) and of the
IGES preprocessor name (no such field on the branch's import record) go.
- doc: a feature result carries no boolean diagnostics on the branch, so
the model cache neither writes nor reads them, and a pattern copy's miss
test runs the intersect as it is. A placed body's worker result is a
Result: a body taken from the previous placement is Ok. A thread's
clipboard remap takes main's arguments, the tapped catalogue names a
metric size by ThreadForm::Metric, and main's tests unwrap the placing
calls that return their error.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each (point, distance) sets back the corner at the body's nearest vertex: its
patch starts that far along every edge there. Tested on the 60 x 40 x 20 box
corner with r5 rounds set back 8 (volume between the convex hull and the
rounds' prisms bounds, 15/36/23 topology, the box's bbox) and on refusals.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Ladder 06/009 had no way to ask for one: "solid.thread has no runout (vanishing thread)
length". With SMLib's smlib_screw_sweep_radial (replay e7acea1):
- smlib: the FFI binding and Brep::screw_sweep_radial, the profile moved off the axis by a
piecewise linear law of the turn; crate test: a tooth moved out 1 mm over three turns has
A 2π (c turns + ∫ offset dt).
- kernel: sweep_helix_radial, sharing sweep_helix's checks (helix_profile).
- doc: `ease_in`, `ease_out` (a length, or ISO 3508's standard 2.5 P or short 1.25 P) fade the
groove over that length, its section moving out of the material as it turns (outward for a
shaft, inward for a nut). ease_in is at the end the thread starts from. The runout lies past
the full-form length: beyond a square stop, before a shoulder up to its clearance; at a free
end inside the face. Runouts longer than the thread are refused. No end gets one by default,
so saved threads keep their geometry.
The cut is the integral over the turn of the clipped section's 2π∫ r dA: engine test (2.5 P
and 1.25 P between two collars) within the sweep's wall tolerance times the runouts' wall area
(0.029 against a 0.036 mm³ bound), ladder 06/009 (004's shaft, a 2 P runout before the
shoulder) graded in closed form (measured 1.7e-6 off).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A thread of some length ran from "the end of the face nearest the pick", a coin toss for a
pick in the middle (rig 602's 4 mm thread meant for the free bottom end ran at the head end,
cutting 1.17 mm into the hex), and every thread end at a face end got a one-pitch overrun,
which at a shoulder gouged the material past it (ladder 06/004 cut 0.9% of its volume out of
the collar).
- kernel: material_in_ring, the exact intersection of a body with a ring about a cylinder
face's axis, by an SMLib boolean.
- doc: a face end is free when no material lies past it, over the span the groove's one-pitch
overrun reaches beyond its first pitch, within the radii the groove cuts (a cap, a chamfer
the groove runs out through, a counterbore). At a free end the groove runs out as before; at
a closed one (a shoulder, a head, a blind hole's bottom) it stops with its widest part clear
of it; inside the face it stops square.
- doc: a thread of some length starts at the end nearest `from`, else at the face's only free
end, else at the end the pick lies within a third of the face of; a pick in the middle of a
face whose ends are alike is refused, naming `from`. A length as long as the face runs the
whole face. `up_to` a plane or a point's place along the axis. New fields `from`, `up_to`
default to none, so saved threads keep their meaning where it was not a coin toss.
- engine: solid.thread takes extent (full|length|up_to), from and up_to.
- ladder 06/003 (now with `from`), 004 and 007 are graded by volume and centre of mass in
closed form (a square stop inside material removes exactly length/P turns, the section being
symmetric); rig 602's method says why the thread is at the bottom, tolerance 0.02%.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The kernel blends exactly the edges given; the feature's tangent_chain (default true, saved
only when false: pending format note blend-tangent-chain) expands each pick to its tangent
chain (kernel::tangent_chain) before the kernel call. A variable or spline round takes its
edge alone. Scripts and MCP passing one pick get the chain again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
SMLib (fork 68e3086) refuses a blend whose rails leave the faces they roll on with
SMLIB_ERR_NO_ROOM, which the kernel reports as KernelError::TooLarge. The kernel's own
pre-checks go: blend_fit's fit check and the 256-station band check of the variable fillet.
The shim no longer follows tangent chains; the Fillet/Chamfer dialog already expands a pick
to its chain (its Tangent Chain option) and stores every edge in the feature.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
extrude_tapered(_between) take a DraftCorner: PreserveCurvature (each wall
continues as itself, a line's plane or an arc's cone, until it meets its
neighbour; SMLib's SM_OC_CURVE_EXTENSION), Tangent (the offsets continued
along their end tangents) or Round (an arc about the corner). A corner
the rule cannot close is refused with its point and why
(solidcraft_smlib::Error::Corner). draft_corners reports, per sharp
corner of a profile, what each rule gives: how far each offset curve runs
on or is trimmed, whether the corner cuts inside the exact offset, or why
the rule refuses it. SMLib pinned at the fork's draft-corners head.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
Owner's decision for strips whose rulings turn freely along their edge (a face's
normal along a free-form face, which was refused): the far edge is not the
offset curve but a B-spline the kernel defines, and the strip is exactly the
ruled surface between the edge and it.
The far curve has the edge's own degree, knots and weights (so the ruled NURBS
needs no knot merging); it runs from the strip's start corner to its end corner
(where the rulings at the edge's ends reach); its interior control points are
given (handles) or the least-squares solution of weighted terms, the
construction's parameters: closeness to the rulings' ends at the edge's
Greville abscissae (W_CLOSE), fairness by arc length (W_FAIR), each end leaving
along the chord to the other end (W_TANGENT, one small function), and at a joint
where the chain runs on smoothly with shared sides the first and second
derivatives by arc length matched to the neighbour's (W_JOIN), in one solve
along the chain, an exact neighbour's far edge matched as it is. Exact strips
are unchanged.
ruled_far_handles gives each free strip's default handles; ruled_surface_with
takes moved ones. The Ruled Surface feature saves them as `far` (per chain
edge, null for the default; pending format note ruled-far-handles), and
surface.ruled takes them as `far`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
Every lib warning under --features smlib outside the files agents are editing
(assemble.rs, sweep_path.rs) pointed at code nothing calls any more:
- body: shell_nesting;
- boolean_record: of_pair and its Traced alias;
- planar_regularize: the generated-wall regularization (and its branches in the
certified one, which was its only other path); planar_sources:
with_shell_planar_sources;
- plane_fit: plane_through, EdgeSampler, planar_face, plane_frame (the patch is
fill's); only loop_area is left;
- polyhedron: solve3; prism: wall_along; offset: Surf::normal;
- record: regions, region; shell_groups: groups, parents and their limits;
- splitface: side_of_point, Analytic, analytic_face, face_sides, canonical;
- spline_conv.rs, all of it (surfaces builds NURBS records); an unused import in
tess/uv.rs.
The tests of deleted items go with them: prism's wall_along test, the shell
grouping test in body_lump_tests, and the generated-regularization assertion in
regularization_protection_tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
The doc puts exact tools into SMLib's boolean as drawn, so nothing asks whether a touching tool
can be joined, batches extrusions, or probes for room to grow a profile.
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
A rig case whose SMLib call fails left nothing to replay it from: nothing called
solidcraft_smlib::set_dump_dir. `solidcraft-cli kernel-cases --dumps=DIR` now points SMLib's
dump folder at DIR/<area>-<case> before each case runs, so a failing call leaves its
arguments and B-reps there (smlib_brep_read reads them back into a scratch C program), and
turns it off after the run. Off without the flag. The kernel exposes it as
smlib_dump_dir, beside smlib_built (the CLI reaches SMLib only through the kernel).
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
face_params and outline are gone (no callers on the record). The fork head adds the 2D planar
boolean (smlib_planar_faces, smlib_boolean2d), the grazing-curve crossing rule (solidcore#480) and
untrim's parameter-box status.
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
A solid body's volume, area and centre are SMLib's mass properties at its documented 1e-4 and its
box SMLib's exact one; a body SMLib cannot measure is an error. The three-mesh measure and the
mesh cross-check beside SMLib's numbers are gone, and with them the kernel switch
(SOLIDCRAFT_KERNEL, Backend): SMLib is the kernel. The smlib feature gates in the crate root, the
validity check and the offset backend go, since SMLib's crate is always linked (answering not
built without the feature).
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
A boolean is SMLib's boolean, its history SMLib's tags: the jittered retries, tool nudges, prism
and coaxial routes, the retry result guard and their diagnostics are gone. A rigid placement moves
the record; any other matrix goes through SMLib's transform (new Brep::transformed over
smlib_brep_transform). Bodies apart join as their records side by side (brep part::compound),
with each piece's faces, sources and history; apart means their exact boxes do not meet. A split
is the intersection with a half-space box, with no faces pushed off the plane.
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
The fork head carries untrim and extend shims with their history tags, the local solvers that
take no value from a failed evaluation (solidcore#481), the B-spline curve radius check
(#475), the torus minor-interval check (#482), the blend rail along its face's rim (#485) and the
intersection path's statuses.
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
assemble (records rebuilt after removing and adding faces) joins; the boolean guards that re-ran
a failed boolean with a changed tool (bore_guard, candidate_reach, sphere_lens, tangent_cylinders,
touching_walls) are gone; the lens-cut result stays as a boolean test. SMLIB_REV at the fork's
integration head with smlib_merge_faces.
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
SMLib's blend builds rolling-ball rounds, corners, loops and vertex spheres in one call; the
modules that built them on the old kernel were deleted by the blends port.
(not built: base in progress)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
A drawn 3D spline reached the kernel only as its sampled polyline (Wire.pts), so SMLib refused
to sweep along it ("not supported yet: SMLib sweep along a smooth path of points"). The doc
holds the spline's fit points; the curve is the centripetal Catmull-Rom spline that
fit_polyline samples. Each span of it is a cubic (Barry-Goldman), and its derivative at a fit
point depends only on that point, its neighbours and the two knot intervals, so neighbouring
spans share it: the spline is exactly one cubic Bézier per span, C1 across the spans.
- geom: catmull_rom holds the curve: samples() (fit_polyline's arithmetic, moved verbatim;
sketch::fit_polyline now calls it) and beziers(), the exact spans; ArcLength on CurveN
(Gauss-Legendre per knot span, Newton for the parameter at a length).
- kernel: PathSeg::Cubic, a cubic Bézier in space (the 3D counterpart of PathSeg::Curve, which
is planar), with its arms wherever PathSeg is matched (start/end, tangents, length, point at
a length, turn, SMLib curve as a degree-3 NURBS, halving). PointPath / PathPiece carry a path
through 3D sketch curves: polylines as points, drawn splines as their fit points (reversed
when the path runs them backwards); PointPath::segs gives SMLib the spans and the lines;
sweep_point_path_with sweeps it (SMLib through sweep_with, so corners, twist, scale, rails and
orientation take the existing routes). A smooth polyline piece (a projected or linked curve,
which the kernel only has as points) is still refused on SMLib.
- doc: path_points builds the PointPath (a wire with three or more fit points is a spline
piece); solid sweep and pipe call sweep_point_path_with; rails, surface sweeps and curve
networks take its points as before.
Checked in scratch (not committed): the spans pass through every fit point and every
fit_polyline sample at its parameter (1e-12), C1 at the joints and round a closed spline;
samples() is bit-identical to the old fit_polyline over 1180 random open and closed inputs;
SMLib sweeps of a radius-2 circle along an open spline and round a closed one match
pi r^2 L (L by independent quadrature of the curve's definition) to 1e-7 and 3e-8.
sketch/wire.rs: the existing closed-spline test reads SPAN_SAMPLES from its new home.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
Body::mesh / mesh_plan / mesh_of and MeshTolerance (chord, display, display_of, export_of)
replace tessellate, tessellate_with, display_mesh*, display_plan*, reference_display_mesh_angular;
every caller in kernel, doc, engine, io, ui-egui, gpu and the CLI renamed mechanically; the
truck-mesher fallback, mesh_shells*, refined* removed; Body::new orients from the planner;
truck-meshalgo dropped from Cargo.toml with the tests of its hooks. Still to fix: trait imports
in tess/*, splitface/lumps/preview/step_in/boolean_record meshalgo uses, measure.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017itGh5fpMEousnZ91cYcVt
With SOLIDCRAFT_KERNEL=smlib, kernel::measure and measure_mass take volume, area and centroid
from SMLib's Brep::mass_properties (integrated on the exact surfaces, accuracy 1e-8). The mesh
measure is still made, as an independent check. Where SMLib disagrees with it beyond the mesh's
own accuracy (1e-3; tangency-labs/solidcore#384), or can't take the body, the mesh measure
stands and the measurement says why. BodyMeasure and MassMeasure carry `measured_by` (smlib |
mesh) and `notes`, and inspect.measure shows both. The box and topology counts stay the mesh's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every blend entry point asks `backend()` and, on SMLib, hands its arguments to blend_smlib.rs:
fillet, fillet_each and fillet_radii (a radius per edge), fillet_g2 (SMLib's G2 blend-curve
section between the round's contacts), fillet_chord (constant-width section), chamfer, and
fillet_variable_spline (the radius law as an SMLib B-spline law; fillet_variable and
fillet_variable_points reach it, and fillet_max, fillet_max_each and fillet_tapered run on these).
Options SMLib has no construction for are refused as "not supported yet", never built by truck:
G2 corner patches, and chamfers with two different setbacks (distance and distance, or distance
and angle; equal setbacks are the equal-distance chamfer).
Each pick takes the body's nearest edge in its own edge order, as truck's blends do, and SMLib is
given that edge's midpoint, so a pick at a vertex never leaves SMLib to choose between edges. The
body goes in tagged as operand 0, SMLib's check must pass, and the decoded result carries SMLib's
history in the body's orders, the blend faces given FaceOrigin::Edge of the edge each replaced
(the picked edges and the rest of their tangent chains). The truck path is unchanged.
Tests (with the feature): a round's volume and its face's origin, a size per edge for rounds and
chamfers, a linear law from either end, refused sizes and picks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
With SOLIDCRAFT_KERNEL=smlib, extrude (and extrude_whole_sides, so boxes and cylinders too),
revolve and revolve_whole, extrude_tapered(_between), sweep and sweep_with, loft,
loft_with_rails, loft_regions and sweep_helix (coils, and the threads' grooves) build their
bodies with the fork's sweeps (smlib_sweeps.rs): the profile's segments go to SMLib as curves
(lines, arcs, Béziers and conics as NURBS), the result comes back through its schema 4 record,
and one SMLib's own check refuses is an error. What SMLib has no sweep for answers "not
supported yet", never truck's sweep: twists, end scales, guide rails and orientation
directions; paths with corners and paths of points; lofts with holes or to a point; draft
extrusions of profiles with holes or with a straight foot.
History is in the builders' vocabulary, found from the result's geometry as provenance.rs does
for operations without exact history: a wall comes from the segments its start edge runs along
(and, for extrusions and revolutions, whose middles it holds half-way), the face in the start
plane where the profile bounds it is the start cap, the one at the end the end cap. Straight
extrusions also carry their faces' construction sources (ExtrudeCap, ExtrudeSide), which the
box and cylinder certificates need. The loft's section matching is truck's own, now
loft_numbered_loops (split out of loft_numbered_wires, unchanged), so both kernels pair the
same vertices.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
With SOLIDCRAFT_KERNEL=smlib (the one kernel switch), shell, offset_faces and thicken send
the body to SMLib tagged as operand 0 (offset_backend.rs) and bring the result back with
SMLib's own history, as booleans do. A result SMLib's check refuses is an error.
SMLib's tags travel through the copies its offset makes, so a shell's inner wall and a
thickened sheet's moved copy read as pieces (Split) of the face they came from. A result face
lying more than half the operation's distance from its tagged face is recorded as that face's
offset (FaceOrigin::Offset) and its edges as where its faces meet; a thickened sheet's walls,
new to SMLib, are recorded as made in place of the sheet edge their edges are tagged with, as
truck's thicken records them. Offset faces pick the faces truck's path picks (a round wall's
pieces together) and pass SMLib one point inside each; the moved faces are Modified.
The histories are checked against the geometry as truck's are (modify_history_tests), with
the analytic volumes, the thicken direction, and the refusal of walls past the L-shaped body's
inradius (8.79) that its box (half extent 10) let through before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
All operations move to SMLib at once, so the selector is no longer the booleans' own:
`SOLIDCRAFT_KERNEL=smlib|truck` (read once, unset is truck, the same refusals for an unknown
value or SMLib in a build without it) is answered by `solidcraft_kernel::backend()`
(crates/kernel/src/backend.rs), which every operation's dispatch calls. The boolean's SMLib path
moves to smlib_boolean.rs. The regression rig's kernel check (`boolean_kernel`) now answers from
the switch, and the rig and CLI set SOLIDCRAFT_KERNEL.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
With the `smlib` feature, SOLIDCRAFT_BOOLEANS=smlib (read once per process, default truck)
sends every kernel::boolean through SMLib: each operand's schema 4 record imported and tagged
as operand 0 or 1, SmBrep's boolean, SMLib's check at every level, the result exported and
decoded (with its geometry check). A result either check refuses is an error. The setting
naming SMLib in a build without it, or anything unknown, fails every boolean instead of
quietly running truck, so a comparison never compares truck with itself.
The history is SMLib's tags renumbered into Body::history's index spaces. The export now
reports where each SMLib face and edge lands in canonical order (its arrays follow
SmBrep::GetFaces/GetEdges, the order the tags use). The result's order is checked against the
decoded solid by face and edge identity, the operands' by re-exporting the imported body and
comparing its numbering with the record's; where either differs the result has no history
rather than a wrong one. The result gets the operands' regularization protection and planar
roles as truck's does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
- `"shells": N` in expected.json grades the total number of shells: a body with a sealed void
holds two, and some joins keep several lumps in one body. Those cases (#3175, #3455, #3526,
#3730) are graded on validity again instead of being marked not valid.
- When SMLib's record conversion says "not supported yet" (revolved surfaces, traced intersection
curves), its check gives no verdict. That's shown but not graded, since it isn't a fault of the
body. solidcraft_kernel::SmlibCheck carries the three outcomes.
- `$OUT` in a recipe is a scratch directory outside the repository, for files a case writes. The
two STEP-export crash cases write there instead of /tmp.
- cargo xtask kernel-cases --results FILE keeps every result as JSON.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- solidcraft-cli and solidcraft-engine forward an `smlib` feature to solidcraft-kernel/smlib, so
`cargo build -p solidcraft-cli --features smlib` gives the rig SMLib's check.
- solidcraft_kernel::boolean_kernel() says which kernel this process runs booleans on.
`default` until the SMLib switch (SOLIDCRAFT_BOOLEANS=smlib) lands and answers there. An smlib
run on a build without the switch reports "not supported yet" instead of grading the default
kernel under SMLib's name.
- An area may keep its skipped list in parts (skipped-<part>.json).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every geometry case the old kernel got wrong becomes tests/kernel-cases/<area>/<id>/: an
engine script or oracle recipe, and expected.json with the right answer, from analysis or a
hand calculation, or else the invariants. Never a kernel's output.
- solidcraft-cli kernel-cases <dir>...: runs a case and grades it. The checks are the outcome
(succeeds, or refuses cleanly), the body count, and one closed shell per body that passes the
record's structure and geometry checks and, with SMLib built, SMLib's check(All). Counts,
volume, area, centroid and box are graded where the case gives them. Each run is pass, fail,
not supported yet or crash.
- cargo xtask kernel-cases [--kernel default|smlib|both]: one process per case, so a native crash
or a hang (--timeout) is that case's result. It writes docs/kernel-cases.md: coverage per area
and source, results per kernel, and the sources looked at and skipped (<area>/skipped.json).
- solidcraft_kernel::validity and smlib_built, which the rig's checks use.
tests/kernel-cases/README.md gives the format.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SMLib B-reps leave the C API as its flat arrays, read through the crate's
one flat wrapper (crate::flat::Flat, copied out as import::Arrays), and
become schema 4 records: shell pairs merged, sheets collapsed, radial cycles
normalised, analytic surfaces as their kinds (narrowing cones kept), loops
reversed where SMLib's surface is inside out, every pcurve from
brep::pcurve, canonical order, then validate() and check_geometry(). Edges
arrive as their own kinds; NURBS edges that are lines or circular arcs (the
primitives' circles) are recognised as a fallback and counted. With the
kernel's `smlib` feature, body_from_smlib() decodes them into a Body.
Adds the cylinder, sphere, cone and torus primitives. Each primitive passes
SMLib's own check at every level, its volume from the exact mass properties
matches the closed form, and it goes round SMLib -> schema 4 -> SMLib with
its counts, volume and a clean check. Truck now meshes a whole sphere
bounded only by its seam (SMLib's sphere) through the display plan.
SMLIB_REV moves to the fork's capi-primitives (the primitives on capi-flat
40bb855).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
SMLIB_REV moves to the fork's capi-static 40bb855 (flat arrays, C ABI 2).
flat.rs wraps them (Flat: arrays by id, Brep::export, Flat::import,
Brep::is_manifold_solid; FFI in ffi/flat.rs). import.rs lays a schema 4 record
out as those arrays: a closed shell becomes one SMLib shell per region it
separates, a sheet's shell holds both sides of each face, the infinite region
is marked void as SMLib has it, pcurves become UV trims on B-spline faces only
(reversed for edgeuses against their edge), planes with non-unit axes go as
their exact bilinear patch. Brep::from_record builds the SMLib body.
kernel: brep_record(body) gives a body's schema 4 record, and smlib_body
(feature smlib) its SMLib body, trusted (no healer: healing moved a
cylinder's volume by 9e-7). Box, cylinder and tilted cylinder keep their
analytic volume in SMLib (1e-9, 1e-6) and are manifold solids; truck's
spheres and tori are revolutions, refused with "not supported yet" for now.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A new crate (L1, only the kernel uses it) builds tangency-labs/smlib's SMLib,
SM_API and C API with the cc crate when its `native` feature is on, from the
commit in crates/smlib/SMLIB_REV (scripts/fetch-smlib.sh fetches just those
sources). The kernel's `smlib` feature turns it on. Without it nothing C++
is built and every call returns NotBuilt, so workspace builds need no C++
toolchain.
unsafe is confined to src/ffi.rs. The wrapper owns handles, maps statuses to
errors, checks the C API version, and sends SMLib's per-call log to the log
crate. Crash reports note the native kernel call in flight.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Extrude: a join from a whole face rounded by fillets sweeps the face itself (#2040)
A face's sketch fits an edge that is no line or arc (the quarter-ellipses
where fillets meet a slanted face) as splines a hair off the body's edges,
so a join from the face touched the body along a face it did not match and
failed with "boolean Union: no result" after minutes. When the profile is
a whole planar face of the one target, the join now sweeps the body's own
face, which shares its edges and is sewn on along it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Extrude: a cut into a whole face keeps the walls that cut and moves the others off (#2040)
A cut from a whole planar face sweeps walls from each of its edges. Where
the face beside an edge turns away from the sweep (n_N . -n_P >= 0) the wall
only touches the body along that edge, or lies in the face beside it, and
the boolean failed on it with "no result". The cut now keeps the walls that
go into the body, exactly (the body's own edges), runs the outline round
the others well outside the body, and starts a little out from the face.
A cut whose walls all only touch takes a slab beyond its depth plane.
Not covered: a cutting wall tangent to the face beside a touching one (a
round's wall meeting an end face, as in the issue's part); the boolean
still fails there.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Extrude: leave a whole face outlined by lines and arcs to the profile's own tool
Only a face whose outline the projection had to fit needs the body's own face as the tool. A square drawn over a box's top already meets the body exactly, and the swept face lost the names a sketch on the new top attaches to (stacked_face_tests, #489).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Run the whole cut tool from a face under the kernel guard (#2040)
Review: `builder::vertex`, `builder::line`, `Face::try_new` and
`builder::translated` ran outside `crate::guard`; only the sweep was
guarded. `cut_tool` now runs all of it under one guard.
Adds an end-to-end test of a cut into a face the body stands out past
below it (the block's top strip between its rounds): it takes only
what lies over the face.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: the slab cut of whole_face_tool, on a spline prism (#2040)
A prism on a closed spline standing on a wider plate, cut 3 into its whole
top face: the slab takes the face's area times the depth and leaves the
plate alone. A test-only counter asserts the slab was the tool, so the test
cannot pass through the profile's own tool. Shortens the name of the
rounded block's top-face cut test.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Kernel: the face's outline as a Vec now that polyline returns a closed chain (#2040)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Tessellation: trimmed faces of display meshes stay on their surfaces
Display meshes triangulate a trimmed face in parameters scaled by the
surface's speed along each, so a strip stretched along one parameter is
not fanned across, and every boundary segment stays a constraint: one that
crosses another is split where they meet. Boolean meshes keep truck's way.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* Tessellation: one boundary rule for every mesh; drop the #1648 retry
A boundary segment that crosses one already added is split where they
meet in Boolean meshes too, which keeps a pinched face (#1648) whole
without the snap-and-retry. Faces whose segments all go in mesh as
before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* Robustness goldens: seeds 1, 23, 55, 64 and 86 re-recorded on current main for #1653
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
* Thread preview test: wait for the stale fit error to clear before the next preview
After switching back from M6x1.3 to M8x1.3, wait_feature_preview stopped at the M6 error still on
screen when the new preview started a few frames later (seen on Linux CI with #1653's meshing).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
* Robustness goldens: the slow seeds 14, 20 and 40 re-recorded for #1653 from the merge queue
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
---------
Co-authored-by: jyviko <2501648+jyviko@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Moving an occurrence places every body under it again, and each placed copy
searched every revolved face for where it sits to turn it away from its seam,
although its source had been turned already. On the Prusa Core One assembly
that search was most of the 12.7 s a move of the top occurrence took. A body
now records that its revolved faces are turned, and world placement
(`place_matrix`) skips the search for such a body. Modelling moves (pattern
copies, hole tools) keep `transform_matrix`, which still turns them in the
copy's own frame: where a face wraps all the way round the turn it finds
depends on those coordinates, and later booleans depend on it.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The retries in boolean_whole_inner mesh every candidate to compare its
volume. A nudged union at the coarse tolerance could come out with a face
on a cone whose trim loop ran metres along the cone's continuation (edges
up to 12.7 m outside a 155 mm part); meshing it took seconds to minutes per
face and once never finished, though its volume (+-1e10 mm3) failed the
check anyway.
check() first samples the candidate's edges (16 per edge) and discards a
candidate with a point further outside the operands' mesh box than 5% of
their size (or a non-finite point): no edge of a union, cut or
intersection can lie there. The reason goes into the new
BooleanDiagnostics::rejected (omitted from JSON when empty). Results of
sound candidates are unchanged.
On the reporting design the PitchTail join drops from 18.7 s to 4.3 s with
the same result; the regression tests synthesize a cone face whose loop
runs off its surface.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* kernel: loft guide rails (#201)
A loft follows one or two guide rails between its sections: stations are added between each two
sections, each the plain loft there moved so that the point a rail guides lands where the rail
crosses the station's plane — turned about the loft's direction and scaled (or stretched across
to the rail) with one rail, carried, turned and sized between two. The sections and caps stay
exact; the smooth loft runs through sections and stations. Solid and sheet lofts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* doc, engine: solid.loft and surface.loft take guide rails (#201)
`rails` (one or two {sketch, curves}) and `rail_fit` (scale | stretch), stored on Loft and
SurfaceLoft (one pending format note), parsed by the sweep's rail parser (now shared) and built
through the kernel's railed loft. Rails on a loft to a point, or on a surface loft with edge
sections or end continuity, are refused as not supported yet. The rails' sketches are walked as
the feature's references and their components placed like a sweep's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* ui: Rail, Second Rail and Rail Fit in the Loft and Surface Loft dialogs (#201)
The rail inputs and fit row the sweep dialogs have, shared through dialogs_sweep; Edit Feature
brings the rails back.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* docs(roadmap): loft guide rails (#201)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* loft rails: review clean-ups (#201)
- Sweep and loft rails share one turn-and-size map (`sweep_path::rail_frame`).
- A station's middle and the loft's direction come from one curve through the sections'
middles (exact derivative) instead of two more plain-loft evaluations per station; at most
33 wires, so railed faces stay light.
- Nearest points on a section use `param_on` (kept only where the search comes nearer); the
size uses `Aabb3`; one B-spline builder; anchors keep their nudge.
- The railed and plain lofts share one path in `loft_with_rails` and `loft_sheet_with_rails`.
- The engine leaves the not-yet-supported refusals to the design's evaluation; both lofts start
their rails from the first section's first point.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* loft rails: refuse rails on sections with holes, with a test (#201)
Main's lofts now take sections with holes (#2447); rails shape only the
outlines, so a railed loft of rings is refused until the rails carry the
holes too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
---------
Co-authored-by: jyviko <2501648+jyviko@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Model cache written by default: exact bodies, stored once, scaled caps (#3053)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* bench-open: say why bodies were not kept
* Exact planes and revolutions; budgets for large bodies (#3053)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Native B-rep: rebuild shells as the kernel made them (closed and seam edges) (#3053)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs: the model cache and the exact native B-rep (#3053)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* bench-open: time a save without the cache too
* Model cache entries are Deflate-compressed (#3053)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Model cache: one that does not fit the design file is left out, never a failed save (#3053)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* inspect.clearance: pairs whose bodies did not change are not measured again
Each clearance check measured every close pair again, though between two
checks usually only a few bodies change or move. Results are now kept per pair,
keyed by both bodies' geometry and placement and the check's settings; a pair
whose two bodies are unchanged is taken from that cache. With 40 overlapping
parts, a check after moving one of them took 0.8 s instead of 23 s (741 of 780
pairs reused). Pairs that could not be worked out are tried again each time.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* inspect.clearance: boxes and placed keys exactly as before the reuse
The pair boxes are taken over the mesh's triangles with finite corners, as a
distance shape's are, so pruning is unchanged. A placed body's identity now
includes its model body's own frame, so two placements are never mistaken for
one another.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* inspect.clearance: a large body is meshed once for all its pairs
Each pair that touches cuts the larger body down to the pair's region with a
kernel boolean, and that boolean meshed the whole body twice: once for its
checks and once inside the intersection. A sheet with 600 faces paid about
2.5 s per pair for meshes that were the same every time.
A body can now keep its meshes for reuse (Body::keeping_meshes); its booleans
take the meshes from there, and the intersection takes its shell meshes
from there too (truck-shapeops: and_with_history_meshed). Clearance uses such
copies for one check. Results are unchanged to the bit. On a 3 x 2 m sheet
with bends and 300 holes, the clip took 0.23 s per pair instead of 2.6 s, and
24 rivets took 5.6 s instead of 16.5 s.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* inspect.clearance: pins on one cylinder with their hole are answered or reported in seconds
A pin seated line to line in its hole, on its axis, and with a blend running into its wall
(a fillet under a head) sent the boolean through minutes of retries that ended unchecked; a
pin a hair off its hole's axis was passed as clear. Clearance now asks first whether the pair
holds such a pin (from the faces' surfaces alone). On its axis the boolean gets a few tries,
which answer plain pins at once; a hair off, or when those tries fail, the shared volume is
worked out apart by coaxial_overlap, which now also takes the zero-offset, equal-radius case,
moves the hole's wall when a blend keeps the pin's from moving, and checks its band more
tightly. What neither answers is reported as unchecked, naming the pin.
The kernel gains with_attempt_budget: the booleans run inside it get a number of tries of
the kernel's own intersection between them, the same on every machine.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* inspect.clearance: the pin tests sit beside the other measured pairs, not at the file's end
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* inspect.clearance: the bodies left whole meet their booleans with their meshes kept
A body that meets several others at one size (a bracket with balls pressed into it) was
meshed again for each pair's intersect: only the clipping used the copies that keep their
meshes. The intersect, the pin work and the union check now take those copies too.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Near-coaxial overlap: ten tries per intersection, enough for a flush head pulled back
#1984's headed pin made before its plate needs more than eight.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Aac7mP9rthK3WsPpPRvCsF
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
jyviko