Every other tessellation-keyword issue and PR, with one line each:
rendering and GPU drawing, UI, performance, display-plan plumbing, and
failures that belong to booleans, fillets, threads, import, shell,
sheet metal or crashes.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
311 ramp cut across a round blank, 784/811 filleted and drilled bodies
for STL export, 818 blind hole in a filleted ring, 819/921 pocket that
splits a wall seam, 1263 concave arc with a knife-edge tip, 1615/1653
crossed lofted tubes, 1648/1850 sphere touching a box edge, 1666/1670
two twisted lofts, 1753 bowl rim cut by a cylinder, 1813/2712 lens face
thinner than the tolerance, 2069 notched ring moved far away, 2528
slot joined with an overlapping cylinder.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
140 join with a pad hanging past the edge, 1219 bore fillet then grooves,
1985/2243 STEP export of two enclosures, 3540 cut with a coincident cylinder
face, 3685/1559 delete a fillet face, 4092 loft + ribs + pad union (two
recipes), 2992 disk and 36 twisted loft blades. Everything else is skipped
with a reason (private designs, other areas, UI, perf only).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
4009 closed shell, 4010 shell open at a hole wall, 2531 point pick after an edit,
3468/3898 offset next to a fillet, 12 split face piece, 566 swept bar, 745/3893
freeform boss (two variants), 3103 tapered cylinder, 3669/2990 loft, 3283/3650
and 2939 curved offsets, 3506 and 1972 split then shell, 3576, 391, 1494, 1525.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Cases: #3702 (a board in the cavity of a case with a filleted rim, cut leaves the case
unchanged) and #258 (STEP cone to a point, with and without an apex vertex, and the 0.5 mm
pin). Everything else in the bundle is skipped with a reason in skipped-other.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>
A test over every design the repository keeps (the action matrix's
corpus: old design files, examples, regressions, demo and typical
designs, the oracle recipes; plus the sample designs, the engine's
design fixtures and the older designs of the hole matrix, taper and
projection tests: 154 designs). Each is rebuilt and recorded in
tests/naming/<group>/<design>.txt: per body the names of its faces and
edges with whether each follows exact history, and every stored name
(feature face picks, fillet/chamfer/bend edges, sheet metal edges, the
face a sketch is attached to) with the face or edge it finds now. A
difference fails with the design and the changed lines; record an
intended change with NAMING_CORPUS_BLESS=1, as FEATURE_WALK_BLESS does.
Face and edge indices are left out on purpose. They are the kernel's
order: truck already gives a fillet's or a boolean's faces in another
order from one process to the next (oracle 07 and 37 changed on every
run), and SMLib will order them differently again, while what must not
change is the names and what references find. Names are listed sorted.
One design's names themselves change between runs on truck
(closed_sweep_torus_cut: level wall halves numbered #1/#2 in kernel
face order); it is listed in UNSTABLE with the reason and its
difference printed instead of failed on.
Records made on this branch (truck), macOS. The whole corpus builds in
about 20 s on 14 cores (designs run side by side).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QWhMwdqJYQKQq8RCLSL1LT
cone_point and cone_point_vertex join the sweep now that #3871 reads a cone
bounded only by its base circle up to its point.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* STEP import: regression sweep over the step-conformance parts of known shape
Fifteen generated parts from step-conformance (CC0): plain, NURBS, seamed,
placed, inch and far-away cylinders, domes, a grooved shaft, a plate with 100
holes, a 3.6 m disc, a 0.5 mm pin, a B-spline half-torus and a washer. Each
reads back with its exact volume, its box and its face count. The half-torus
on a TOROIDAL_SURFACE and the cones to a point are left out until they read
right.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* STEP conformance sweep: the pin's volume as FRAC_PI_8 (clippy)
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>
truck placed a face that no intersection divides (every face of a shell
that meets nothing of the other) by counting the faces a ray crosses,
taking each face's surface as stored. A revolved body's reversed faces
counted the wrong way, and an inward shell (a void, a cut's inverted
tool) was read as a solid: a cut beside a sealed void came out empty,
refused or, with a tool whose bounding box covers the target, taken with
the body gone. Intersections of such bodies came out empty as well.
In an AND the crossings now count as each face is oriented, and an
inward shell holds what lies outside it. An OR keeps truck's count.
The corrected cut let a folded offset through: an arc edge next to a
face it meets tangentially had its ends moved past its middle, and the
arc went the long way round (robustness seed 73's shell). The offset
refuses an arc whose middle crossed its chord, as it does a straight
edge that turned round.
Goldens: seeds 0, 45 and 77 had a cut refused as empty that now takes
what it should; their volumes are checked by hand in the PR.
A face pinched to a point, where its boundary touches itself or another of its loops within
TOLERANCE (a sphere touching a box's edge leaves one on the box), had a constraint refused by the
CDT, and the fallback cut across the face: it came out with no triangles or without those at the
pinch. When a constraint is refused, the vendored truck-meshalgo now puts every boundary point
within TOLERANCE of a segment into it (at most 8192) and triangulates the face again. Faces whose
segments all go in mesh as before.
The regression tests live in crates/kernel/src/thin_face_tests.rs, and the #1648 property test in
boolean_props.rs is no longer ignored. Robustness seeds 20 and 55 are recorded with their Linux
values.
Co-authored-by: MateuszSychXpl <151755914+MateuszSychXpl@users.noreply.github.com>
* Fix large modeled threads and thin crest reconstruction
Helix stations follow the pitch (radial chord error under 1% of it, 48 to
720 stations), and the faceted rebuild retries at the BSP's precision when
the usual weld tolerance collapses a thin crest facet.
Rebased onto main as one commit (the branch's own commits touched UI files
that later moved to #2263 and #2264). Adapted to main's UnitVec3 thread
axis: `*k` in thread_groove and `*cylinder.axis` in the new flange test.
* Fix icon review revisions for same-timestamp edits
On Windows, Git's default core.autocrlf checks the icon masters and the
palette out with CRLF, and `cargo xtask icons` failed four ways:
- the accepted-icon hashes (lint_icons.py and the review page's
review_state.master_hash) were taken over the files' bytes, so every
accepted icon read as changed; both now use master_hash, which hashes
the text with CRLF read as LF;
- icon_palette.revision hashed the bytes while save_colours hashed the
text it read, so every palette save was refused as stale; revision now
reads the text too;
- the tools and their tests read and wrote text in the locale's encoding
(cp1252 can't write U+2194); they now say utf-8;
- the icon inventory listed UI files with backslashes; it writes /.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
#2247 changed 16 robustness seeds (bodies N -> 1, and some seeds then build
different steps; seed 54's new shell fails on #2247's parent too). Recorded
on Linux.
Designs saved before #2247 that combined a split-off piece by name (tools
Posts, Posts (1)) failed on the missing piece and left both posts off the
case. A combine now skips a missing `X (n)` when `X` is a tool too or the
join's target, with a warning. Format note for the meaning change.
#2064 (multi-profile extrusion) merges coplanar faces the boolean used to
leave split. Seeds 67 and 94 change face/edge counts at the same volume;
seed 13 has four fewer edges before its random fillet, which then picks
different edges by index. Every step still passes the validity and
volume-direction checks. Pull request CI runs every 8th seed only, so
main's full run was the first to see these.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* feat(geom): surfaces through curve networks, and spline spaces to build them
Adds spline_space (clamped B-spline spaces, Hermite interpolation, exact
embedding by Bézier degree elevation and knot insertion, a monotone cubic
for reparameterising) and network (loft, Coons and Gordon surfaces through
curves in one or two directions, with G1/G2 cross-boundary conditions
against a neighbouring face). Part of #763.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(kernel): boundary surfaces from sketch curves, 3D curves and edges
kernel::boundary_network turns sketch chains, 3D sketch polylines and edge chains
into network curves, gives G1 frames of the faces along open edges
(EdgeFaces), builds the B-spline face (sides split at chain vertices so a
join shares the body's edges) and joins it into a surface body by stitching.
G2 is refused for now (not supported yet). Part of #763.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(kernel): second derivatives of B-spline surfaces in the vendored truck-geometry
BSplineSurface::uuder and vvder multiplied the second differences by p
instead of p·(p − 1), as BSplineCurve::der2 does, so every surface of
degree p ≥ 3 (NURBS too, through its homogeneous surface) reported
1/(p − 1) of its second derivative: a cubic surface's curvature read as
half. This is why a G2 boundary surface on a cylinder measured half the
curvature it carries (#763). New tests hold both to finite differences
for degrees 2 to 5.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* feat(kernel): G2 boundary surfaces against a body's face
The network always carried the face's curvature across a G2 edge; the
refusal came from measuring it with truck's halved surface second
derivatives (fixed in the previous commit). EdgeFaces now reads the
face's second fundamental form from the exact second derivatives
instead of differencing the normal. A new test holds the curvature
across a cylinder's edge to the cylinder's within 0.1%, also with a
sheared far curve; G3 is refused (a network holds up to G2).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* test(kernel): a boundary patch with straight corners measures its own area
A stadium's arcs meet its lines at 180°, so the Coons patch is singular
at its corners. From such a corner the reference mesher's boundary walk
ran onto the B-spline's extrapolation past its domain (a planar spline
passes through every point of its plane there), and the face measured
289 mm² for its 158.5 (#763). The vendored mesher's bounded projection
(nearest_projection, #2043) keeps the walk on the domain; this holds it
there, by the faces, the measuring mesh and the display mesh. The
display path (#1653) was never affected.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* feat(geom,kernel): three-sided boundary surfaces (one side collapsed to a point)
A network curve may be a PointCurve: the apex where the two sides beside
it meet. The curves through it cross it at the point itself, its row of
the net collapses there, and the kernel's face leaves that side out of
its loop (three edges, one vertex at the apex). One point per patch, as
a side; refused elsewhere, when it moves, or with a continuity
condition. Tests: a triangle and a quarter disc filled flat with their
analytic areas (40 mm², 25π), by geom and by the kernel's faces and
meshes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* feat(doc): boundary surfaces through curve networks and loops of three or four
FeatureKind::SurfaceNetwork saves a network as `u` and `v` curves, each
a sketch chain (planar or 3D), a chain of a body's edges with an
optional G1/G2 continuity, or a point (the apex of a three-sided
patch), and an `operation` that can join the surface into the body of
its edges. A SurfaceBoundary of three or four sketch curves, refused
until now, builds as the flat Coons patch of its loop (three collapse
the side opposite the first curve). The format change is one pending
file; feature-walk goldens record the new variant's sketches, bodies
and points.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* feat(engine): surface.boundary through loops of three or four and curve networks
The command keeps its loop form (two to four curves of one sketch; three
and four now build the flat patch of the loop, crown refused there) and
gains a network form: `u` and `v`, each curve a sketch curve id, a
{sketch?, curves} chain (planar or 3D), a {body, edges, continuity?}
edge chain (g1/g2 with the face beside it) or a {point} for a
three-sided patch, with `operation` new or join. Edits may move between
the two forms (timeline.redefine accepts the command now that it has a
dialog).
Tests: analytic areas of loop patches and of a hyperbolic paraboloid
bounded by four 3D lines (every mesh vertex on z = xy/10), a loft
across two sketches, G1 and G2 measured across a cylinder's edge with
inspect.continuity (angle < 0.1°; curvature jump 1e-16 with G2 against
0.027 with G1), 26 hostile networks refused without panicking, and u/v
added to hostile_params_never_panic.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* feat(ui): Boundary Surface dialog
Curves and Cross curves take sketch curves and edges (picks that meet
end to end in one sketch or body are one chain); Continuity and
Operation (New Body or Join). Two to four sketch curves of one sketch
with nothing across are the loop patch; anything else is a network.
Edit Feature brings either back (per-side orders, crown and stations
the dialog does not show are kept). Strings in en, ja and zh.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* docs: ROADMAP M8 and the bridge patch's notes for boundary networks (#763)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* test(oracle): a boundary surface through a loop, thickened (#763)
Case 83: four sketch curves (a square's three sides and a half circle
for its top) make one boundary surface, thickened 2 mm; analytic values
from tools/oracle-analytic/cases_surfaces.py (V = 2A, area 2A + 2P,
centroid by moments). Recipes gain boundary_surface, thicken and remove.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
* refactor: boundary surfaces, review cleanups
Project onto the face once per derivative evaluation (the frame is
passed to natural() and reused by scale()), keep each track's
finite-difference step instead of rescanning its table, export the
kernel's edge-pick and chain caps and the default station count for
the engine, doc and dialog to share, map TooLarge spline errors to
KernelError::TooLarge, reuse the engine's operation parser, and skip
the dialog's sketch lookup when the picks are no loop.
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>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
On a busy CI machine (several fork jobs under gVisor at once) a web
offline launch took over 90 s and the icon review server took over ~2 s
to answer. Launches now get 240 s and one retry; the review server gets
up to 60 s to start.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#853 recorded seeds 13, 71 and 94 with the face and edge counts its own branch built
(40/101, 19/50, 27/63). With main's kernel they still build as they did before #853
(41/103, 20/52, 26/61), so every full goldens run has failed since it landed. This
records them again; #853's seed 20 change and the other 93 seeds are unchanged.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Main's CI has failed since 14:47 on `robustness_goldens::all_3`: seed 55 (a
filleted box cut by spheres and patterned) used to end with an invalid body and
now passes the validity check, and its volume moved by 1.3e-5 relative.
The outcome changed between 2ae21143 (main's last green run) and 7f9074d6,
where the kernel meshing change is #1670 (twisted ruled faces divided to the
chord tolerance). The volume changed between 8a055afd and 31b2ddc0, where
#1753 changed truck-meshalgo's triangulation (the faces of an edge share its
points exactly). The kernel's validity and volume checks measure those meshes.
Pull requests run every 8th seed, so only main's full run saw seed 55.
Only seed 55 is recorded, with the values main's CI printed on Linux.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LGeCa5jWWdroMWvXxN3LLM
Seed 31: #1581 refuses a fillet selection with an earlier round's smooth seam instead of re-rounding another nearby edge. Seed 83: #1316 was recorded before its branch took in #1495, whose shell keeps the wall beside a concave edge.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Chamfer: refuse a result that grew a convex edge's part (or shrank a concave one's)
A chamfer of a curved edge that runs off a face it skims (robustness seed 57's
ball rim) built a part bigger than before. Checked from the two faces' normals
at the edge; rounds and unclear angles are left to the wedge bound.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Action matrix: the sheet metal tray's relief corners refuse a 0.25 mm chamfer
Each is a 2.5 mm concave edge (the corner of a corner relief, between two
planes at right angles): a chamfer there fills at most 0.25²/2 × 2.5 = 0.08
mm³. What was built took 0.95 mm³ away; the chamfer guard now refuses it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Robustness goldens: seed 57's wrong-way chamfer is now refused (#1390)
On main with #391, seed 57's chamfer grew the part; this refuses it, so the run goes on
to its later steps and ends ok.
---------
Co-authored-by: Phanoob <phanoob@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
A point within the point tolerance of a boundary curve's end is that end in the boolean, whatever its parameter, so text cuts no longer leave unmatched sliver vertices (#391). Robustness goldens re-recorded for seeds 0, 34, 39, 41 and 57 with reasons.
A new face-based Extrude Join (and outward Press Pull) joins the picked face's body by default and stores that scope, so an edge-touching neighbour is no longer tried first or absorbed. Explicit targets and participants still win. Related to #226.
A round or chamfer whose edge chain runs on across a curved seam (the end circle of a lofted tube whose axis bends) is now built: the section solve keeps its contact on the seam through a plane taken afresh at the seam's nearest point at each step. Straight seams are unchanged. Robustness seed 41 is re-recorded: its second fillet, refused before, now builds. Fixes#1213.
Co-authored-by: gvocale <gvocale@users.noreply.github.com>
* Action matrix regressions of #1255: knuckle rounds, thin holes, a 45° flange hole, an axis after an edit
- Curve blends: past the end of a curved edge the run of sections along the end
tangent wrapped round the ring of centres; it is shortened while it still
crosses the end face (the hinge knuckle's cap rounds at 0.728 mm).
- Boolean retry guard: each body's volume is measured finer than its smallest
radius, and a volume out of bounds is measured again ten times finer before
it refuses (a 0.133 mm drill measured -30272 mm³; a plastic part's coarse
mesh was off by more than the slack).
- Sheet metal: a drill cap's certificate allows the evaluation noise its fitted
tilt shows (a hole square to a 45° flange).
- Construction geometry on a named face uses the face found by name, not a
second pick at its nearest point (an axis through a corner round after a
parameter moved it).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Retry guard: measure the target, tool and result at one tolerance
Measured at different tolerances, the chords' error on the faces they share
didn't cancel: a blind hole in a long D-profile bar measured 20 mm³ more than
the bar.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Action matrix baseline from this branch's full run: 37 actions pass again, none fewer
Blessed with cargo xtask action-matrix --from DIR --bless from the 16 shards of
run 38673859774.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
#1164 joins a body that sits flush on a rounded rim into it. In seed 13 the
rectangular pattern's second box (x 5..7.5) sits flush on the half-ring's top
face with a 27 mm2 contact patch. It used to stay a separate shell (one body,
three shells); now it joins the ring (area -54 mm2, twice the patch). The
fillet edge and hole position the generator picks next come from the new
geometry, so the end state differs: 3 bodies, 2647.95 mm3 (was 4, 2300.91).
Every step stays a sound solid, and no other seed moves.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Action matrix: nightly shards cut by measured cost, heavy designs in slices, and a determinism check
The full matrix ran as four fixed quarters of the corpus and two of them hit the 60-minute
job limit. Now action_matrix_shard runs shard k of n (ACTION_MATRIX_SHARD=k/n), planned from
tests/action_matrix/costs.json: designs that take more than five minutes are split into slices,
and the pieces go largest first to the least loaded shard. The nightly runs 12 shards and uploads
each one's outcomes and timings; cargo xtask action-matrix --from DIR --bless records them in the
baseline and costs.json without running the matrix again. action_matrix_determinism runs a sample
twice and lists any action whose outcome changed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Action matrix: costs measured on runners, 10 shards, determinism on the lighter designs
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Action matrix: baseline and costs from a full run on main, 16 shards, timeouts listed
The baseline is regenerated from the 10 shards' results of a full run on 01b1b705
(cargo xtask action-matrix --from DIR --bless): 8 new designs, 302 actions that pass now,
names that changed, and 40 actions that passed before and fail now, each listed in #1255
(22 fillets refused since #949, 4 hinge blends since #1101, 9 on the loft cut, 5 single).
costs.json now holds every design's measured time, and the nightly runs 16 shards of about
17 minutes. An action that passes in the baseline and times out is listed but not judged,
and stays passing when blessed: two runs of the same build differed only on 10 actions that
took 50-84 s against the 90 s limit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Action matrix: slice heavy designs by a hash of each action's name
Slicing by index gave some slices only fillets on oracle 41 (28 slices, a multiple of the
actions per edge), three times the planned time; two shards ran past their limit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>