Skip to content

SolidCraft: an open-source parametric 3D CAD application built in pure Rust. Desktop, web, and automation through MCP.

Wave 483/509: Kernel rig: the two G0 window patches graded on the box the part they close keeps (and 3 more)

jyvikoc35be9d20e1a
diff --git a/tests/kernel-cases/sheet-metal/296-hole-in-a-base-with-two-adjacent-edge-flanges/expected.json b/tests/kernel-cases/sheet-metal/296-hole-in-a-base-with-two-adjacent-edge-flanges/expected.json
index 7b3b863b02..0967020c04 100644
--- a/tests/kernel-cases/sheet-metal/296-hole-in-a-base-with-two-adjacent-edge-flanges/expected.json
+++ b/tests/kernel-cases/sheet-metal/296-hole-in-a-base-with-two-adjacent-edge-flanges/expected.json
@@ -4,7 +4,8 @@
"tangency-labs/solidcraft#355"
],
"what": "an 80 x 40 base with a 30 mm flange on the y = 0 edge and a 15 mm flange on the x = 80 edge, and a round hole through the base beside them: sew failed with a non-manifold edge contact",
- "source": "invariants",
- "method": "the hole is cut straight through the flat base, clear of both bends and of the corner, so it must give one valid closed body; the corner geometry of two adjacent flanges has no simple closed form, so no volume is graded",
- "body_count": 1
+ "source": "analytic",
+ "method": "T = 2, R = 2, 90 deg flanges, setback R + T = 4 (K = 0.4 moves only the flat pattern, not the folded part). The earlier flange keeps its length and the later one ends at its bend, and the later one trims the base under the first bend (the convention oracle 75-sm-bracket-holes is graded on): base x 0..76, y 4..40 (5472); bend 1, a quarter annulus of radii 2..4 (3 pi a mm), x 0..76 (228 pi); flange 1's panel y 0..2, z 4..30, x 0..80 (4160); bend 2 y 4..40 (108 pi); flange 2's panel x 78..80, z 4..15, y 4..40 (792); the hole r 3 at (25, 20) straight through the base, clear of both bends and the corner, takes 18 pi. V = 10424 + 318 pi = 11423.026464",
+ "body_count": 1,
+ "volume_mm3": 11423.026463841552
}
\ No newline at end of file
diff --git a/tests/kernel-cases/sheet-metal/978-edge-flange-beside-a-ledge-as-tall-as-its-setback/expected.json b/tests/kernel-cases/sheet-metal/978-edge-flange-beside-a-ledge-as-tall-as-its-setback/expected.json
index b6d90092fd..5f1042e201 100644
--- a/tests/kernel-cases/sheet-metal/978-edge-flange-beside-a-ledge-as-tall-as-its-setback/expected.json
+++ b/tests/kernel-cases/sheet-metal/978-edge-flange-beside-a-ledge-as-tall-as-its-setback/expected.json
@@ -3,7 +3,11 @@
"tangency-labs/solidcraft#978"
],
"what": "a 90 degree inside flange of radius 1 on an edge that has a ledge beside it exactly as tall as the flange's setback (3 mm): sew failed with a degenerate edge",
- "source": "invariants",
- "method": "T = 2, flange R = 1: setback R + T = 3 and the ledge (x 97..100, y 0..3) is exactly that tall, so the trim line runs along the ledge's top; the right answer is one valid closed body (the ledge goes with the setback strip), with no graded volume because the flange's extent along the ledge has no single reading",
- "body_count": 1
+ "source": "analytic",
+ "method": "T = 2, flange R = 1, inside: the setback R + T = 3 strips y 0..3 off the base, and the ledge (x 97..100, y 0..3) lies wholly in that strip, so the base left is x 0..97, y 3..60 (11058). The bend (a quarter annulus of radii 1..3, 2 pi a mm) and the panel (leg 20 - 3 = 17, 34 a mm) run along the edge either as far as the base's remaining edge (97) or the picked edge (100), the reading the case leaves open: V lies between 11058 + 97 (2 pi + 34) = 14965.468975 and 11058 + 100 (2 pi + 34) = 15086.318531; a flange missing, or the base left untrimmed by the setback, falls outside",
+ "body_count": 1,
+ "volume_mm3": [
+ 14965.468,
+ 15086.319
+ ]
}
\ No newline at end of file
diff --git a/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-240-deg-window-in-a-cylinder/expected.json b/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-240-deg-window-in-a-cylinder/expected.json
index 017bac7a06..d76f3ce08f 100644
--- a/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-240-deg-window-in-a-cylinder/expected.json
+++ b/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-240-deg-window-in-a-cylinder/expected.json
@@ -3,7 +3,19 @@
"tangency-labs/solidcraft#759"
],
"what": "a window 240 deg round a cylinder's wall (r10 x 40, z 15..25), the faces the cut made taken off and the window patched g0 and stitched",
- "source": "invariants",
- "method": "the window (as the g1 case) filled G0: the fill meets the window's edges; its shape is the fill's own, so one closed valid solid is the right answer",
- "body_count": 1
+ "source": "analytic",
+ "method": "the box x -5..11, y -11..11, z 15..25 cuts the wall where 10 cos a >= -5 (|a| <= 120 deg), as in the g1 case; filled G0, the fill meets the window's edges and its shape is its own, so no volume is graded. The wall left below z 15, above z 25 and round a = 180 deg reaches x = +-10, y = +-10 and z 0..40, and a fill of a rim lying on the cylinder stays inside that box: bbox -10..10, -10..10, 0..40",
+ "body_count": 1,
+ "bbox": {
+ "min": [
+ -10,
+ -10,
+ 0
+ ],
+ "max": [
+ 10,
+ 10,
+ 40
+ ]
+ }
}
diff --git a/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-saddle-window-in-a-torus/expected.json b/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-saddle-window-in-a-torus/expected.json
index 7fafe45a83..cc75244616 100644
--- a/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-saddle-window-in-a-torus/expected.json
+++ b/tests/kernel-cases/surfaces/cov-patch-g0-fills-a-saddle-window-in-a-torus/expected.json
@@ -3,7 +3,19 @@
"tangency-labs/solidcraft#759"
],
"what": "a window 240 deg round the saddle-shaped inner wall of a torus (R20, r8) between z = -3 and 3, its cut faces taken off, patched g0 and stitched",
- "source": "invariants",
- "method": "the box x -6..13, y -13..13, z -3..3 opens the inner wall where 12 cos a >= -6 (|a| <= 120 deg): its rim is two arcs on z = +-3 each way and two curves on x = -6, and it bounds no height field. G0: the fill meets the rim. The fill's shape is its own: one closed valid solid is the right answer.",
- "body_count": 1
+ "source": "analytic",
+ "method": "the box x -6..13, y -13..13, z -3..3 opens the inner wall where 12 cos a >= -6 (|a| <= 120 deg): its rim is two arcs on z = +-3 each way and two curves on x = -6, and it bounds no height field. G0: the fill meets the rim. The fill's shape is its own, so no volume is graded. The box opens only the inner wall, so the torus keeps its outer equator (r 28, x and y to +-28) and its top and bottom circles (r 20, z +-8), and the fill of a rim on the inner wall stays inside that box: bbox -28..28, -28..28, -8..8.",
+ "body_count": 1,
+ "bbox": {
+ "min": [
+ -28,
+ -28,
+ -8
+ ],
+ "max": [
+ 28,
+ 28,
+ 8
+ ]
+ }
}
diff --git a/tests/kernel-cases/sweeps/4280-circular-pattern-of-sweep-cut-groove-on-hub/expected.json b/tests/kernel-cases/sweeps/4280-circular-pattern-of-sweep-cut-groove-on-hub/expected.json
index b65c543054..c429109a92 100644
--- a/tests/kernel-cases/sweeps/4280-circular-pattern-of-sweep-cut-groove-on-hub/expected.json
+++ b/tests/kernel-cases/sweeps/4280-circular-pattern-of-sweep-cut-groove-on-hub/expected.json
@@ -3,7 +3,8 @@
"tangency-labs/solidcraft#4280"
],
"what": "a round groove (r2) swept as a cut along the curved profile of a revolved hub, then circular-patterned 6 times about the hub's axis: 'boolean Cut: no result' in the pattern",
- "source": "invariants",
- "method": "the six grooves run down the hub's curved face 60 deg apart and don't meet, so the hub stays one valid closed body. The groove follows the surface (its centre on it), which has no closed-form removed volume",
- "body_count": 1
+ "source": "analytic",
+ "method": "the hub: the profile x 0..30 at z 0, up x = 30 to z 5, the arc through (30, 5), (18, 11), (10, 20) (centre (38.75, 37.5), R 33.657280, the side away from it, x(z) = 38.75 - sqrt(R^2 - (z - 37.5)^2)), then z = 20 to the axis, revolved about z: V_hub = pi (30^2 x 5 + Int_5^20 x(z)^2 dz) = 29625.699466. Each groove takes the r2 tube about that arc (length 25.613619, flat ends square to it at both ends, its centre on the hub's surface) where it lies in the hub: Int_0^L ds Int_0^2 rho drho Int (1 - rho cos psi / R) [in the hub] dpsi, psi from the arc's normal toward its centre, integrated numerically from that exact geometry (Gauss-Legendre in s and rho, each inside run of psi to its crossings): 162.452357, steady to 3e-6 relative when the grid is doubled (half the tube is 160.935; the turn round the hub's axis and the end wedges move it). The six grooves, 60 deg apart at x >= 8, don't meet: V = V_hub - 6 x 162.452357 = 28650.985325; one valid closed body",
+ "body_count": 1,
+ "volume_mm3": 28650.98532494651
}