Files
solidcraft/crates/kernel
brett19andClaude Opus 5.5 3b4adeb73b solidcore: a variable round is built again, as the union of its balls
With #2262 and fillet_gn_section_generator on, solidcore builds a
variable round as the union of its balls, so fillet_variable's
refusal from sync-2 (9f64cce54, solidcore#2250) comes off. Laws with a
corner stay refused (solidcore#430); a law whose balls fold is refused
by solidcore (NoRoomReason::BallsFold) and arrives as Error::NoRoom.

Against closed forms, solidcore and SMLib d01d7b0 agreeing to 1e-11:
- 1 to 3 along 30 mm removes 27.858907 against the balls' union's
  27.857434 (great circles 27.898239); a law of one radius 2 removes
  25.753574 against (1 - π/4) 4 · 30 = 25.752220: both 5.3e-5 over,
  the solver's surface fit, as on the C++ kernel.
- the rig's 1 to 5 cone: 11956.21991 against 11956.22243 (2.1e-7).
- kernel tests: a_linear_radius_runs_from_the_start_end (18.5378),
  variable_radius_on_a_box_edge, down_to_zero (22.5949),
  taper_follows_the_room (the wedge 5245.68 against the balls'
  union) pass.

The refused test becomes a_variable_round_is_the_union_of_its_balls
(the two closed forms, either direction, a folding law refused, a pick
miss named first). the_wedge_round_keeps_its_volume held 5223.0386,
measured before the fork's balls' union; it now takes its closed form
from removed_by_balls.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SdcgEu4Lr3moCCisdZShKz
2026-11-01 18:32:30 -07:00
..