Skip to content

solidcore: a curved Setback that no fill certifies up to twice its setbacks is refused untyped, naming no size #16398

Open

Description

@brett19

Found while measuring #13832 step 5 (#16370): a curved Setback that no fill certifies up to twice the rule's setbacks is refused as an untyped failure that names no size.

Repro (test hip_roof_apex_of_unequal_rounds_is_refused_as_data in solidcore/crates/solidcore/tests/issue_2343_curved_setback.rs on #16370, #[ignore]d because it fails on main):

  • Part: a hip roof with one curved slope. It is the intersection of a prism along x (section: z >= 0, the plane z = 6 + 3y/4 for y < 0, and the circle of radius 5 about (y, z) = (-3, 2)) and a prism along y (the triangle (-6, 0), (0, 6), (6, 0) in (x, z)).
  • The apex V = (0, 0, 6) is a convex vertex of four edges. Its faces are three planes and a convex cylinder.
  • Round the four apex edges 1, 1.25, 1.5 and 1.75 together (fillet_setback_beside_curved_faces on).

On main (f910ff8, MEASURED):

  • classify_setback takes the corner as a Setback, with rule setbacks [1.575, 1.875, 2.25, 2.625].
  • No G1 fill up to degree 9 certifies at 1x or at the growth's first trial, 2x. At degrees 6 to 9 the QP solves, but every net fails the exact curvature certificate (certified_sign / t_nonneg checks). Degree 5 is infeasible, and 3 and 4 have no affine solution.
  • grow_curved returns SetbackRefusal::no_room("... no setbacks up to twice the rule's ...", Some(0.0)) with no .typed(...). The session then returns SolidError::Failed { operation: "blend::fillet_executive::session::finish", reason: Blend(Failed("the fillet body is not built")), status: 1001 }, with fit_setback() = 0.
  • The same path serves the asked form ("no setback up to twice {a} ..."), e.g. asked 4.0, and three-round corners.

Expected: a typed SolidError::BlendRefused that carries the cause and its site (the vertex). If no size can be named, say so in the type: §7.3 asks that a refusal names the size that fits, from the cause.

Related gap (MEASURED, not a wrong body):

  • Rounds 1, 1.1, 1.2, 1.3 are refused the same way.
  • 1, 1, 1, 1 and 1, 1, 1, 1.25 build and grade: check_brep level 2, seams G1 to 4.3e-12 rad against the exact normals, convex, contained.
  • The planar twin, with the cylinder replaced by the plane z = 6 - 3y/4, builds 1, 1.25, 1.5, 1.75.
  • So four mildly unequal rounds beside a cylinder never certify, because the QP's sector rows do not imply the net certificate there. The refusal is on the conservative side. Whether a certified fill exists for these radii is not shown.

Typed-errors tracker: #13790. Curved Setback hub: #13832.

Written by Claude Code (Anthropic) on behalf of @brett19.

Metadata

Metadata

Assignees

No one assigned

    Labels

    needs-triageno priority yet; a maintainer sorts it with /triage

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions