Expand description
Tolerances.
See docs/DATA_MODEL.md §5 and §9. Two separate things live here:
- Global constants: the thresholds at which the kernel decides two
things are “the same”.
Tolerancescarries them, parameterised by the model’s unit scale. - Per-entity tolerances:
Tolerance, the radius attached to an individual vertex, edge or face, together with the containment rule that relates them.
The unit scale is explicit. Kernels commonly hard-code a confusion
tolerance of 1e-7 with an undocumented assumption that models are in
millimetres, which then misbehaves silently on models authored in metres or
inches.
Structs§
- Tolerance
- The tolerance carried by a single vertex, edge or face: the radius within which the entity is considered to lie.
- Tolerances
- The threshold constants the kernel decides identity by, for a given model scale.
Constants§
- ANGULAR
- Two directions closer than this in angle are parallel. Dimensionless.
- APPROXIMATION
- Accuracy target when fitting a curve or surface to data, at unit scale.
- CONFUSION
- Two points closer than this are the same point, at unit scale.
- DEGENERATE
- A length below which an entity is degenerate rather than merely small, at unit scale.
- INTERSECTION
- Convergence target for intersection algorithms, at unit scale.
- P_
CONFUSION - Confusion in parametric space. Dimensionless.
Functions§
- check_
containment - Check the containment rule
tol(vertex) >= tol(edge) >= tol(face)for a boundary relationship.