Skip to content

Next Gen STEP AP242 Native-ish BREP Geometric Modeling Kernel

2.8 KB
Raw Blame

M4 intersection authority audit

The audit covered Kernel.Core, BRep Boolean routes, analytic cone/plane support, CLI line/surface sectioning, clipping and triangulation, Panel boundary and mate checks, Drawing HLR, SurfaceMeshIR, STEP import/recovery, and restricted-field trim prototypes.

Area Classification Authority assessment M4 action
TransverseConePlaneIntersection bounded constructive, analytic healthy bounded construction: caller intent fixes cone/world-Z section and hyperbola family unchanged; regression tested
BrepBoolean and recognized family builders topology-authoring, bounded recognized cases contained legacy behavior; topology is explicitly requested by a Boolean and routed through recognized families/JudgmentEngine unchanged; generic query has no dependency on it
STEP analyzer plane sections and ray hits internal query-like; analytic plus tessellated fallback contained analysis behavior; section output is explicitly requested and provenance identifies fallback unchanged
STEP importer polygon crossings/recovery internal topology recovery, approximate planar predicates future architectural risk because imported loop-role recovery can affect topology; already bounded by tolerance, diagnostics, and JudgmentEngine documented, not refactored in M4
Firmament restricted-field/marching contours approximate/sampled, internal-only contained legacy prototype; records say numerical-only/not exportable and BRep topology not emitted unchanged
Profile arrangement line/arc intersections topology-authoring inside explicit profile construction healthy bounded construction: authored profile intent requires splitting/reconstruction unchanged
Planar triangulation, display-loop checks, feature-band clipping internal-only planar predicates healthy implementation detail for already-authored topology/display unchanged
Drawing projection/HLR segment cuts approximate display-only no model authority unchanged
SurfaceMeshIR and display tessellation sampled/approximate visualization no topology authoring; analytic BRep remains owner unchanged
Panel boundary self-crossing and G0/G1/G2 mates sampled validation evidence healthy query-like validation; does not create topology unchanged; Panel patches dogfood M4
BRep interference and AABB overlap generic query-like, bounded observational verification/candidate pruning unchanged

No release-blocking path was found where an arbitrary M4 numerical witness directly becomes an edge, trim, face, or Boolean result. The principal future risk is STEP recovery, where approximate planar crossing facts participate in import normalization; that path is explicit, diagnosed, and utility-selected, so a broad M4 refactor would increase risk without strengthening the new public predicate.