Relationship of IfcGradientCurve.EndPoint, IfcSegmentedReferenceCurve.EndPoint with zero-length segment

[aothms]

As noted by @RickBrice https://github.com/IfcOpenShell/IfcOpenShell/pull/3982/files#diff-be0a3e3e00f8d8a24a29c355bd022eb663c4310e27251191374b9aa6d05ab4b4R100

This seems redundant.

If the geometry definition is also present, then each of the zero-length segments shall have a IfcCurveSegment counterpart - of length zero.

https://ifc43-docs.standards.buildingsmart.org/IFC/RELEASE/IFC4x3/HTML/concepts/Product_Shape/Product_Geometric_Representation/Alignment_Geometry/Alignment_Geometry_-_Segments/content.html

Should we deprecate these EndPoint attributes, as:

• not consistent with business logic layer, where we can only have zero-length segments
• not consistent with compcurve for horizontal, which doesn't have this attribute

CC @evandroAlfieri @peterrdf

[evandroAlfieri]

future release?

[peterrdf]

Maybe, still something to think about. My higher level view would be that such data would be better fitted in schema semantics, i.e. the attribute EndPoint compared to specific use of a curve segment, i.e. last curve segment with length 0. I am not sure what is the best option at the moment but I like having such an attribute better compared to 'rules' on use of existing properties, still I see also your point.

[evandroAlfieri]

any warning or "temporary" check we can introduce for files validation? (till the time where this will be fixed)

[RickBrice]

This may not be a viable solution, but I'll share my thought. The zero-length segment requirement doesn't seem necessary in the business logical because that information is not for geometry, but rather checking business-logic rules like "is the design speed too fast for this curve?". The end point of the IfcAlignmentHorizontal/Vertical/Cant occurs at the end of the last item in the list of IfcRelNests.RelatedObjects. The zero-length segment is not needed to identify the end point.

Without the zero-length segment in the IfcCompositeCurve/GradientCurve/SegmentReferenceCurve.Segments the EndPoint becomes non-redundant and relevant. EndPoint could be promoted to IfcCompositeCurve and a rule defined that states EndPoint is required for IfcSegment.TransitionCode = .DISCONTINUOUS.

It may be too late to do something like this. There may be implications for backwards compatibility that I don't yet understand. It's just an idea.

For validation checking, my first thought is this:

1. Verify zero length segments are provided as specified
2. Confirm geometry of zero-length segment between business-logic and geometry are the same
3. If EndPoint is present, confirm it is the same as the zero-length segment in the geometry.