Bad IfcGradientCurve.EndPoint and IfcSegmentedReferenceCurve.EndPoint not detected

[RickBrice]

IfcGradientCurve and IfcSegmentedReferenceCurve both have an option EndPoint attribute. When provided the EndPoint should exactly match the placement of the zero-length segment. When this is not the case, a best practice issue should be raised by the validation service.

Attached are two nearly identical files. The only difference is I manually messed up the EndPoint on the file ending with "_Bad_EndPoint" to test the validation service on this point.

Generated_Cant_CosineCurve_100.0_1000_300_1_Meter.ifc.txt
Generated_Cant_CosineCurve_100.0_1000_300_1_Meter_Bad_EndPoint.ifc.txt

[civilx64]

We could also make this a best practice rule that the EndPoint attribute should always be left blank. I remember some discussions about an implementer agreement to that effect but not sure where or even if it was documented.

[RickBrice]

Either solution would be fine. This is an insignificant boundary case that has an undefined resolution.

[civilx64]

I did some research and could not find anything concrete re: leaving it blank. But it should be relatively easy to check the value against the start of the zero-length segment as you had suggested.