Fixes ALS017 (IVS-282) #328
Conversation
…ment representations The length of each segment corresponds to the total length along the parent curve. However, for continuity calculations, we need to provide the total length along the *base* curve. (IVS-282)
features/steps/utils/geometry.py
Outdated
| u = abs(self.segment_to_analyze.Placement.Location.Coordinates[0] - \ | ||
| self.previous_segment.Placement.Location.Coordinates[0]) |
There was a problem hiding this comment.
Isn't this a bit cyclical? We want to assess basically whether segment to analyse and previous segment align, but if we do that based on assessing the respective placements we would miss some cases where the length of previous is actually not enough to reach the placement of current?
In other words: this assesses continuity of vertical in the Y axis, but doesn't account for gaps in the domain of the vertical profile, Is that a problem?
There was a problem hiding this comment.
Potentially. I had gotten wrapped around the axle with confusing length along the base curve versus segment length of the parent curve. Using the placements was the quickest way to resolve this as well as confirm my understanding of the piecewise function implementation.
The cleanest way forward IMHO is to add a property endPoint with type taxonomy::cast<taxonomy::matrix4d> to curve_segment_evaluator given that we have both length_ and projected_length_ available. Then (I think) we can just call pwf.end_point() from python to get the calculated point at the end of the segment - versus dealing with the complication of length versus projected_length on the python side.
I'm keen to take this on as it will confirm my burgeoning C++ capabilities 😉.
There was a problem hiding this comment.
IfcOpenShell PR#6006 now provides correct values of function_item.length() for vertical alignment (IfcGradientCurve) segments. Our calling code was adjusted accordingly.
Updated to current geometry evaluation methods for alignment segments. Adjusted IFC test files to reflect corrections to direction vectors and parabolic segment lengths. Updated CI pipeline to use the latest IfcOpenShell build. (IVS-282)
|
I also made some adjustments / corrections to the unit test files:
|
ifcopenshellcorrects issue with parabolic vertical curves in imperial unitsuinevaluate(u)for calculating placement transforms at the ends of alignment segments.