Regression: CurvedSegment Ribs are too small and don't fit the hullcurve

[AIRCAP]

Commit ba13b70 reintroduced #34 - breaks CurvedSegments

[AIRCAP]

Probably tessellate is the wrong way to fix this. Shape.BoundBox should not be used anywhere in an execute() function, since its exact values

a) depend on whether the part has been tessellated
b) depend on the GUI Shape meshing settings

to make the shape calculation independent of GUI settings in FreeCAD preferences function Shape.optimalBoundingBox() - which does not rely on tessellation. this could worst case be a lot slower than Shape.BoundBox, which uses the tessellated GUI mesh to calculate an approximate boundingbox

[AIRCAP]

to be exact, the right call to give the same values seems to be
optimalBoundingBox(False,False)

[AIRCAP]

actually its not that easy. I just had a SubShapeBinder to a sketch that for the Z boundingbox reports +- epsilon even with (False,False) with optimalBoundingBox() (or optimalBoundingBox(True) it gets over a millimeter in (nonexisting length)

that means for the calculation whether a boundingbox is "flat" in a dimension which CurvedArray does all the time, the Shape's tolerance has to be taken into account (epsilon has to be set to 2* tolerance - not Precision:Confusion() )

[AIRCAP]

I take it back, even a 2D part with high tolernace optimalBVoundingBox(False,False) doesn't seem to ever get thicker than 2e-7 - so increasing epsilon by factor 4 should be sufficient

[AIRCAP]

let's do factor 10 just to be on the safe side

[AIRCAP]

#53 should be a good workaround for this i hope