curveBoolean Bug near Tolerance

curveBooleanFail.3dm (22.8 KB)

Screenshot 2021-04-13 at 20.54.40

reproduce: select all, call curveBoolean, click somewhere in the left region.

I know that it is beyond or close to document tolerance - but as all curves are lines / polylines i would expect a proper result.

_intersect finds all expected intersection-Points.

thanks for having a look at it. - kind regards -tom

Hi Tom - I guess I do not understand - this is indeed looking at distances well below tolerance - if you decrease the tolerance, then it works. I’m not sure how else it could work…


Dear @pascal thanks for having a look at this.
I think tolerances should only take account / be applied where they are needed - for sure some algorithms that find their solution by recursion can use tolerance as stop-value.
But operating with 2d-Lines and finding intersections and curve parameters should be some simple matrix / vector stuff and not depend on a tolerance-mechanism.
ok - so maybe it is not a bug, but a question:
why are tolerances applied in this case, where the vector-math behind it could give a fast and more precise result?
kind regards -tom