Dear tcxhb
I work on this API sometimes. Is there a specific case where you want to log a bug? If so, please consider submitting an incident to customer support.
In general, the algorithm has two steps: feature detection (i.e., finding places on the ellipse that need to be checked for the closest point, using boxing, or some other strategies), relaxation to a precise answer (e.g., with something like newton's method).
It is rare, but concievable, that there were problems in either phase of the algorithm. Without looking at specifics, it is difficult to tell what the problem is.
One should also keep in mind, that the closest point query (which point on a wire is closest to a point), may have multiple answers: e.g., if the point is exactly in the center of a circle there will be infinitely many equally correct closest points! If this sort of situation is occuring, it may also cause problems for the algorithm. Having multiple closest points is a corner case that may not have beenconsidered by the original implementors.
without more information, that is the best I can tell you. Please either post a sat file here, or log a bug for this, so I or another Spatial employee can resolve the bug for you.
best regards
Eric Zenk
