Inconsistent return types in GeoPath.track_distance about nvector

[pbrod]

Ok. I fixed the issue in commit f9bfc79

Now GeoPath, GeoPoint, Nvector and ECEFvector and Pvector return scalars for the case where the input is not actually arrays of points but just single objects.

[pbrod]

Thanks for the notification on this inconsistency.

Actually, both GeoPath(pointA, pointB).track_distance(method='great_circle') and
GeoPath(pointA, pointB).track_distance(method='euclidian') returns an array of m = max(len(pointA), len(pointB)) values.

Note this works either if len(pointA)==len(pointB) or len(pointA)==1 or len(pointB)==1.
In other words it returns all paths between point A and B even if one or both are given as an vector of points..
However, GeoPath(pointA, pointB).track_distance(method='exact') raises a ValueError for m>1.

Also
GeoPath(pointA, pointB).cross_track_distance(pointC, method='great_circle') and
GeoPath(pointA, pointB).cross_track_distance(pointC, method='great_circle') both returns arrays of length m
where m = max(len(pointA), len(pointB), len(pointC)) and either len(pointA)==len(pointB)==len(pointC) or len(pointA)==1 or len(pointB)==1 or len(pointC)==1.

Thus, I am inclined to document this vectorized behavior and make sure GeoPath.track_distance(method='exact') respond in the same way.

[freol35241]

@pbrod

Thanks for the explanation! I actually had no idea GeoPath supported vectorized operations.

Still, in order to not confuse the user (me 😉) to much it would be nice to return scalars for the case where the input is not actually arrays of points but just single objects. Compare for example with numpy where functions such as numpy.add returns a scalar if it is provided with scalar inputs but an array if it is provided with arrays as input.

While on the topic of type inconsistencies, I also noticed that GeoPoint sometimes returns latitude and longitude as scalars and sometimes as length 1 arrays. I assume this has a similar reason/explanation?