Calculating distances with np.nan gives incorrect result

Running the following code:

from shapely.geometry import Point
import numpy as np

a = Point(1, 1)
b = Point(np.nan, np.nan)

a.distance(b)
a.hausdorff_distance(b)

will output np.nan for the hausdorff_distance which I believe is correct but the distance function outputs a huge number. Shouldn’t it output np.nan?

[jorisvandenbossche]

Opened https://github.com/libgeos/geos/issues/569 on the GEOS side

[jorisvandenbossche]

All the example cases from above, both with empty Point or empty GeometryCollection, and both with distance and hausdorff_distance, now return NaN with the latest GEOS version on the main branch.

So I think this can be closed?