Closed benbovy closed 1 week ago
We should fix that there if needed.
I think I assumed that an empty Linestring would have no polyline, not an empty polyline? But that assumption might have been wrong?
Yeah the creation functions added in #51 currently add one empty polyline, but it is probably cleaner indeed to implement the empty case as no polyline here (there might actually be another bug since actually Geography.m_empty
is set to true when there is no polyline rather than when there is an empty polyline).
In any case, "LINESTRING EMPTY"
is being tested on the s2geography side, so I assume the creation methods there also create a PolylineGeography with an empty vector
Yes we should fix this in the Spherely creation functions.
Looks like the CI failing tests are caused by the changes in https://github.com/paleolimbot/s2geography/pull/37, which now raises an exception for empty S2Polyline. @jorisvandenbossche do you think we should fix that in s2geography or work around it in spherely to support empty linestrings?