UserWarning from geopandas' to_file() with GeoSeries.notna()

[mwtoews]

I'm seeing a UserWarning with geopandas-1.0.0 while using the to_file() method.

/tmp/py310/lib/python3.10/site-packages/pyogrio/geopandas.py:523: UserWarning: GeoSeries.notna() previously returned False for both missing (None) and empty geometries. Now, it only returns False for missing values. Since the calling GeoSeries contains empty geometries, the result has changed compared to previous versions of GeoPandas. Given a GeoSeries 's', you can use '~s.is_empty & s.notna()' to get back the old behaviour.

To further ignore this warning, you can do: import warnings; warnings.filterwarnings('ignore', 'GeoSeries.notna', UserWarning) has_z_arr = geometry[geometry.notna() & (~geometry.is_empty)].has_z

And checking the code here: https://github.com/geopandas/pyogrio/blob/af292e579572a6a33a22a6403873b8e7b0a9d7f6/pyogrio/geopandas.py#L523

A simple fix would be to apply the suggestion from the warning.

has_z_arr = geometry[~geometry.is_empty & geometry.notna()].has_z 

[martinfleis]

Thanks!

The snippet won't fix it as it just inverts the order of conditions but notna() will emit the warning anyway. We shall suppress the warning here.

[mwtoews]

Thanks for the tips, makes sense to follow the second suggestion. I can prepare a test and PR for this one.