Joris Van den Bossche
Joris Van den Bossche
This is green now. It's certainly not the nicest PR, but I think it should be good to merge this soon, since it will make it actually possible to run...
Failures are GDAL/Fiona related installation issues (following the CMake change in conda-forge, I assume)
Note for myself -> still need to add this to the whatsnew file
A related question that might indirectly answer this issue: assume that long term we change GeoDataFrame to error if there is no geometry input (like we are doing for GeoSeries,...
_If_ we would like to handle empty GeoDataFrames, thinking about a possible way to do this: currently we set the `_geometry_column_name` to `"geometry"` even if there is no geometry column....
@chaoyanghe you have a problem with the installation of fiona / gdal. Can you give more details how you installed geopandas and its dependencies? (and can you show the output...
@zlqs1985 can you show the output of the install command?
Yes, it's not the idea to leave out pyproj for the typical installation of geopandas, just to make it *possible* to run geopandas without having it installed. That's also what...
Our native `read_parquet`/`to_parquet` (using pyarrow directly, instead of through GDAL) are _much_ faster, so I don't think we will remove those anytime in the near future (and the same for...
To give one example to illustrate the order of magnitude difference (but the exact difference will depend a lot on the characteristics of the data): ``` In [2]: %time df...