That would definitely be welcome! I'm not sure exactly what you were envisioning, but I think it makes the most sense to have a geoparquet as a distinct type, separate from parquet.

In case they are useful places to start, on the pin_read side, boards use the load_data function to bring in the data. And then when pin_write is called, it is the save_data function that does the actual handling depending on data type.

Definitely makes sense to add as a distinct type.

Given that all GeoParquet files use the same file extension .parquet and are valid Parquet files in their own right (it is a sub-specification), there could be some argument that using "parquet" with pin_write should automatically detect that the object is a GeoDataFrame and so use gpd.to_parquet instead of pd.to_parquet. But this is really just for convenience and could be added later. For now, I'll just focus on adding it as a distinct type.