It's difficult to teach users why they should have to use first(geometrycolumns(x)), and it's not easy to remember when starting out.
Would it make sense to have a convenience function geometrycolumn(x)? Then things like df[:, Not(GI.geometrycolumn(df))] just work, without exposing users to the annoyances inherent in geospatial datasets.
It's difficult to teach users why they should have to use
first(geometrycolumns(x)), and it's not easy to remember when starting out.Would it make sense to have a convenience function
geometrycolumn(x)? Then things likedf[:, Not(GI.geometrycolumn(df))]just work, without exposing users to the annoyances inherent in geospatial datasets.