Hvplot and vdims

dwr-psandhu · October 22, 2024, 12:10am

I typically use hvplot to get started and then try to customize it using holoviews methods. However one difference is that hvplot (especially for geo=True) drops all the other value dimensions by default.

github.com

holoviz/hvplot/blob/90783783b91b9d42a6970bf402e012ec9fd3e73e/hvplot/converter.py#L2908


      
              if is_geodataframe(data):
                  x, y = ('Longitude', 'Latitude') if self.geo else ('x', 'y')
              elif self.gridded_data:
                  x, y = self.variables[:2:-1]
              else:
                  x, y = data.columns[:2]
          
          redim = self._merge_redim(
              {self._color_dim: self._dim_ranges['c']} if self._color_dim else {}
          )
          kdims, vdims = self._get_dimensions([x, y], [])
          if self.gridded_data:
              vdims = Dataset(data).vdims
          element = self._get_element(kind)
          cur_opts, compat_opts = self._get_compat_opts(element.name)
          for opts_ in [cur_opts, compat_opts]:
              if 'color' in opts_ and opts_['color'] in vdims:
                  opts_['color'] = hv.dim(opts_['color'])
              # if there is nothing to put in hover, turn it off
              if 'tools' in opts_ and kind in ['polygons', 'paths'] and not vdims:
                  opts_['tools'] = [t for t in opts_['tools'] if t != 'hover']

Unless added to the hover tooltips. Isn’t it better to just do what the default holoviews elements do which is to add all other columns to the vdims. This helps with customization of that object later.

Thoughts?