Rather than explicitly storing the bounding box and adding the extra step of injecting non-standard attributes, we should instead make the first four rows of the Points table "special". The first four rows should instead correspond to the points whose coordinates are on the extremes of the axes. We can then always load the first row group to identify the bounding box of the points, and we can update the frontend logic so that it knows that the z-order curve sorting begins after these initial "special" rows.
|
def sdata_points_write_bounding_box_attrs(sdata, element) -> dd.DataFrame: |