recursively calling this is doing a deep copy of the whole subtree at every level. Might be better to only do the copy once at top level.

Hi Fil, I also have problem using d3.geocentroid and d3.geopath with this Geojson encoded in WGS84 planar coordinates. I use turf.centroid instead of d3.geocentroid for calculating center of the polygon but don't know how to overcome the problem with d3.geopath?

Can you share a notebook showing the issue?

Here is the notebook: https://observablehq.com/@cadasa/geojson_encoded "geojson encoded" is the one with d3.Geocentroid and d3.Geopath problems

Oh, yes. The issue with that dataset is not linked to the line resampling problem of this notebook. Rather, it is because the winding order of its polygons is not normalized. turf doesn't care about that, as it works in planar coordinates, but D3 makes a difference between clockwise and anticlockwise. See https://github.com/d3/d3-geo/issues/138#issuecomment-392842991 or https://bl.ocks.org/mbostock/a7bdfeb041e850799a8d3dce4d8c50c8 for an illustration of the issue. I'll send you a fix as a suggestion to your notebook.

thanks Fil!