For some cities only a square is displayed

[sebkur]

For some cities, the expected suburb boundaries do not appear, instead there only a giant square on the map.

These are cities for which it doesn't work here:

• Augsburg
• Braunschweig
• Bad Belzig
• Wiesenburg
• Chemnitz
• Potsdam

for most other cities everything works fine.

I have the impression it might have something to do with the extend of the city, the ones above are all rather small ones, so that the zoom level is rather high (i.e. zoomed in).

[sebkur]

I have looked into the Chrome Developer Tools and looked for something suspicious happening but didn't stumble upon anything that looks like it could be the cause for this. I don't think it is a problem with the data as others who also experience the problem told me they used to work before.

[epaulson]

@sebkur can you check out the geojson winding order of those - see what we did in #332

[sebkur]

Thanks, indeed, that seems to do the trick. See #342 for Bad Belzig. When this passes review I can also fix the others.

[sebkur]

I also detected a few more broken ones:

• albuquerque.geojson
• arlingtonva.geojson
• copenhagen.geojson
• durham.geojson
• melbourne.geojson
• taiwan.geojson

Those also have some polygons that are not in clockwise order, however I wasn't able to see that anything is broken in the web app:

• chapel-hill.geojson
• korea.geojson
• london.geojson
• olympia.geojson
• poland-parks.geojson
• san-diego.geojson

[epaulson]

re: the cities that aren't ordered right but display correct - I'm unfortunately not enough of a GIS/D3 expert to know exactly why the old version of d3 (v3) had no problem with these files and why the modern version of d3 (v7) does - I've not been able to find a documented change in d3 that says it's more strictly enforcing the order or some other change that would help us understand what exactly to look for.

[sebkur]

I guess it might have to do with @mbostock's reply further down on the thread you you linked to https://github.com/d3/d3-geo/pull/79#issuecomment-281130284:

Related, it’s not sufficient to say that the exterior ring should be clockwise (or anticlockwise) because D3 uses spherical rather than planar equirectangular coordinates. Better to clarify that exterior ring for polygons smaller than a hemisphere should be clockwise; polygons larger than a hemisphere will use the opposite winding order. In other words both winding orders are valid but have opposite meaning.

I'm also still not quite getting this yet. Spherical polygons, planar polygons, hmm, which one's do we have at hand?

[epaulson]

Right - but what I didn't understand was what changed in D3 between v3 and v7 - wasn't it always using spherical coordinates?