Point in polygon check failing

[asinghvi17]

For a polygon with multiple holes, the point in polygon check seems to be failing.

import GeometryOps as GO, GeoInterface as GI
using GeometryBasics

rings = [[Point2f.(r*cos(θ), r*sin(θ)) for θ in LinRange(0, 2π, 90)] for r in 5:-1:1]
p = GI.Polygon(rings)
poly(p)

xmin, ymin = minimum(ax.finallimits[])
xmax, ymax = maximum(ax.finallimits[])
points = [Point2f(x, y) for x in LinRange(xmin, xmax, 500), y in LinRange(ymin, ymax, 500)]
f, a, p = heatmap(xmin..xmax, ymin..ymax, GeometryOps.within.(points, (p,)))
poly!(a, p; alpha = 0.6)
f

[rafaqz]

This isn't just "mutiple holes" its holes inside polygons inside holes! I don't know if anyone handles this?

[asinghvi17]

It's an example in the Hao-Sun paper

[asinghvi17]

Also earcut and Cairo, since Makie can display it

[skygering]

Is this a valid polygon though? Everything post the first ring is supposed to be a hole. So really this should be "validated" and split into three polygons, the first two each having holes and the last being a circle... Thoughts @asinghvi17? Do we still want it to work for this or should the polygon just be validated (once we have that functionality 😢 ).

[asinghvi17]

Hmm that's a good point. By the even odd rule it should look like the first plot...let me check what the databases define this as first!

[asinghvi17]

So to go back full circle (haha) it looks like this is an invalid polygon by a conservative reading of GIS standards. We could warn or error if necessary, but we should have a fix that lifts these up to become multipolygons!