Union of triangles is not always equal to the convex hull when using VoronoiDelaunay?

[nako95]

I'm using this package on Julia 0.5.2 and I sometimes get that the union of the triangles in the triangulation does not equal the convex hull of the points, which is a property of Delaunay triangulation. I include an easy example to show this:

using Plots, VoronoiDelaunay

min_coord=VoronoiDelaunay.min_coord
max_coord=VoronoiDelaunay.max_coord  

function pointsRescaledCoords(x,y) #To be used for tessellation
    Point2D[Point((max_coord-min_coord)/(maximum(x)-minimum(x))*(x[i]-maximum(x))+max_coord,
     (max_coord-min_coord)/(maximum(y)-minimum(y))*(y[i]-maximum(y))+max_coord) for i in 1:length(x)]
end

#First shape
X = [0.0; 0.0; 1.0; 0.6; 1.0]
Y = [0.0; 1.0; 1.0; 0.8; 0.0]
display(plot(Shape(X,Y), opacity=.5,xlims = (-0.05,1.05),ylims=(-0.05,1.05)))

tess = DelaunayTessellation()
push!(tess,pointsRescaledCoords(X,Y))

#Display delaunayedges
x, y = getplotxy(delaunayedges(tess))
display(plot(x-min_coord,y-min_coord,xlims = (-0.05,1.05),ylims=(-0.05,1.05)))

#Second shape
X = [0.0; 0.0; 1.0; 0.6; 1.0]
Y = [0.0; 1.0; 1.0; 0.79; 0.0] #Fourth element changed from 0.8 to 0.79
display(plot(Shape(X,Y), opacity=.5,xlims = (-0.05,1.05),ylims=(-0.05,1.05)))

tess = DelaunayTessellation()
push!(tess,pointsRescaledCoords(X,Y))

x, y = getplotxy(delaunayedges(tess))
plot(x-min_coord,y-min_coord,xlims = (-0.05,1.05),ylims=(-0.05,1.05)) 

When using the first shape the union of the triangles in the triangulation does not equal the convex hull of the points but when using the second shape they do. The only difference between the shapes is that I have changed the fourth element in Y from 0.8 to 0.79 in the second shape.

Am I doing something wrong? It seems to me that somethings strange is happening when I do the triangulation on the first shape and I don't understand why.

[skariel]

see issue #16 , this is known behavior. The solution will be to use 3 points instead of 4 as the initial embedding.