Port JTS #448

[pramsey]

Fix Voronoi polygon generation by adding robust Triangle circumcentreDD

[dbaston]

Adds a ~40% performance penalty in my quick test (bin/perf_voronoi). I wonder if the DD circumcenter can be used only when needed?

[dr-jts]

Adds a ~40% performance penalty in my quick test (bin/perf_voronoi). I wonder if the DD circumcenter can be used only when needed?

A few possibilities come to mind:

• I think the robust circumcentre code is only need for Delaunay triangles which are right isosceles triangles. So perhaps a filter can be created to detect this condition (up to a tolerance) and fall back to the double-precision calc if not.
• alternatively, detect when the circumcentre lies "close" to a triangle edge and if so recompute with higher precision
• perhaps some inspection/experimentation with the circumcentre calculation would reveal a way of creating a filter based on intermediate values during the computation (e.g. if the denom is very large or the numx and numy are very close)
• maybe very close Delaunay triangle circumcentres could be detected and snapped together. This could perhaps be done by processing each triangle after circumcentre calculation, and comparing centres of each adjacent triangle. Alternatively, if close circumcentres are found then recalculate using high precision.

So lots of options to try!

[pramsey]

@dbaston see if the little pre-filter condition I've added takes some of the pain away

[dbaston]

@dbaston see if the little pre-filter condition I've added takes some of the pain away

Yep, performance looks the same as master now. Thanks!

[pramsey]

Pushed to master

[dr-jts]

@pramsey It looks like this heuristic doesn't work (mea culpa, since it was my suggestion). We should remove it, I think.

[dr-jts]

The tolerance here should be much smaller for this test to actually work. Say 0.1, or even 0.