Volodymyr Agafonkin

@egtw-rgb not an admin, but I did see and appreciate it. Thank you!

Polygon clipping is a notoriously hard problem to do robustly! I attempted it several times through the years and failed miserably. Perhaps some day we'll get there.

Here's the current performance breakdown:  Query structure search only takes 3.5%, whereas it's the main target of optimization of Seidel's randomized idea in the paper (improving tree balance, building...

Worth noting that even with the current bottlenecks, it's already significantly faster than poly2tri.js. Sure it produces much worse triangulation quality, but for some applications triangulation performance is much more...

Update: 

Performance is much better now. Splitting trapezoids is still a bottleneck though. ``` bash typical OSM building (15 vertices): seidel x 74,740 ops/sec ±0.69% (98 runs sampled) libtess x 22,402...

Good point! Indices instead of coordinates are also needed to support THREE.js use case, so will take a look.

Very cool! Plugged it into the UV mapping notebook and while we'll have to find another fixing scheme, it looks very interesting:  I guess the only issue that could...

For comparison, 6-pole UV mapping looks like this: 

Yeah, I agree. The question is how expensive this will be, and whether we can fine a compromise. E.g. if I find out that we can improve precision with only...