Barend Gehrels

Okay, thanks for testing! I tested with geographic / andoyer. I'll look further next Wednesday. I'm about to publish a PR anyway. If I still have that problem, with that...

Okay, apparently it happens, for this specific case, with `thomas` when using `long double` in the strategy calculations. I can reproduce it with this program (also adding some decimals to...

This is the output, for me (it might be environment dependent): ``` N5boost8geometry8strategy12intersection19geographic_segmentsINS1_6thomasELm2ENS0_3srs8spheroidIeEEeEE N5boost8geometry8strategy12intersection19geographic_segmentsINS1_6thomasELm1ENS0_3srs8spheroidIeEEeEE p: 147.588 q: 0.13477 int: 0 how: d POINT(2.10235e-317 6.9528e-310) N5boost8geometry8strategy12intersection19geographic_segmentsINS1_6thomasELm2ENS0_3srs8spheroidIeEEeEE p: 147.588 q: 0.13477 int:...

The `thomas` was defined in the unit tests (by myself but long ago). I made it flexible now. Indeed, sometimes a turn is missed with `thomas`, sometimes with `andoyer`, both...

https://github.com/boostorg/geometry/pull/1058

> I'm building a spatial extension for sqlite, > in the same vein than spatialite > but with Boost.Geometry instead of GEOS. Perfect!!! That is good news.

Interesting, thanks. I can reproduce it. It's related to internal calculations doing this, and there are many because you get an intersection at every node. Due to FP precision the...

> @barendgehrels are you OK with this PR? See my review remark, basically yes, but can you look at my comments I made earlier?

I can reproduce it. Output (in slightly modified test program): ``` 70: issue_1231 : MULTIPOLYGON(((3.33333 8.33333,5 5,5.31429 4.37143,-1.21132 -7.95472,-3.40491 -7.48466,0 5,3.33333 8.33333),(5 5,4 6,0 1,5 5))) 70: .../boost/libs/geometry/test/algorithms/set_operations/intersection/./test_intersection.hpp:120: error: in...

It's difference is also wrong, both clockwise and counterclockwise. I'm making a PR for the testcase (not for the fix)