Barend Gehrels

Well done! I can confirm that it is faster now, sometimes much faster, and that the `correct` is fixed for the random "junk" polygons ``` pentagram areas 33.5 d 25.6158...

I'm looking at the code. It might be possible to change `dissolve_find_intersections` into just using `detail::self_get_turn_points::self_turns`. You will then get all turns back, and you can iterate through them to...

> The odd even approach can have the same performance as the other approach. But for this, the difference should be fixed. Yes I noticed the new issue. Will look...

Thanks, but I don't think it will fit as such in the library. As you say - the difference of A - B (both being box) can be a polygon....

Good points. Do you plan to implement intersection (not yet mentioned) and union_ (you mention that - but via difference)? Could we maybe add it first to extensions, instead of...

I'm OK including them, but with some notes. Not including them was done on purpose. Boost.Geometry is a concept based library. It supports models fullfilling the rules of the concepts....

I glanced through the sources and sometimes (not too often) we include them already. For example in the new discrete_frechet_distance.hpp. I think that should change (it looks not necessary anyway)....

**Point 1:** About `using pol2d = boost::geometry::model::polygon ;` Because you specify the polygon type as clockwise (first `true`), the right side is the interior and the left side is the...

> Excellent. So two additional test cases to make sure regression does not occur? Indeed. The related PR #889 can be merged first. Then (or actually I'm starting already locally)...

> Take your time. Next week i am away on summer break. I wont be able to test for a while. Thanks! It seems to be a one line fix...