Generating a valid loop after splitting up a polygon at the antimeridian

[boulder-pika]

Greetings -

I am very much new to this library, but would like to use it to break up polygons that span the antimeridian. To that effect, I've reviewed a related post at #439.

But I have a question about how to turn the outputted edges into valid geometry. Start with the test code at

s2geometry/src/s2/s2boolean_operation_test.cc

Line 1757 in b99a794

"# # 90:0, 0:180, 0:-120, 0:0, 0:120, 0:180");

:

// The northern hemisphere.
TestMeridianSplitting("# # 0:0, 0:120, 0:-120",
"# # 90:0, 0:180, 0:-120, 0:0, 0:120, 0:180");

One can see the resulting path is: 90:0, 0:180, 0:-120, 0:0, 0:120, 0:180. But if I wanted to create an S2Loop with this path, it would be invalid per the comment at

s2geometry/src/s2/s2loop.h

Line 72 in b99a794

// Loops are not allowed to have any duplicate vertices (whether adjacent or

:

// Loops are not allowed to have any duplicate vertices (whether adjacent or not).

The above path visits 0:180 twice.

And per the comment at https://github.com/google/s2geometry/blob/b99a7941052c1347ed84a508669d3155c2963b20/src/s2/s2loop.h#L65C1-L67C9:

// An S2Loop represents a simple spherical polygon. It consists of a single
// chain of vertices where the first vertex is implicitly connected to the
// last.

I understand that there is an implicit 90:0 at the end of the path to close the loop. Hence, the 0:180 should not be interpreted as a closing vertex which may justify its appearance twice.

The S2 library appears to be outputting an invalid path when splitting a polygon, preventing me from using it in external geojson applications. Can somebody help me overcome this conceptual problem? Is there a way to generate a valid loop from this path? Any code / examples to point me at would be beneficial. As stated, I'm very new to this library and am wagering there is some piece I am missing.

[jmr]

I think you need to use S2LaxPolygonShape as mentioned in #439 (comment).