daid

The newer version of Arcus is here in this repo, but if you installed an older version in your python installation, then the newer version does not properly override this.

Ok, I think I see why that fix fixes it. I'm using `std::sort`: https://en.cppreference.com/w/cpp/algorithm/sort Which specifies that the elements need to have the `Compare` requirement: https://en.cppreference.com/w/cpp/named_req/Compare Which in turn, says...

Ah, that's a good pointer. My suspicion is then that there is an issue with this function: https://github.com/daid/SeriousProton/blob/master/src/io/network/address.cpp#L87 Can you see if it works if you make that function empty?...

In debug builds, esc is a quick&hard exit. But you can use the home key instead

Problem isn't concurrany. Problem is that newer versions of "gcov" seem to output a "grand total" at the end, and cpputest is taking a LOT of assumptions on the output...

Extra note, if you just ignore the "make gcov", but still enable "CPPUTEST_USE_GCOV" in the makefile. The tool "gcovr" can parse the generated files just file. In my case, from...

I think this can be fixed easier then normal fixing, but just changing the shader to assume that the normal could be flipped as well.

@Ghostkeeper That's how the normals are calculated right now. The problem is some applications just don't keep the winding order properly at all (as with the above model) OpenSCAD also...

FYI: Models with just all the normals 100% inverted are rare. Models with some of the normals correct, but some wrong are more common.

``` difference() { cube(20, center=true); cube(10, center=true); } ``` OpenSCAD code. Result is a cube with a hollow center. But the center has an improper winding order. Good luck ;-)