Edge Case
Hello there reader. Today I acquired some more information regarding the weird edge collision bug. Where moving along a straight wall into a diagonal one causes an entity to get stuck. My observations are that this really only seems to happen at these specific edges. Not on other diagonal tiles that have five vertices in their collision shape. It doesn’t happen if you move a little bit away from the wall and then into the diagonal one.

The weird thing is that the first iteration of collision checking correctly calculates the normal, etc. However the other three iterations that follow (4 iterations total) are not actually returning the correct normal. I suspect that the tiles simply don’t line up as I expect. Maybe floating points are the cause for this. Whatever it is though, I’ll have to dive into even more detail and closely inspect what goes wrong.

Last modified: February 3, 2023