Quirky ray cast method

SIsilicon28

While creating the collision system for my game project I realized that I needed concave polygon shapes to act single sided, i.e. the collision normal returned by a RayCast is the same no matter which side it hits. To do this, I decided to use RayCast.get_collision_shape() to get the specific shape that the ray collided with. If the shape is a concave polygon, then I'll find the face that the ray cast's collision_point corresponds with. Else I use the regular ol' get_collision_normal() method. Now here's the problem with get_collision_shape. Instead of returning the shape's index, it gives the index of the triangle of the concave shape. While that saves the time to actually find the triangle, it means that a static body can only have its first shape interacted this way. Is this a bug, or intentional behavior that I'm not aware of?

TwistedTwigleg

Sounds like it could be a bug or there should be another function that returns the shape’s index rather than the triangle index. According to the documentation, get_collider_shape is supposed to return “... the shape ID of the first object that the ray intersects ...”. To me this implies that the function returns an ID for the shape, not an ID for a triangle. Maybe the functionality changed but the documentation wasn’t updated?

I wonder if the using DirectSpaceState gives the same results or not.

It might be worth opening a bug report on GitHub, or maybe a feature request on the proposal library if it is intended functionality.