2D CircleCast Centroid causing infinite loop.

Questions & Answers
1

Hi every, first time poster here.
I have the following code to make a simple 2D Ball (circle) bounce around some colliders on a scene, without using any RigidBody (Unity’s Physics are a bit of an overkill for what I need).

void FixedUpdate()
{
	Vector2 pos;
	Vector2 prev_pos;
	Vector3 rot;
	
	pos = transform.position;
	speed.y -= gravity;
	
	if(!was_hit)
	{
		float ray_length = speed.magnitude;
		Vector2 dir = speed.normalized;
		Vector2 prev_dir = Vector2.zero;
		RaycastHit2D hit = Physics2D.CircleCast(
			pos,
			radius,
			dir,
			ray_length,
			collision_mask
		);
			
		if(hit.collider != null)
		{
			bool on_going;
			float bounced;
			float perm;
			
			on_going = true;
			bounced = 1;
			
			while(on_going)
			{
				prev_dir = dir;
				prev_pos = pos;
				
				perm = 2.0f * Vector2.Dot(dir, hit.normal);
				dir = dir - (hit.normal * perm);
				
				print (
					(hit.point + (hit.normal * radius))
					+ " :: " +
					hit.centroid
				);
				//pos = (hit.point + (hit.normal * radius));
				pos = hit.centroid;
				
				ray_length *= (1 - hit.fraction) * bounce;
				
				bounced *= bounce;
				
				
				if(ray_length > 0)
				{
					hit = Physics2D.CircleCast(
						pos,
						radius,
						dir,
						ray_length,
						collision_mask
					);
					
					if(hit.collider == null)
						on_going = false;
				}
				else
				 	on_going = false;
			}
			
			
			rotation = Vector2.Angle(prev_dir * -1, dir) * speed.magnitude;
			
			pos = pos + (dir * ray_length);
			speed = dir * (speed.magnitude * bounced);
		}
		else
		{
			pos = pos + speed;
		}
		
		transform.position = new Vector3(pos.x, pos.y, transform.position.z);
	}
	
	was_hit = false;
	
	rot = transform.localEulerAngles;
	rot.z += rotation * Time.deltaTime;
	transform.localEulerAngles = rot;
}

My problem is that the line 46 [pos = hit.centroid] causes this for no apparent reason.
The weird part is that if I uncomment the line 45 [pos = (hit.point + (hit.normal * radius))] above, the code works perfectly.
The even weirdest thing of it all is that the comment above those 2 lines prints the exact same value for both (when the working code runs, since I don’t know how to get prints under infinite loops).

Any ideas why this is happening?
This is not exactly a critical matter, since I already have a solution, but it is one of those things that simply bugs me.
Why wasting calculations, when the thing is already done?

Also, it would make the code simpler because I need to relocate it to make it a function used more than once per frame, and with more complex uses.

2

Ok, this was a silly question because I found the problem.
When I assign the new position to the centroid, I am actually positioning the ball to the first point of contact.
If I cast from that point to anywhere, the first point of contact will be the exact same because it is a point of contact.

The reason why I haven’t detected this before with plain prints is because I did a Vector2 print and not a float print for each of the individual values.
When a Vector2 is printed, the float values are rounded to make them look better
So that tiny difference in the values is never seen.

Maybe this will help someone else, but it is over all a very silly error.