feat!: add normals to `CollisionData` in events

[tqwewe]

Adds normal data from ContactManifoldData::normal to the CollisionData struct.

Benefits

• Normal vector added to collision data

Downsides

• I had to remove Copy from CollisionData due to the normals being in a Vec<Vec2>. We could just provide the first normal Vec2 as an Option<Vec2>, but I'm not sure whats the best option for this.

---

Fixes https://github.com/jcornaz/heron/issues/188

[jcornaz]

I had to remove Copy from CollisionData due to the normals being in a Vec. We could just provide the first normal Vec2 as an Option, but I'm not sure whats the best option for this.

Arf. That's a quite big downside indeed. But the benefits are high too. I'll try to figure out how to move forward (which may simply be: releasing a breaking change).

[jcornaz]

Maybe only provide the first normal as an option as you suggested. At least it doesn't force us to make a breaking change, and it should be enough for most use-cases. It may even make it easier to use actually.

Even if we end up changing our mind and break the API later to introduce a Vec, then we would have at least deferred breaking the API until we really needed it.

What do you think?

[tqwewe]

I think it should cover most cases. If someone creates an issue for it, then it could be easily changed, or maybe just released in the next major/breaking release.

I'll go ahead and update it to Option for now

[tqwewe]

Although, it seems like Vec2 does not implement Eq... so even as Option<Vec2>, it will still be a breaking change regardless.

[jcornaz]

Although, it seems like Vec2 does not implement Eq... so even as Option, it will still be a breaking change regardless.

Arf... I'll think more about it. Don't change anything yet.

[tversteeg]

Although, it seems like Vec2 does not implement Eq... so even as Option<Vec2>, it will still be a breaking change regardless.

Can't you manually implement Eq and ignore the Vec2?

[jcornaz]

Can't you manually implement Eq and ignore the Vec2?

That would be an incorrect equal implementation IMO. If two event datas don't have the same normal, then they are not equal.

[tversteeg]

Can't you manually implement Eq and ignore the Vec2?

That would be an incorrect equal implementation IMO. If two event datas don't have the same normal, then they are not equal.

You could also compare the normals in the eq manually.

[jcornaz]

You could also compare the normals in the eq manually.

Vec2 only implements PartialEq, because in some cases a vector is not equal to itself. So we can only at most implement PartialEq. Implementing Eq is not possible, even manually, because we still have to account for the fact that Vec2 only has a partial equivalence relation.

[tqwewe]

Out of curiosity, why wouldn't Vec2 already implement Eq? In what cases would a vec not equal itself?

[jcornaz]

In what cases would a vec not equal itself?

Vec2::NAN is not equal to itself, because f32::NAN is not equal to itself, because (in math) x/0 is not equal to itself (because it is undefined).

That's the most obvious case, but I think there are other cases when dealing with infinity or subnormal numbers. But I don't remember well, I'd have to check.

[tqwewe]

I see, essentially the same reasons why f32 doesn't implement Eq I guess