Open ChrisJanssens opened 1 year ago
in practice you probably still want to apply a gravity toward the "ground" ? So I'm not sure this matters much ; but an epsilon isn't needed I think indeed.
While I think this makes sense, more context about this code can be found here:
The commit message is explicitly about forbidding exactly horizontal movement, I'm not sure wheteher it's fully thought out, or was done only to follow the epsilon "mindset" which was already here before. Tagging @janhohenheim if you're around and able clarify ? 🥺
@Vrixyz I remember doing that, but I don't remember my reasoning. Tending towards following the epsilons that were already present. If it's inconvenient, feel free to revert it :)
I still agree with this change, but this only adds a fix when snapping is enabled, so we'll need to gather more information about it before validating.
This PR https://github.com/dimforge/rapier/pull/715 explores alternatives (or additions)
It looks like we are taking the dot product of the movement and up direction here to ensure the vector is negative to the up direction. Is this necessary? I was using grounded to determine when my character controller can "jump". Allowing this value to be zero or lower allows my character controller lateral movement along a collider that is on the opposed side of my character to that of the up direction.