As reported on [Discussions|https://discussions.unity.com/t/how-to-use-new-vision-inputs-from-beta-0-5-0/309343/23], there is an issue with gaze/pinch input in virtual reality (CompositorServices, in Apple parlance), where the second pinch is not reported as ended until the first pinch has ended. So for example, the following sequence of events occurs:
# Pinch index finger and thumb on left hand
# Observe `Began` and `Moved` events from SpatialPointerDevice pointer #0
# Pinch index finger and thumb on right hand
# Observe `Began` and  `Moved` events from pointer #1, and continued `Moved` events from pointer #0
# Release pinch on right hand
# Observe `Moved` events from pointer #0, but no expected `Ended` event from pointer #1
# Release pinch on left hand
# Observe `Ended` events from both pointers

The expected behavior is that the `Ended` event will be sent at step #6, but it is not. This issue can be replicated in a Swift app based on the Xcode project template for visionOS metal rendering. The issue has been reported to Apple with the feedback ID `FB13351130`, and needs to be resolved on their end. I'm opening this issue so that it is visible to users who may encounter the same problem, and we can use it to track its status. When Apple releases a version of visionOS that fixes the issue, we'll close it out.