https://github.com/immersive-web/webxr/blob/master/input-explainer.md#appendix-a-proposed-partial-idl

XRInputSource has been updated to expose it's target and grip poses as two spaces which can be queried with XRFrame.getPose rather than querying them through the input-specific XRFrame.getInputPose. Chrome's implementation of XRInputSource should be updated to match the latest changes, and XRFrame.getInputPose should be removed.