Future wireless networks will witness ubiquitous human-machine interactions, where extended reality (XR) is expected to be a key scenario in next-generation mobile systems. In this article, we examine the integrated sensing and communication (ISAC) framework in XR, where a reconfigurable intelligent surface (RIS) may assist user (UE) positioning and communication. Specifically, a practical positioning algorithm based on multiple signal classification (MUSIC) with the aid of specially designed RIS configurations is conceived. Furthermore, we formulate the joint optimization of the UE beamformer and RIS phase shifter to maximize the channel capacity under Cramer-Rao lower bound (CRLB) constraints, which is solved by alternating optimization with gradient projection and manifold optimization. Finally, we use simulation results to demonstrate the feasibility of the conceived positioning algorithm and corroborate the effectiveness of the proposed optimization approach.