In this article, joint communication and sensing (JCAS) is combined with reconfigurable intelligent surfaces (RISs), which validates a novel RIS-based JCAS (RIS-JCAS) system. The proposed system performs monostatic sensing to detect a potential user, after which communication with the user is established. The corresponding RIS-cascaded channel is formulated, based on which an algorithm is created, to handle radar target detection and communication channel estimation while multiple RISs operate simultaneously. A joint beam training process by utilizing the identical beamforming pattern on multiple RISs is created, which increases the time efficiency significantly. The detection accuracy is subsequently further improved through a novel fine-tuning method. To practically validate the aforementioned algorithm, a RIS-JCAS system is demonstrated using a mmWave testbed with a bandwidth of 1 GHz centralized at 26.2 GHz. Besides combining the widely adopted orthogonal frequency-division multiplexing (OFDM) scheme with high-directional RIS beams, the system also allows the use of variant RIS flat-top beam shaping combined with orthogonal chirp-division multiplexing radar (OCDM) waveforms. This idea keeps the sensitivity of the RIS-JCAS system while broader beams with wider coverage but lower gain are utilized. Overall, accurate 4-dimensional sensing results are derived, which strongly assists the communication channel estimation and realizes the core benefits of the proposed RIS-JCAS system.