Based on the DIS result, the HLA framework has been defined to achieve interoperability of independent simulators. Concurrently, and for the most part, independent of that effort, the parallel and distributed simulation community has attempted to define synchronization protocols for the correct execution of parallel simulation as-fast-as possible. Building parallel time-constrained federates within an HLA framework is not an easy task. We identify the potential difficulties: one or several federates, when and how to advance the federate's time, how to handle RTI notifications, etc., and present our experiences with adding HLA features into the parsec parallel simulation language.