We investigate an extension of predictive control, PID controllers, and Smith predictors to some classes of linear delay systems which are most encountered in practice. The predictive part is provided by extending flatness-based predictive control. Generalized PI regulators lead to a 'PID and Smith predictor'-like scheme which permits us to easily handle (deadtime-dominant) unstable systems. Numerous examples with thorough simulations demonstrate the performance of the proposed control strategy.