Real-time Systems, in particular regulation systems, are often specified using the notion of running modes. On the other hand, the family of synchronous languages [1] has been very successful in offering formally defined languages and programming environments for safety-critical systems. We are particularly interested in the data-flow language Lustre, well-suited for the description of regulation systems. Recently, we proposed to extend Lustre with a new construct devoted to the description of running modes in regulation systems. This language extension is based upon the mathematical model of mode-automata [13]. We now have a running implementation of this extension. In this paper we comment on a case-study proposed by SAAB, and argue for