Continuous Time. All employees who have a break of less than 60 minutes either before or after their regularly sched- uled or assigned work will be paid at continuous time without a guarantee of a minimum two (2) hours until the work is completed.
Continuous Time. This section sets out a problem which it has become clear needs attention; we are not yet in a position to say exactly how the project will resolve this problem but we are well aware of [Kop97] which offers a valuable reference to technology in this field. Several of the systems of interest within the project interact with the physical world to try to control variables which vary continuously. The speed of a car being influenced by a “cruise control” system (cf. Chapter 6) fits this description. Of course, the cruise control system does not directly influence the speed. The simplest system would require sensors and actuators linked to physical parts of the car. In fact, in a modern car, a cruise control system interfaces to other quite complex systems to bring about the changes required in the components that actually cause acceleration and deceleration. In order to design a discrete control system, there has to be a bridge between digital signals and the continuously varying phenomena. One ap- proach is to assume from the outset some sampling frequency and thus to describe the problem in discrete terms from the first specification. There is a school of thought that says this should not be the only available approach. Indeed there are reasons to doubt the wisdom of a premature (and pos- sibly arbitrary) move to discrete time. It might be that determining the sampling frequency is a design decision that should be made during design (not necessarily at the outset). The sampling frequency will obviously de- pend on the rate of change of various quantities. In fact, it is likely that 3There is a wider field of investigation regarding how HJJ could fit with current PFs work further being only an optional tool. There is a phase of requirements elicitation, prior to the definition of the system boundaries, that could take advantage of the (mental) structure imposed by the approach. Although this is not the mainstream of the HJJ research, we are confident that some progress will be possible later in the project. one will need to record assumptions (as rely conditions) about the maximum rates of change at different timing granularities (cf. TimeBands [BH09]).4 If we do proceed to include specifications of continuously varying phe- nomena, we shall clearly require a coherent development method to move to the Event-B notation.5 There are also obvious tool requirements to be considered.
