%T Dynamicity through Occam and TDS %A D. Millot, J. Vautherin %E J. Wexler %B OUG\-11: Developing Transputer Applications %X A parallel program running on a parallel machine involves a logical network of processes and a physical network of processors. When both networks are known at design time, a static mapping of the logical network on the physical one can take place. Dynamicity arises when one of the networks is not entirely determined before execution starts. This can happen for instance when the logical network involves dynamic creation of processes, or when the physical topology cannot be defined at design time (this is the case in a multi\-user context, as the configuration a user gets is affected by the other users\[rs] computations, or when the programmer wants to develop a generic program that can be executed on any physical topology).When using Occam to design a network of processes that has to be mapped on transputers, things have to be decided before execution starts :\- the network of processes should be composed of an explicitly bounded number of processes due to the fact that, unlike CSP, Occam does not allow recursion,\- the topology of the transputer network should be known, for processes are explicitly mapped onto processors. One has to investigate the physical topology in order to write the configuration statements, and a change in the topology entails modifications in these statements. An application is therefore dedicated to a configuration. It would nevertheless be very attractive to design software that could run on an unidentified topology, trying to make the best of available processors. Such software has to realize a dynamic placement and therefore includes a phase supposed to investigate the configuration, load the different codes on the appropriate processors, then dynamically start their execution.Our aim is to develop suitable tools to write that kind of application. As a first step, we investigate the potentiality of Occam and TDS in this field.