Modeling robust computations in a data-flow-like style

In this paper we present a model for describing robust computations and for this purpose we define the outline of a programming language which can be used for exploiting robustness, high parallelism and modularity. Constructs are delined for the management of non-determinism and to allow for history sensitive computations. The computational model which is used resembles a data-flow model where the flow of computation is data driven. Non-determinism is expressed by defining constructs based on Dijkstra's guarded commands, and hislory sensitive computations are modeled by allowing cycles to be present in the computation graph. Computations described by this language are shown to be intrinsically robust, because they satisfy Denning's principles, and either to insertion of consistency checks or the structuring of redundant computations may be used to provide fault detection.