Programs that process streams of information are commonly built by assembling reusable information-flow components. In some systems the components must be chosen from a pre-defined set of primitives; in others the programmer can create new custom components using a general-purpose programming language. Neither approach is ideal: restricting programmers to a set of primitive components limits the expressivity of the system, while allowing programmers to define new components in a general-purpose language makes it difficult or impossible to reason about the composite system. We advocate defining information-flow components in a domain-specific language (DSL) that enables us to infer the properties of the components and of the composed system; this provides us with a good compromise between analysability and expressivity. This paper presents DirectFlow, which comprises a DSL, a compiler and a runtime system. The language allows programmers to define objects that implement information-flow components without specifying how messages are sent and received. The compiler generates Java classes by inferring the message sends and methods, while the run-time library constructs information-flow networks by composition of DSL-defined components with standard components. © Springer-Verlag Berlin Heidelberg 2007.
CITATION STYLE
Lin, C. K., & Black, A. P. (2007). DirectFlow: A domain-specific language for information-flow systems. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4609 LNCS, pp. 299–322). Springer Verlag. https://doi.org/10.1007/978-3-540-73589-2_15
Mendeley helps you to discover research relevant for your work.