An Automatic Technique for Selection of Data Representations in SETL Programs

Publisher Summary This chapter highlights an automatic technique for selection of data representations in SETL programs. SETL is a very-high-level programming language supporting set-theoretic syntax and semantics. It allows algorithms to be programmed rapidly and succinctly without requiring data-structure declarations to be supplied. The chapter describes a new technique for automatic selection of appropriate data representations during compile time for programs with omitted declarations. It also presents an efficient data representation selection algorithm, whose complexity is comparable with those of the fastest known general data-flow algorithms of Tarjan and Reif. The level of a programming language is determined by the power of its semantic primitives, which influence the ease and speed of programming in the language profoundly. The chapter also describes how the SETL compiler chooses data structure implementations. SETL supports logical quantifiers and the abstract data type set as the prominent high-level features of the language. Programmers are intended to use these features as much as possible and leave more specific data structure and control structure decisions to the SETL compiler. Most other very high-level languages start by incorporating these same features. SETL is a very-high-level programming language supporting set-theoretic syntax and semantics. It allows algorithms to be programmed rapidly and succinctly without requiring data-structure declarations to be supplied. Such declarations can be manually specified later, without recoding the program, to improve the efficiency of program execution. We describe a new technique for automatic selection of appropriate data representations during compile time for programs with omitted declarations and present an efficient data representation selection algorithm, whose complexity is comparable with those of the fastest known general data-flow algorithms of Tarjan and Reif.