Revised selected papers of the 22nd International Symposium on Implementation and Application of Functional Languages (IFL 2010), Alphen aan den Rijn, Netherlands, volume 6647 of Lecture Notes in Computer Science. Springer, 2011.Online version
Relational database management systems can be used as a coprocessor for general-purpose programming languages, especially for those program fragments that carry out data-intensive and data-parallel computations. In this paper we present a Haskell library for database-supported program execution. Data-intensive and data-parallel computations are expressed using familiar combinators from the standard list prelude and are entirely executed on the database coprocessor. Programming with the expressive list comprehension notation is also supported. The library, in addition to queries of basic types, supports computations over arbitrarily nested tuples and lists. The implementation avoids unnecessary data transfer and context switching between the database coprocessor and the programming language runtime by ensuring that the number of generated relational queries is only determined by the program fragment’s type and not by the database size.
Also available: Slides