Filtering SQL via the principles of pandemonium

Abstract

In this paper, we re-introduce Pandemonium, an early specification for parallel processing through semiautonomous agents. The biggest advantage of the Pandemonium approach is its simplicity, which is achieved by dividing tasks among many computational units. Using Pandemonium as a metaphor, we design an interpreter for general-purpose filtering of text. To demonstrate the applicability of our design, we show how to partially parse and correct a sequence of badly ordered SQL commands. This example is not artificial; the badly ordered commands were generated by a commercial CASE tool for database development and the command set was too large for manual correction. A Pandemonium-style approach has many advantages over using full-blown parser generators and rule-based systems for such tasks.