Anomaly detection in dynamic programming languages through heuristics based type inference

Abstract

The growth of agile methods has helped to popularize dynamic programming languages. Their concise syntax and the flexible type handling are well suited for the needs of agile development. Flexible type handling is advantageous in the application of polymorphism, a key ingredient of agile methods. Coincidentally, flexible type declarations have resulted in the loss of a significant tool for assuring software quality. Static typing has long been considered a key tool for identifying anomalies. Proponents of agile methods have come to depend upon a combination of static analysis and software testing to assure code quality. Existing static analysis tools have given little attention to type information. We have identified an approach that combines type inference based heuristics with existing static analysis tools in the identification of potential anomalies. This approach provides the programmer with an assessment of the likelihood that the code contains an anomaly. The goal of this approach is to help improve the quality of the resulting code while retaining the advantages already observed with dynamic programming languages. An initial evaluation of this approach resulted in potential type related anomalies being correctly identified 86% of the time while existing approaches were correct only 36 to 64% of the time.