Trust and Dependence Analysis

Peter Ørbæk

July 1997

Abstract:

The two pillars of trust analysis and dependence algebra form the foundation of this thesis. Trust analysis is a static analysis of the run-time trustworthiness of data. Dependence algebra is a rich abstract model of data dependences in programming languages, applicable to several kinds of analyses.

We have developed trust analyses for imperative languages with pointers as well as for higher order functional languages, utilizing both abstract interpretation, constraint generation and type inference.

The mathematical theory of dependence algebra has been developed and investigated. Two applications of dependence algebra have been given: a practical implementation of a trust analysis for the C programming language, and a soft type inference system for action semantic specifications. Soundness results for the two analyses have been established

Available as PostScript, PDF.


[BRICS symbol] BRICS WWW home page