Deep Type Inference for Mobile Functions
Stephen T. Gilmore · Edinburgh Research Explorer (University of Edinburgh) · 1999
. We consider the problem of assessing the trustworthiness of mobile code. We introduce the idea of deep type inference on compiled object code and explain its usefulness as a method of deciding the level of security management which a unit of mobile code will require. 1. INTRODUCTION The mobile agent paradigm is emerging as a leading programming paradigm for the next generation of networked computing architectures. A mobile agent can be deployed for evaluation of a computation on a remote host. This remote evaluation can both improve data locality and make economical use of available network resources. However, the only reasonable security policy for a computational resource provider to adopt is one which considers all computations of non-local origin to be potentially hostile. Thus programming languages such as Java 1 [1] where the notion of mobility is native to the language enforce sandboxing of mobile code. This prevents anti-social behaviour such as the creation and deletion...