Applications ofConcrete-Type Inference

Precise information about concrete types has numerous applications including increased program understanding, application extracting, and various optimizations. We present a concrete-type inference tool for Smalltalk in a practical setting, demonstrating that it can reduce the runtime memory requirements significantly. Specifically, we have implemented our system in the Resilient Embedded System, a development platform for embedded systems. We also demonstrate how our tool can be used to catch potential runtime errors at compile time. Our type inference system is based on the Cartesian Product Algorithm [1]. We discuss sources of imprecision in existing algorithms and present a new technique to reduce imprecision from default initialization of instance variables. Furthermore, we discuss the applications of concrete-type inference and how they can help programmers rapidly develop faster, more stable programs with smaller memory footprint. Especially the latter property is significant when building programs for embedded systems.