CAREER: Clamp - Language Support for C-Level Abstraction, Modularity, and Portability

University Of Washington, Seattle WA

Investigators

Abstract

CCF-0447697 Dan Grossman University of Washington CAREER: Clamp: Language Support for C-Level Abstraction, Modularity, and Portability This research involves the design, implementation, and evaluation of new programming-language technologies (in a new language called "Clamp") for modular and portable low-level software. The goal is to provide a language for programmers that allows for machine-specific assumptions where necessary and a powerful system for ensuring such assumptions are never made implicitly. Doing so should make it easier to build robust software infrastructure and detect defects early in the software-development process. Automated support for checking low-level invariants should also make it easier to teach reliable-software idioms to students and professionals. Key technical difficulties in providing a safe-but-convenient low-level language are support for alias analysis (determining when program expressions refer to the same memory) and compile-time arithmetic (reasoning about the possible results of a numeric computation). Therefore, the Clamp infrastructure is intended to make it easy to use different approaches to these problems with minimal change to the Clamp implementation.