Today's diversity of languages, libraries and virtual machines ought to be a boon for programmers, but instead they fragment our infrastructure.

The resulting integration problems continually limit the re-usability of code and the insight of tools; one major source of problems is the division between "managed" language virtual machines (VMs) and the Unix-like "native" environment. I'll outline my research into an evolutionary way to bridge this chasm, based on extending Unix-like services to become "VM-like". I'll then focus on one specific area of technical work: the libcrunch system for run-time type and memory safety checking. This instruments unsafe C code using techniques reminiscent of safe virtual machines, and provides an efficient yet memory-safe implementation of C (not a dialect). I'll briefly discuss some other in-progress applications of the infrastructure, including a precise whole-process garbage collector and a "wide-spectrum" multi-language programming environment.

Stephen Kell is a Research Associate at the University of Cambridge. His interests include most topics related to programming.

His work's emphasis is on improving the flexibility and expressiveness of infrastructure software, from programming language implementations downwards. He received his PhD from the University of Cambridge in 2012, and has done postdoctoral work in Oxford and Lugano.