Unifying Multiple Layer Activation Mechanisms Using One Event Sequence

Abstract

Different context-oriented programming languages try to capture contexts with respect to different things, including a computation, an object, and a device that executes a program, by providing different layer activation mechanisms. When we want to exploit all of those different kinds of contexts at the same time, it is not clear how the effects of those contexts should be combined. We develop LamFJ, a calculus for expressing various layer activation mechanisms. It replaces the with and without expressions in ContextFJ with four expressions that fire context change events, which models changes of each context. LamFJ is not only powerful enough to express multiple layer activation mechanisms but also clearly defines combined effects of those mechanisms. In addition to the supported layer activation mechanisms in the paper, namely imperative activation, per-object activation and dynamic scoping, we aim at supporting other mechanisms like reactive and structural activation with small extensions.