Customizing JavaScript Virtual Machines for Specific Applications and Execution Environments

Abstract

Recently IoT (Internet of Things) technology, which gathers various data from sensors and controls distributed devices among a network, is getting more and more attention. The eJS (embedded JavaScript) project aims to make it possible to write IoT programs in JavaScript and reduce difficulties and complexities in the development of IoT applications. The eJS project offers a framework named eJSTK to generate JavaScript virtual machines (VMs) called eJSVMs, each of which is customized to a specific IoT application executed on a specific IoT device equipped with limited resources. This paper presents two new customization capabilities of eJSVM. One is to choose internal data structures of eJSVM between those for 32-bit environment and those for 64-bit environment. The other is to choose a suitable garbage collection algorithm used in a generated VM among four different algorithms. Experimental results showed the effectiveness of the new customization capabilities.