Characteristics of an Agent Scripting Language and Its Execution Environment

  • Authors:
  • Danny B. Lange

  • Affiliations:
  • -

  • Venue:
  • COORDINATION '99 Proceedings of the Third International Conference on Coordination Languages and Models
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

Experience gained from IBM's Aglets project has led us to the conclusion that Java is far from being a language ideal for agent development. We discovered significant deficiencies in Java related to agent mobility, messaging, and security. Most importantly, we recognized that the Java programming language and Java-based agent frameworks do not provide a sufficient set of high-level abstraction mechanisms for efficient agent development. At last but not least, we also found the Aglets system put an undesirable emphasis on code mobility which is only required in a certain class of agent-based applications. Consequently, we have taken steps to address these issues in the creation of new agent scripting language and an associated execution environment. This new scripting language is based on the XML Internet standard. That is, the scripting language is optimized for manipulating and producing XML data and it is itself represented in the XML format. Drawing ideas from from popular Tcl and LISP, it is an extensible, persistent, and adaptive scripting language for creating agent-based applications. The execution environment consists of a network of communicating agent servers based on the HTTP and TCP/IP protocols. Agents can safely be hosted in this environment which has been built to support long-term hosting of agents. The environment also provides a graphical agent workbench for developing and testing agents. The execution environment is in its entirety written in Java for optimal portability.