[ News | IPCA | Mirrors | Add | Search | Mail | Help | WoTUG ]
23-24th September 1996
at University of Kent at Canterbury
Organised by World occam and Transputer User Group (WoTUG)
This document transcribes a series of emails in May, 1996, which account for my learning experience with Java threads. The motivation for this arose from the WoTUG-19 conference at Nottingham-Trent University, which included a pre-conference weekend for the WoTUG committee that was convened to examine technical relationships between the Java and occam programming languages. Java had drawn itself to our attention as it had drawn itself to everyone's attention. However, what was of special interest to us was that a major part of its hype echoed many of the key messages that the occam community had been trying to get across (without, exactly, total success) for the past decade. For example, the need for security to be built in to the language (motivated in the case of occam by the requirements for safety-critical embedded applications, and in the case of Java by the danger of pulling executables straight off the web). Another aspect that struck a chord was the binding of a model for concurrency directly into the language (which gives much greater awareness to the system designer of what is actually being designed, and information to the compiler that can be used to prevent a wide class of error and to optimise code generation). The particular problem exercised in these emails is the construction of robust and efficient glue objects (channels/buffers) for connecting active processes (i.e. Java objects containing one or more threads). It proved surprisingly hard to get this right.