By William Fellows
Sun Microsystems Inc’s real-time Java aims got a shot in the arm last Thursday when NCITS, the National Committee for Information Technology Standardization, voted not to sponsor the Real Time Java Working Group and its proposals for extending Java for real- time applications. The decision lessens the likelihood that developers will eventually have to choose between two competing real-time Java specifications or target both. The future of RTJWG, the so-called Java ‘splinter group’ was being discussed in an atmosphere of crisis as we went to press. Some insiders expect it to eventually morph into something other than a standards initiative. In front of the meeting was a motion to proceed without using Sun’s intellectual property, though some members were arguing that it would not be different enough from what NCITS voted down. RTJWG, whose 35 members include Hewlett-Packard Co and Microsoft Corp, argue that Sun has too much control of how Java standards are set and wants the process to be independent of Java’s owner. Sun COO Ed Zander said last week that HP’s aim with RTJWG is to upset the apple cart, and that its effort is having a disruptive effect on Java. NCITS’ decision will slow down RTJWG’s progress and should give Sun and its own real-time Java partners, notably IBM Corp, time to get their own real-time Java expert group up and running. The IBM and Sun-led expert group was formed as part of Sun’s new community licensing and development model for Java. IBM was particularly enthusiastic about the opportunity, however genuine concerns appear to remain among the real-time community who worry that Sun, for which the real-time market is not as important as it is to others, may eventually deliver a watered down real-time Java specification. Moreover, the real-time Java community as a whole still has to address the concerns of the US Department of Defense whose standards organization hosted a meeting of RTJWG and Sun supporters last Monday to express its concerns. The DoD, for which real-time Java will be important in its future IT development, cannot support two real-time Javas, nor a specification that is too weak for its requirements. RTWG applied to NCITS in December which duly formed a technical committee for real-time computer systems, called R1. No votes from IBM, Sun, and Sybase led to a second ballot of members and the RTWG proposal fell short of the two-thirds required to pass according to NCITS procedures. Only seven of RTJWG’s claimed 35 members (25 are said to have made their membership public) voted in favor. Both Sun and RTJWG use as their starting point documents created by the National Institute of Standards and Technology’s Requirements Group for Real-time Extensions for Java. It met in San Diego last week but is said by insiders to be proceeding much more slowly than it should.
Sun blesses NCITS’ new Jini-esque work
RTJWG set out on a different path from Sun last July a few months after HP set the ball rolling by offering a ‘clean room’ embedded Java machine called Chai not dependent on Sun’s intellectual property (some published reports suggest it is based upon Microsoft’s Java virtual machine). HP and others believe that Sun is unfairly prohibiting non-licensees from participating in the development of Java specifications. HP is dissatisfied with the pace at which Sun is advancing plans to enhance non-core Java APIs, saying it is holding up development of the market for embedded Java technologies. RTJWG said it remains committed to its original objectives and at a meeting last Monday agreed upon a better draft of its ‘scarecrow’ specification for real-time Java extensions. The R1 group was meeting as we went to press to decide upon a course of action. What has been agreed is that a committee led by Mitre Corp’s Douglas Jensen will meet in Hawaii on February 23, 24 and 25 to discuss a new proposal for R1, this time focused on distributed real-time Java, which includes technologies such Sun’s Jini and JavaSpaces, but not limited to them. Mitre for example is already testing other kinds of mechanisms. Because R1 will produce a technical document and not a proposal for a standard it Sun has already given the project its support.
