The Object Management Group has picked a shortlist of five technologies that it will consider for use as its Object Request Broker, the mechanism that enables objects to make and receive requests and responses transparently in an object-orientated environment. The five finalists are Cambridge, UK-based Architecture Projects Management Ltd, Digital Equipment Corp, Hewlett-Packard Co/Sun Microsystems Inc, the Data General spin-off HyperDesk Corp and NCR Corp/Object Design Inc. Dropped from the original seven contenders are Groupe Bull SA and Distributed Software Engineering Tools Inc. From these five a single technology will be recommended for selection at an Object Management Group meeting in June. The votes show that the joint Hewlett-Packard-Sun submission was most strongly supported with 17 cast in favour, followed by HyperDesk and NCR-Object Design both with 15, DEC with 13 and Architecture Projects with 12. Bull and Distributed Software were eliminated with 10 and two votes respectively. Apart from the contenders themselves, the voters were Data Access, Intel, AT&T, Informix, Visix, Objectivity, Ellemetel, Tivoli Systems, Mentor Graphics, Genesis Development, Philips, Microsoft, Netwise, Constellation, Aldus and Unify, all of whom had a yes/no vote available to cast on each submission. However not is all a bed of roses at the Framingham, Massachusetts-based Group. Several of the participating companies subsequently voiced loud reservations about the nature of the process, and indeed the Object Request Broker itself. Microsoft for one said a fair evaluation of the submissions is impossible [because] the submissions do not appear to be addressing the same set of issues in a uniform manner. The definition of object is unclear. The scope of the Broker is unclear. Without this common base, an apples-and-oranges comparison becomes the only viable possibility; certainly evaluation on technical merit is too difficult. Microsoft believes part of the problem is the weak definition of objects and services for the Object Broker, and that if the Group is really seeking to define a standard that guarantees interoperability, then some common notion of object model is required. Distributed Software, though it sounds like sour grapes, supports Microsoft’s position, saying that the selection process may be moving too fast in light of the fact that amongst Object Management Group members there is still no consensus to what the object model should look like, or indeed on the scope of the Broker itself, given that most of the submissions go well beyond the Brokers scope given any definition. Another non-submitter shares Microsoft’s worries but reckons that the Group already knows the problems – but questions whether it has the guts to do anything about it. We may know we have a train running full steam in the wrong direction, some may even have ideas what to do about it. But, I have yet to see the commitment necessary, by the number of people necessary to put the train on the right track.
