At this week’s Xhibition in San Jose, California, the Common Open Software Environment firms – Hewlett-Packard Co, IBM Corp, Santa Cruz Operation Inc, Sun Microsystems Inc, Univel Inc and Unix System Laboratories Inc (and probably Digital Equipment Corp as well by then) – will present COSE’s raison d’etre to the world in the form of […]
At this week’s Xhibition in San Jose, California, the Common Open Software Environment firms – Hewlett-Packard Co, IBM Corp, Santa Cruz Operation Inc, Sun Microsystems Inc, Univel Inc and Unix System Laboratories Inc (and probably Digital Equipment Corp as well by then) – will present COSE’s raison d’etre to the world in the form of a white paper prepared by Unix Labs’s Andy Nagle. It describes – at great length but with little substance – how the companies will attempt to accelerate the open systems process. In a nutshell, the white paper proposes that vendors get together and agree on common specifications and behaviour, application programming interfaces and test suites for various technologies before taking them forward as recommendations to industry standards bodies like X/Open Co Ltd and the Object Management Group for ratification, branding and certification. The paper describes a new layer of informal processes that will be used to achieve these goals. Squabbling
The aim is to cut out the time-consuming lobbying – or squabbling – that currently takes place within the standards bodies and do the shouting at private gatherings of vendors. Nagle, unfortunate fall guy for the white paper, says it will ensure that debate in the standards bodies will be about the specifications themselves, rather than inter-company feuding over stuff like feature sets. This will enable standards bodies to review and ratify specifications more quickly and, it is hoped, cut the time to market for implementations. Nagle says the COSE process is needed because existing standards bodies can’t keep up with the pace of technology development. COSE says its process will enable independent software vendors to focus their development resources on need rather than the systems and looks and feels they must target, as well as providing an expanded market for them to aim at. It will reduce costs and benefit users by reducing fragmentation, the paper argues. COSE envisages branded software that is plug-and-play compatible between vendors. It will demand detailed written specifications that govern its correct behaviour and says customers should not have to go to a particular supplier to get a full implementation of it – that it should be available from more than one vendor. Here COSE cites OSF/Motif and Novell Inc NetWare Unix client as examples, both now freely available X/Open specifications in their own right, and currently being implemented by a range of suppliers. COSE plans a different brand for each specification, but says there is no plan to brand for a complete environment. Describing its role as a catalyst to existing open systems processes, COSE proposes no new organisations and says existing Unix International, Open Software Foundation, X/Open and Posix technology request procedures and working groups should be retained. Although COSE reminds us that it is neither an initiative nor a group nor a club nor a consortium, it wants ad hoc COSE working groups – formed from major industry interest groups – to prepare formal specifications based on agreed recommendations resulting from what it calls common directions. The white paper would have us believe that end users demanding more commonality between different implementations of the same technolgy will drive suppliers and independent software firms toward a single technology direction.
By William Fellows
Such an impetus will become a general, or common direction, when it is supported by a sufficient number of vendors. Leadership of a common direction can come from any quarter, often intially through private meetings of a few vendors. A direction becomes a general direction only with an ample following made evident through public endorsements, and eventually through votes that formalise a specification. A common direction could be one in which vendors combine existing technology and write a new specification – like the proposed common desktop environement; endorse a de facto standard – like Motif; or set up a work group to write a draft specification of a common set of application p
rogramming interfaces. The rambling case which presents in the white paper identifies some of the problems that may ultimately lead to its curtailment. While COSE must steer clear of US anti-trust laws for obvious reasons, it desperately needs the marketing collateral of a club, a group, a consortium or an initiative behind it. As long as it can’t market itself then it will always lack teeth, detail, usability and clarity – all the things missing from this confusing white paper. On the other hand it must maintain consensus and a complex system of cross-checking between participating firms – content of much of the subtext of the white paper. COSE already has the difficult task of trying to explain and justify a new set of industry forces, processes and alignments that for all intents and purposes do not exist. Because of these problems, some insiders argue that COSE shouldn’t have bothered with the white paper. They say the market has been almost frozen in anticipation of the common desktop environment and that COSE should have concentrated on getting that specification out. Some are also concerned that the drive to standardise will lead COSE firms further and further up the technology path and scare off software companies that are or could be developing innovative technologies in the same sectors. Moreover, doesn’t it seem strange that the Unix vendors should be trying to teach the rest of the world a thing or two about about open processes? At UniForum meetings and other conferences and gatherings not so long ago, users were pleading for a common Unix kernel. Their pleas fell on deaf ears. Now, some years later, the prospect of those very same users being wooed to a new rival, Windows NT, is forcing those same hard-of-hearing Unix suppliers to offer up a half-baked common front-end patched together by an organisation that doesn’t exist.
The white paper could almost be taken as description of the process by which a handful of Unix vendors might steamroller their preferred technologies through the standards bodies. Perhaps some of the blame for the white paper can be laid at the feet of Digital Equipment Corp. It said it wanted documentation and pieces of paper describing the COSE process before it would decide whether to commit itself and some of its technologies, to the effort. DEC’s particiation was expected to have been confirmed last week but executives were still meeting to discuss the implications of white paper as we went to press. The firm says it wants fully to understand COSE and the movement behind it… figure out what impact COSE will have on customers and whether it requires a change of strategy or products. The likely answers are none, no and no, according to one source, who says COSE’s emphasis on openess and commoness fits nicely with DEC’s own thinking. Other insiders say DEC is more interested in staying up with the standards game and less concerned with COSE’s technology specifications. DEC’s support for COSE is expected any day.