The two issues that top the agenda whenever two or three Unix people are gathered together are the proposed Open Look user interface, and Open Software Foundation alliance of industry heavyweights. Right in the middle is thrusting young Sun Microsystems, so who better to try to put them into perspective than the keeper of the company’s technical soul, Bill Joy?
Unix was already 10 years old in 1977 when the young Bill Joy first came across it in its PDP-11 incarnation. Today, he is the self-assured technical spokesman of Sun Microsystems Inc, the Mountain View, California company that he helped found, and which is now the growing and probably the most highly regarded company in its field. Earlier this month, Joy visited the UK to give an impromptu talk at the European Unix User Show, focussing on two of the major talking points of the event: the proposed Open Look iconbased user interface for future implementations of Unix, and the Open Software Foundation that has brought IBM, DEC, Hewlett-Packard, Apollo and sundry hangers-on together in the unholiest of alliances to try to derail AT&T’s thundering train that now seems to have Sun on the footplate as its energetically stoking fireman. Bemused People are for and against Open Look without ever having seen it, said Joy, seemingly bemused by the politics that have entered his field of technology over the past few months. At least at the show visitors could actually see the product demonstrated for the first time – or at least a pre-beta version as it was described by the demonstrator – and Joy was keen to reinforce the first sighting with his explanation of the methodology behind its development. Since Apple Computer turned an idea from Xerox Corp’s Palo Alto Research Center into a commercially-desirable product, it has been recognised that a graphical user interface needs its own look and feel, a toolkit to implement that look and feel, and a graphics library. Taking its lead from the Andrew project at Carnegie Mellon University, Open Look added another element, a window system platform providing multilevel input/output, a graphics model, and device independence, using either News/Postscript from Sun itself or the X Window System from the Massachusetts Institute of Technology. But, claimed Joy, Open Look has aimed at a level of consistency of operation that he says the Mac interface has not reached. Open Look uses direct manipulation, without a menu bar at the top or function keys to break the paradigm, he said. Some have criticised Open Look for its rather ungainly analogies, such as the elevator scroll bar, thumb tacks to pin menus on the screen, push buttons, and (in the early versions at least) an alarming three-dimensional zoom effect for error message notices (the whole idea of the thing is to make the computer accessible to people who are scared to death of the computer, so it doesn’t seem to be a very good idea to frighten the life out of them the first time they make a mistake). But Joy’s opinion is different. It’s a high quality graphics design, not like Presentation Manager, [the Microsoft user interface for OS/2] which is kind of ugly and not aesthetically pleasing.
Joy went on to analyse the various approaches that manufactu rers have used to carve out their own chunk of the computing marketplace. The traditional one is to develop your own technology and sit on it, like IBM, DEC, and Apple have done. Sun’s approach was different; it used what I call ‘blitz krieging’, or attacking by licensing your technology to ev erybody, in the hope that it will become established. Ac cording to Joy, standards are set in the marketplace, not by committees, so it’s vital to get your technology out there as fast as you can. The blitz-krieg technique presents a problem to the proprietary vendors: they either ignore it and do nothing, thereby risking a dramatic loss of market share; adopt the technology and admit they were wrong (like Data General, said Joy); or take up the third option, that of kidnapping and burying the competition. This, he said, descri
bes the Open Software Foundation. You take the standard up, but you say that it’s not interesting. But isn’t the Open Software Foundation anencouraging move towards setting a fully independent standard? It’s easy to agree on a subset of standards, said Joy, but no one will sign up for getting rid of the incompatibilities between the various proprietary extensions, say in the individual Fortran offerings of the Software Foundation members. System V sets out to do that with Unix – and that is why it was so hard for all those companies, with their proprietary extensions, to accept it. Scorn Warming to his theme, Joy poured scorn on the objections of the Group to Sun’s controversial relationship with AT&T, which gives it great influence over the future development of Unix. Faced with the near total commercial failure of its 3B family of Unix supermicros, AT&T last autumn decided that the follow-on product it had in development simply wasn’t going to rescue the situation, and almost in desperation turned to the company that seemed to have most of the answers, including thorny ones like how best to bring the diverging strands of Unix together into a single standard, and what would be the likely best platform for next generation Unix systems. AT&T agreed to use the Sun Sparc RISC microprocessor architecture, brought Sun in on the development effort for the next release of Unix, and cemented the relationship by agreeing to bankroll the ambitious company’s next spurt of growth by putting up cash for new shares. They said we would optimise Unix for the Sparc processor – I wish I knew how to do that, commented Joy. The only secret is to write a good compiler. And as for Sun having an 18 month lead in porting new developments, Sun is actually shortening its lead by opening up its own technology. It takes only two months to port Unix, not eighteen months – the time the Open Software Foundation has set for itself to develop a complete environment from scratch.
