X `Window Of Opportunity' Opens Wide

Products

Focus

X `Window Of Opportunity' Opens Wide

X-Window's Long-term Future Looks Clear, But Spots Remain To Be Polished

Mike Azzara
UNIX Today!

March 20, 1989

Predicting is hard-especially the future.

-Niels H.D. Bohr, creator of the Bohr Atom model, 1938. Even Niels Bohr, however, would have to acknowledge that the future of X-Window networks appears downright dazzling.

Nearly every hardware system vendor supports X host software; more than a dozen available packages add support for X's desktop component to PCs and other workstations; roughly half a dozen dedicated "X terminals" have hit the market in the last five months. In addition, Novon Research-which has been commissioned by OSF, among others, to study X and user interfaces-believes that the major hardware vendors are preparing X terminals of their own.

Still, several significant hurdles lie ahead for X:

* rampant confusion over exactly what X is and does;

* precious few X applications or X end users exist today;

* competition from similar, proprietary solutions; and

* some industry experts-like Steve Jobs and Bill Joy-continue to insist that X is "brain dead."

The confusion stems from several quarters, even though the fundamental ideas behind the X-Window System, developed by MIT as part of the DEC- and IBM-funded Project Athena, are simple. X provides a vendor- and operating system-independent way to distribute an application-involving graphics or not-across a network, so that applications processing can be done in a host and display processing can be done in a desktop device.

MIT's X nomenclature, though correct in theory, kicked off the confusion. In X, the desktop software component is called the "server" and the host component is the "client." Confusion results because the rest of the world has already become comfortable with the client-server database architecture, in which the host is server and the desktop its client.

Technically, however, it makes perfect sense. The terms "server" and "client" are derived from queuing theory, in which a server is someone or something that takes requests from many clients and serves them. X-Window's desktop software acts as a display server, managing and controlling access to a bit-mapped screen, explains Paul Evans, engineering systems administrator at Loral Instruments, San Diego. X provides multiple windows, each of which can be a session with a different application and host. Therefore, each host is a client contending for access to that screen, even though the host is actually executing the application code.

Even more confusing is that X is often called a universal "user interface standard." Yet, there are still myriad factions warring over the right to define their user interface as the standard. How can both be true?

Both are true because the X-Window System, in the course of distributing an application, provides all the tools necessary to build a multiwindow, icon-driven graphics user interface. In fact, different companies have used these tools to build many different multiwindow, icon-driven user interfaces, each with its own slightly different look and feel-and each declared standard.

This, according to Novon Research director Brian Boyle, results in both good news and bad news. First the bad news: real end users may not realize the benefit of a common way of interacting with all applications unless some company or organization really can set a look-and-feel standard on top of X; the good news is that because these X interface-building tools are common, any application that is written to X is portable to any computer network supporting X.

But there is still hope that a standard user interface could be set. Not only are several companies and/or consortiums (AT&T, Sun, IBM, Microsoft, UNIX International, OSF) trying to set such a standard on their own, but standards bodies are getting into the act. So many, in fact, that an exhaustive list of them all would be prohibitive for space reasons, but they include the IEEE, the International Standards Organization, ANSI, the Human Factors Society, the National Institute for Standards and Technology, DARPA and X/Open.

The second hurdle for X-its lack of applications-is likewise moving well along toward resolution. Boyle says Novon's research has uncovered large numbers of programmers working on X-based applications in user companies, VARs, software developers and systems integrators. In fact, most X terminal vendors-Visual Technology, Acer Counterpoint and Network Computing Devices-report the large majority of early installations are going to software developers.

NCR recently jumped into the X terminal market at UniForum (see story this page).

At Loral Instruments, developers chose to use X in their system for real-time analysis of test data from aircraft, spacecraft and launch vehicles. In this application, data are downloaded from the craft in the air to a front-end telemetry system, which uses TCP/IP protocols to pass that data to a graphics application in a VAXstation 2000 workstation running Ultrix. Evans said the aerospace companies Loral serves have a bias in favor of DEC gear.

The graphics application processes the data and then acts as an X client to pass the resulting information to the X server for display. In this case, client and server execute in the same workstation, a common phenomenon during these early days of X development, according to Network Computing Devices executive vice president Judy Estrin.

Loral is investigating the possibility of distributing the client and server, and is a beta test site for NCD's X Display Station. The company already uses the NCDs-along with its Sun Microsystems workstations-for developing code for the application, Evans said. About half a dozen of these telemetry systems are installed in what Evans described as "Mission Control rooms" that can have up to five or six of the real-time telemetry front ends and as many as two-dozen workstations.

Loral's Evans explained that in 1986-87, when Loral chose to go with X, there were not many technologies to choose from. Loral primarily considered Sun's NeWS (Network Extensible Window System) and X.

The X-Window System won simply because it was vendor-independent, even though Loral was already a Sun customer and Evans believes NeWS to be more sophisticated. The X Consortium, including DEC, DG and IBM, had recently formed with 13 charter members and made a big media splash in support of X.

Evans, Sun's Bill Joy-and even many of the founders of the X Consortium-acknowledge that its formation was directed precisely against Sun and NeWS. It was a move to prevent Sun from setting a network windowing standard the way it had already set its Network File System (NFS) standard.

The result of that political action is that, today, X has become a universal de facto standard, which even Sun has promised to support. That's apparently been a bitter pill for Joy, who says NeWS is far superior. He describes NeWS as an on-screen version of the PostScript page-description language (ironically, Sun in November won a contract from MIT to add 3-D graphics capability to X).

A DE FACTO STANDARD

Whether inferior (i.e., brain dead) or not, X clearly has become the thing that is commonly available and on which most vendor-independent networked applications will be based, according to Boyle. "Marilyn Monroe may have been perceived as a dumb blonde, but that didn't stop her from becoming this country's de facto standard sex symbol," said Boyle. "And if you look at X, even the competitors are being drawn into the fray."

He points out that X's major proprietary competitor will soon run on X. Presentation Manager, a component of IBM's Systems Applications Architecture (SAA) developed originally for OS/2, will run on X in PM/X, being developed jointly by Microsoft and Hewlett-Packard. "And [OSF's] Motif is essentially a synthesis of these two competing technologies-X on the inside and PM on the outside," Boyle said.

The only major holdout is Steve Jobs and his company, NeXT. They're important because of media impact, not because they have any current influence in the market. NeXT's Next Step user interface is based on Display PostScript technology, not X, and Jobs has echoed Joy's sentiments regarding X's clinical death.

Nonetheless, even Jobs has said NeXT will support X if it becomes a widespread customer requirement. That, according to Boyle and many others, appears to be only a matter of time.