Will lights on Broadway outshine Java?
X Window Consortium reveals the successor to X11R6
In the not-too-distant future as envisioned by Oracle, Sun, Netscape, and countless Java programmers, cheap Internet appliances have swept through corporate networks and are appearing in a variety of forms as entertainment devices in homes.
The casualties of the Java-powered Internet appliance tidal wave will include Microsoft, Novell, Banyan, PC cloners, and others, including the X Window Consortium. With local display handled by an embedded Web browser, and servers sending applets instead of running applications and processing GUI commands, the Unix community will have relegated X to the Smithsonian.
In response to this new vision of future computing, the X Consortium is disclosing its own view of a post-X world. X11 R6.2, is the final release of the X Window System as we know it. Paul Lavallee, director of marketing for the consortium, said the group thought it important to find a way to ride the Internet's wave of popularity. In the fall of 1995 the consortium briefed members of the X Technical Advisory Group on "Broadway."
Stated most simplisticly, Broadway will treat every Web browser as an X desktop, within which it can run X terminal windows. Voila. With no changes to applications, every X application, and some non-X applications, will be Web enabled, Lavallee said.
There are significant differences between Broadway's vision of a Web-enabled computing world and the vision held by Java-powered approaches. It remains to be seen whether both will be accepted as complementary technologies or if one will lose to the other.
"The execution models are fundamentally different," said Bill Yoder, Hewlett-Packard's representative to the X Consortium Technical Group (XCTG). The applet model of Java assumes that some code will be downloaded to the client desktop and run locally. Broadway's entire application will run on the host, and only the display will run locally inside a window in the Web browser.
Digital's representative to the XCTG, Rob Lembree, said the issue will boil down to "data locality" -- how much data has to be manipulated and where those compute cycles take place. If an application needs to manipulate a lot of local data, it makes more sense to crunch that data with an applet. Conversely, if the application is working with centralized data, the Broadway model is ideal.
Perhaps more importantly, if Broadway delivers on its design goals, existing applications can run over the Web without modification. Java applications, on the other hand, are a whole new code base. That raises the investment protection red flag, Yoder said, the knowledge investment -- training, experience and code -- in existing applications.
"I think there is no question Broadway will be the platform of choice for legacy applications," Yoder said. "The question for software vendors is, if they're going to significantly enhance an application, or write a new application, will they do that with Java, or will they simply write it the way they always have, and assume Broadway will take care of the distributed Internet functionality."
By leveraging the X Window System, Broadway also provides a far more mature imaging and graphics model than Java, said Ed Lee, product manager at Integrated Computer Solutions. The imaging extensions to X, both for 2-D and 3-D graphics are mature and well-suited for distributed applications in that they include compression technology. The notion of extensible widget sets is also something that has matured over the years in the X Window's world, Lee noted.
From a technical perspective, most sources interviewed for this story say Java-like strategies and Broadway can indeed be complementary. The issue is mind share and resources.
"It all boils down to where you put your resources," said Hal Jespersen, a senior staff engineer with SunSoft and Sun's representative to the XCTG. "In an ideal world, everyone would invest in Java and Broadway. In a world of competing resources, I wish the Consortium would have completed some of the interesting technology pieces they were working on, but have now shoveled into the Broadway project."
Jespersen pointed to remote audio support, enhanced security, and the low-bandwidth X (LBX) projects -- all of which predated Broadway -- as highly valuable technology components about which Sun is enthusiastic. And all components could have been done without fundamentally recasting the X Window System as a Web-enabled distributed applications platform.
"Sun does not have a firm position on Broadway," Jespersen said. "I remain to be convinced whether it can all be put into a single package that solves customer problems in a satisfactory way. The No. 1 concern is performance."
Yoder and Lembree, who are far more enthusiastic about Broadway, agreed that performance will be the key issue.
The Consortium's Lavallee says that LBX, now renamed X.Fast, is concentrating on circuiting, caching, and compression techniques to significantly boost throughput. He expects applications will run satisfactorily on connections as slow as 28.8 kbps, and said early lab tests are delivering promising results.
"We are trying to leverage existing applications to this new computer desktop -- to let people use those applications without modification," Lavallee said. The result, he said, will be far greater hardware and software independence, and a much greater level of functionality on corporate intranets.
No doubt Xtech '96, being held in San Jose, February 12-14 -- the same time UniForum is in San Francisco -- will give many more people a chance to learn what Broadway is all about. Then we'll see if it starts generating as much excitement as Java.
Lavallee expects the X Consortium to deliver a beta release of Broadway
to Consortium members in May 1996. A production release for the general
public is slated for August.
--Barry D. Bowen
About the author
Barry D. Bowen (firstname.lastname@example.org) is a business computing analyst and writer with the Bowen Group Inc. (http://www.nas.com/~xeno). Reach Barry at email@example.com.
If you have technical problems with this magazine, contact firstname.lastname@example.org