Unfinished Business: Java

From the beginning, Java has suffered from hype. Because its early promises were so vast, it couldn't possibly measure up to the claims. (Not even dark chocolate is that good.) It was impossible for Java to achieve all of its goals. But unfortunately, one area in which I think Java has lost big-time is cross-platform support. And I sure wish it hadn't happened that way.

In the summer of 1996, an IBM executive declared to me that "Java will make operating systems irrelevant" and predicted that by 2000 most desktop applications would be 100 percent pure Java. (OK, raise your hand if you run more than one Java app regularly.)

It's understandable that IBM, with so many operating systems to support, would yearn for a cross-platform solution. But IBM drank the Java Kool-Aid to the bitter dregs, taking the matter far beyond audience comprehension. I recall one IBMer suggesting to a software developer that he write a device driver in Java. In September 1996, IBM executive Mike Lawrie encouraged end users to believe that ViaVoice might soon be ported to Java.

It's not that I'm anti-Java. Quite to the contrary: I believe that Java is an excellent tool for the right tasks. Unfortunately, the promise of "write once, run anywhere" was broken so long ago that nobody remembers where the warranty card is.

Recently, I reviewed a Java application that ran fine on Windows 9x and NT. But the program crashed under OS/2 and hung the Macintosh. I didn't have the time or energy to watch it fail under Linux. If the vendor was only writing for Windows, why didn't they just write a native Windows app?

Even when a Java solution ostensibly promises to support everyone, browser and Java version inconsistencies make life frustrating for both user and provider. An online correspondent sent me the excuse-laden "use another OS" e-mail he'd received from tech support at SmartMoney.com, when its Java stock charts wouldn't run correctly under his chosen OS. He voted with his feet and found another site to support him. Yet, I can commiserate with the service provider, because - as stated eloquently at www.javaonthebrain.com - "Anything beyond 'Hello World!' will create a problem on some platform."

Yet, I don't see anybody working to fix this situation. The tacit solution seems to be "just run Windows," which is diametrically opposed to the original goal. As evidence, note that IBM says it will no longer upgrade the OS/2 version of VisualAge for Java. After pushing OS/2 customers to nonproprietary technologies, IBM is yanking developer support for the major nonproprietary technology available to those customers. Nor is IBM claiming poor sales; instead, an IBM spokesman justified the move by telling me that the tools were all on Windows, AIX and Linux. Tell me: How does that serve your OS/2 customers?

"Write once, run anywhere" was a lofty goal, but one worth pursuing. If vendors and service providers aren't going to attempt it anymore, why bother with Java?

Esther Schindler is technology editor of Sm@rt Partner.