What's Not So Hot with Java
October 6, 2008
HP 3000 owners are now considering what a 2009 without HP's resources will look like, since the vendor is closing off its MPE/iX labs at year's end. The demise of HP's 3000 lab efforts already has a precedent: the vendor's abandonment of Java on the 3000. The strategy could play directly into HP's migration desires, leaving MPE/iX software frozen while HP hangs on to the code which others could improve to satisfy 3000 sites. The biggest irony might be that Java is the most prevalent open source product in the world, but it needs HP to release its source to gain freedom again for 3000 sites.
This language promised a "write once, run anywhere" future when Sun first introduced Java in the middle '90s, a portable programming platform to deliver on the dream of "open systems." Even though open systems needed to wait until Linux and ubiquitous Intel hardware established the concept, HP leaped in by 1997 with a Java/iX implementation, and in later years touted a small number of 3000 customers making use of the language.
But once HP 3000 companies didn't swarm toward the solution, the vendor's diminishing lab staff had to turn away from the language as well as needed updates. Java/iX has been frozen by HP at version 1.3 for more than five years, a version which becomes less useful with every month HP hasn't touched it. Of course, that will mirror the hands-off future defined for HP's 3000 labs, a group of wizard-like MPE/iX engineers being put to work on other operating environments.
We wrote about this issue this spring, interviewing the last HP staffer to add something to Java/iX, Mike Yawn. At the time that Yawn "owned" Java/iX, he was passionate about reporting from the annual JavaOne conference, as well as presenting in 1998 the prospects of graphical interfaces on the cutting edge for the language.
Java, as it turns out, was one of the first projects which the OpenMPE advocacy group identified as a way for an outside lab to help 3000 owners. The language has a lot of momentum in the IT world. Today Charles Finley of Transformix, a migration company working to move 3000 shops to other platforms, said Java has become a lot better than what HP left on the 3000 years ago.
What Java is missing might not ever be recovered for the HP 3000 community, simply because the vendor did its own, proprietary work to create the Java Virtual Machine for MPE/iX. JVM is indispensible in getting Java to serve as an engine of commerce and transactions. Finley said
There are some native pieces to Java that are proprietary to HP. It does not seem to be possible to port a newer version to the HP 3000 without access to those sources, and when it was last discussed in my presence, HP was unwilling to give anyone access to the code.
On that last point there may still be hope; HP might offer access to the proprietary Java modules as part of a third party licensing arrangement. But Finley's company has had engagements with migrating customers that show how far Java has slipped under HP's 3000 stewardship.
"There was never a decent version of X development tools available on the HP 3000, so the graphical tools are not available on the HP 3000," Finley said. HP tried to introduce Swing, a graphical interface tool, for the 3000 — but once again, in the late 1990s, HP 3000 sites were far more interested in getting code Y2K-ready than creating code in an emerging language. It's not as if Java is now useless on the 3000, but comparing it to any other version shows why HP discontinued support of Java/iX during 2007, even though the language is still included in releases of the operating system. Finley reported today on the relative utility:
We have used Java on the HP 3000 to do a few little tasks and the version that is there is still useful to an extent. That said, we use Java on Windows, Linux, HP-UX, etc. and there are many things one is able to do with Java that are not possible with Java on the HP 3000
Other issues Finley mentioned about Java, like being a resource hog on the 3000, or having still-Spartan documentation, or being removed the bounty of free applications that could "run anywhere," could be resolved with 1. A 3000 emulator running faster than any current HP 3000; 2. Giving the documentation over to a third party like OpenMPE; 3. Making Java/iX current with the world's release by releasing the source code.
Keeping an open source solution proprietary does appear to contradict the concept of open source, even if the reason for HP's decision is a disappearing 3000 lab. Perhaps Java/iX can become the test case for how HP will license for open development a piece of the 3000's Fundamental Operating System.