JavaSoft in the News

This week, Javasoft made news with its involvement in the Mars Pathfinder mission, the release of JVM version 1.1.3, and the preview release of Java Foundation Classes.

We've already intimated that Java had NASA's attention; they made public statements about it some months ago now. We're not surprised that NASA was one of the first big organizations to take a shot at serious Java applications.

The organization is unique among computing groups in the United States federal government, where often ridiculous hiring and bidding policies virtually ensure that only the worst available talent is hired for a given job. The unique opportunities NASA affords its employees have served as incentive when decent salaries have not--affording the space program some of the best talent in the world.

We've seen evidence of several different projects at NASA benefiting from Java; this week, the focus is on the Web Interface for Telescience (WITS) applet, as telescience is the order of the day for the scientists attempting to pilot a small robot that's lately (and very extravagantly) been transported to the surface of Mars.

Of course, the applet and underlying control system are still in development, and what's currently available is a demonstration, populated with actual Pathfinder data. Nasa claims intent to use the system for actual robotic control for a mission targeted to arrive on Mars in the year 2001.

Piloting a robot on Mars, we probably needn't point out, poses what are probably the most formidable control system challenges yet faced by humankind. Hopefully, by then, they'll have fixed the rest of the bugs in the JVM.

We do seem to have been due for another maintenance release of the JDK, and now we've got one--1.1.3 is out. We read the changes posted by the Java team, and were a little surprised at how few there were, although the list we saw may be deceptive or inaccurate.

Summing it up: Three race condition problems in the AWT, the broken profiler in java_g, a particularly nasty security problem with the type verifier, and performance problems related to AWT font metrics all got attention in the release.

The practice of releasing additional, even flakier bits of testbed technology with these releases has now been formalized with what Javasoft is calling "Technology Preview Packs." We're very amused.

The download page offers a JIT compiler (you guessed it, the Symantec JIT) and a native thread support package for Solaris. The latter, for those eager to jump on Sun for doing what Microsoft is doing, is not for providing direct access to Solaris threads, but rather for allowing the Solaris JVM to take advantage of the kernel-level threads in Solaris. Interesting.

As for the Java Foundation Classes, Sun's timing couldn't be better. Their arrival, and the stir they're beginning to cause, illuminates the discussion of foundation classes we've been having for the past few weeks.

It's unfortunate, we realize, that all this marketing hype is preventing things in Java from being descriptively named. We'd much rather they called it: "AWT version 3," "Extended AWT," or at least just "Swing" (an allusion to Duke Ellington's famous quote, "It don't mean a thing, if it ain't got that swing.").

For what it is, Swing is big, and it's very, very complicated. It's designed to unify a lot of funky and strange user interface functionality into one big package. Our biggest concern at the moment is how skillfully it was integrated with the current user interface facilities of Java.

We will, of course, dauntlessly pick it apart, all the while making snide comparisons to Microsoft's Application Foundation Classes, in very short order.