Volume Number: 13 (1997)
Issue Number: 9
Column Tag: Viewpoint
by Eric Gundrum
There has been more hype surrounding Sun's Java than anything else I remember about the computer industry. I am not entirely sure why Java has generated so much excitement. I suspect it has something to do with the promise of truly portable computing; that is, write code once and run it anywhere and everywhere. Unfortunately Java is still very far from realizing that promise. In fact, it is about as close to that promise as bad diner coffee is to a cup of home-brewed Sumatra.
Don't get me wrong, I want a product to do what Java promises just as much as everyone else. I would love to be able to sit down in a cafe with a public Java terminal, log into my web site, read mail, review my financial accounts, work on a few articles and write some code. Doing this from any computer, anywhere, and at any time would be wonderful. This capability would likely result in computers becoming more ubiquitous than telephones. However, Java is not up to it just yet.
In recent weeks, I have been doing some research into Java. This language seems well designed when compared to the likes of C++ and the other common object-based languages. The Java designers seem to be well versed the various object-oriented languages and pulled the best principles and techniques from many of them.
I particularly like the appropriate mix of keywords and punctuation. People cite C++ as an obfuscating language partly because it tends to use punctuation for everything; just look at how many different meanings there are for a colon. C programmers complain about Pascal's use of keywords for everything. They get irritated by having to type "BEGIN" and "END" an infinite number of times when simple braces will do just fine.
Java's syntax looks a lot like C++ on the surface, but it includes some of the verbosity of Pascal where the extra verbiage can be a real benefit. Java's designers did not go overboard; they use C-style braces to define scope, so you won't get tired fingers from typing "BEGIN" and "END" all over the place. However, they added "implements," "extends," and other keywords to make the structure and meaning of the code clear from the text, rather than relying on your memory and the terse symbols of C++.
Another nice improvement is the absence of header files. Classes are declared in the same file in which they are implemented. The class declaration is included in the resulting binary object. Want to use a Java class in your applet? Just "import" the library file containing the class. The lack of header files, coupled with dynamic binding can eliminate many of the common problems coordinating the build process on multi-person projects. It also means fewer source files to maintain.
One of the most touted improvements is having memory management built into the language. No longer do we have to waste so much time tracing down rogue handles or other memory errors. Unfortunately, the implementations of garbage collection in most Java environments still leave a lot to be desired, but this is just an implementation issue. Sooner or later the Java vendors will provide more efficient garbage collection to gain performance better than their competitors.
The biggest problem with multi-platform Java development is that there are too many flavors of Java and they don't quite work the same. This means that we must test our applications on all of the available Java platforms if we want to be sure it works for the multi-platform audience. Java programmers have reported to me that they can spend as much as 80% of their development time tracing and resolving platform-specific idiosyncrasies. Unfortunately, it looks like this problem will get much worse before it gets any better. There are many more implementations on the way, including those embedded in microchips.
What we need is an unambiguous specification for the language and libraries, and a certification process to ensure each implementation abides by the spec. For the moment, Sun seems too busy pumping the Java media wave to deal with this issue. (Sun's "100% Java" campaign is an example of this hype.) If this problem is not solved soon, it will likely be what stops the seeming endless momentum of Java.
Java is a nice evolutionary step beyond C++, but it is not the be-all and end-all of languages. Java incorporates features other languages have had for years, bringing these features into the mainstream, but the implementation is still immature. In time, Java should improve significantly, the programming community will become familiar with these "new" features and they will demand them of whatever succeeds Java. First, we have to get them working in Java.
Apple has made a lot of noise about Java being important to their future success, and to ours. This makes sense. If Apple's hardware business dries up, and their OS business dries up, what's left for them to do with their valuable brand name? Apple has been positioning themselves to be an important player in the Java sandbox.
Even so, Apple's Java strategy is way behind that of the rest of the world. Currently, Mac developers have to wait far too long if they want to use the latest Java technologies on the Mac. Many developers don't wait, they use another OS. Hopefully, Apple is fixing this problem; only time will tell.
