May 23, 2005, marks the 10th anniversary of Java's initial public launch. Ten years ago, Java's claim to fame was going to be its incorporation into Netscape Navigator, but that isn't really where the story began. Java was originally created at FirstPerson, a Sun-chartered endeavor, and was the brain child of James Gosling. FirstPerson created Java, which was called Oak back then, to enable them to build a prototype set-top multimedia device called *7. The name "Oak" is said to have been derived from an Oak tree that James Gosling could see from his window, and the *7 device starred a little character that we all know today as Java's mascot, Duke. Although the device never reached the consumer market, it spawned the technology that has so radically changed the software landscape. Four major and 10 minor releases later, we arrive at the present day.
Java Present: Java 2 Platform, Standard Edition (J2SE 5.0) "Tiger"
When the Java 2 Platform, Standard Edition (J2SE 5.0, code-named "Tiger") came out back in September of 2004, I have to admit I was excited. Sun had really dialed in on several areas where I had always perceived that the language was weak, most notably the handling of primitive types, the lack of type-safe enumerated types, and C-style IO formatting. Sun also added some rather powerful, although not always well-understood, constructs such as generics and metadata. Going beyond the language itself, Sun added a new API for monitoring and managing the JVM. And, of course, what self-respecting release of any software these days would be complete without promises of performance improvement? So, in general, the J2SE 5.0 release was a strong one, providing most developers, including me, with enough compelling features to warrant putting time into our project schedules for learning and upgrading to the latest and greatest.
Java Future: Java 2 Platform, Standard Edition (J2SE 6.0) "Mustang"
The next major release of the core Java 2 Platform, Standard Edition (J2SE 6.0, code-named "Mustang") also sounds very promising. Currently, Sun is projecting a release date of mid-2006, which may be a little aggressive, but it all depends on what actually does and doesn't make it into the release. Sun is taking a very pragmatic approach to managing the scope of the release. There are two avenues for features and functionality to get into Mustang. The first avenue is through Java Specification Requests (JSRs). Essentially, this is the process through which Sun handles the changes to API libraries currently existing within the core platform or the addition of new libraries to the core platform. The JSR process can be rather bureaucratic, and although Sun claims to have no influence over it, few people outside of Sun actually believe that to be the case. The second avenue for features and functionality changes is through bug fixes. In fact, there is even a mechanism in Sun's bug database for you to vote on the bugs that you would most like to see fixed. Although these votes are just advisory in nature, it is good to know that somebody is listening.
So obviously, the biggest changes to the platform are going to be encapsulated in JSRs, and to manage which JSRs are included in the release and which are not, there is of course a JSR. I warned you this could get a trifle bureaucratic, but it is actually a pretty good way of managing things. So you essentially have one parent JSR that holds reference to everything that is currently expected to be in Mustang. This meta JSR is known as JSR 270: J2SE 6.0 (Mustang) Release Contents, and as of the writing of this article, it contains 10 JSRs.
Let's take a brief look at each of these JSRs that has been explicitly slated for the Mustang release. Note that the order of the JSRs here is just an ascending list based on JSR number and in no way should be construed as an importance ranking. (In addition, there will probably be a JAXP JSR, but the number has yet to be determined.)
JSR 105: XML Digital SignatureThe W3C has a standard for digital signatures, and this will be an API to incorporate XML digital signature services. This can of course be done already, but the goal here is to create a standard.
JSR 199: Java Compiler API
Currently, your favorite IDE calls the Java compiler using a command line, which is rather awkward. The idea here is to add an abstraction layer around the compiler that the IDE can call programmatically. In the short-term, this will make life much easier on the IDE and will allow for the exposure of more complicated constructs that would simply be impossible to manage through a command line interface.
JSR 202: Java Class File Specification Update
This is a collection of "under the hood" changes to class files. The most notable one is to increase the max size that a class file can be in order to support the class files put out by some JSP generators. Unless you have run into this particular problem, this JSR probably isn't of much interest to you.
JSR 221: JDBC 4.0
To me, this is the most important JSR of the bunch. JDBC 3.0 was a step in the right direction, but since the release of J2SE 5.0, we now have some new language features that need to be leveraged into JDBC. Both generics and metadata will be leveraged into JDBC 4.0 and should help developers write cleaner code by reducing casting and redundant code. Developers often fall into two camps when dealing with relational databases. The first is the "I want an abstraction layer so thick I don't even want to know what is on the other side." These sorts of developers love mechanisms like container managed persistence. The other camp is the "Just give me an easy way to write SQL statements that are intimately coupled to the database scheme." Although somewhat less than architecturally sound, pragmatically this is the way that a lot of development is done, and the authors of this JSR are acutely aware of the situation and willing to embrace the more direct access methods.
JSR 222: JAXB 2.0
This was a close runner-up to the JDBC 4.0 JSR as being the most important JSR. XML, like relational databases, is a technology that is here to stay. JAXB 2.0 is all about making it easier to work with XML (specifically, XML schemas).
JSR 223: Scripting for the Java Platform
The name of this JSR is a little confusing, but the concept behind it is solid. Scripting languages like PHP are very popular, but they don't have an easy mechanism for accessing Java classes. The focus of this JSR is to bridge the gap between scripting languages and Java.
JSR 224: JAX-RPC 2.0
JAX-RPC 1.0 reminds me a lot of early versions of JDBC; it's really powerful, but clearly there has to be an easier way of doing a lot of this. Additionally, it seems to only scratch the surface of what needs to be done. If you intend to read only one of the JSRs to get a better sense of where Sun is going, this is the JSR to read.
JSR 260: Javadoc Tag Update
Javadoc was one of the first Java technologies that made me say to myself, "Wow, how did I ever get anything done before I had this?" However, my enthusiasm and interest faded over time, not because of disillusionment with Javadoc but rather because the class browser mechanism in IDE just got that much better. These days, I mostly use Javadoc only for perusing third-party APIs. However, the minor changes proposed in this JSR will make Javadoc more useful in these situations.
JSR 268: Java Smart Card I/O API
Smart cards are a big deal for Sun and Java. However, since smart cards are currently outside of my industry (and probably yours also), I have a hard time getting too excited about changes to an API that I will probably never use.
JSR 269: Pluggable Annotation Processing API
In order to understand this JSR, you most first understand JSR 175: A Metadata Facility for the Java Programming Language. Note that the terms "annotation" and "metadata" are synonymous. In some ways, I'm still undecided on metadata. While I understand it as a solution to the problem of defining the extra contextual information that we need for constructs like EJBs, part of me shivers and thinks about the evil abuses of the C/C++ preprocessor and hopes that metadata will stay on the straight and narrow path. So JSR 269 is thus about creating an API to make annotation processing easier.
In addition to the official JSRs, a ton of bug fixes are slated to go into Mustang. Most of these bug fixes are what one would typically think of as a bug fix. For example, Bug 4811096 (Allow mixing of heavy and lightweight components), something that in my opinion should have been addressed a long time ago, fits nicely into the definition of a typical bug. On the other extreme is Bug 5106661 (Improve Windows Look and Feel), which to me seems more like an enhancement than a bug fix. However, since it does not require an API change, it does not merit its own JSR.
One theme seems to bind all of these JSRs together and pretty much sums up the entire Mustang philosophy: Take what has already been built and make it easier to use. Sun has really laid a strong foundation with Tiger but has yet to leverage all of its potential across the board.
So now that the general scope of Mustang has been set, how does Sun plan on doing development? Well, Sun has been under increasing pressure to open-source Java, but doing so would put at risk many of the key principles, like cross-platform compatibility, that Sun has been so diligently maintaining. So Sun is working on a compromise. As of November 2004, the company has been doing weekly drops of both the binaries and source code for Mustang. While this is a step toward open source, this is not open source. The code can be used under a license that only allows you to modify and use it within your own organization. Of course, Sun has provided a mechanism for submitting your changes back (called Project Peabody) that then gives Sun the rights to distribute your contribution. Sun refers to this sharing and contribution of code as "transparency," but it should not be confused with open source and is not likely to appease the open source advocates.
Ten years of Java evolution have gone by quickly, yet it is sometimes hard to recall what it was like programming before there was Java. The J2SE 5.0 Tiger release has some truly useful foundational language changes. Although most of these changes can be readily used by developers now, they will become more valuable as they enable many of the features of J2SE 6.0 Mustang. Although Mustang is not due to become an official release until mid-2006, Sun is already doing weekly builds and making the source code available under a special license and is actively soliciting feedback and source contributions from the community. Not only is this good for Sun, but it gives us an unprecedented view of exactly what will be in Mustang. While Sun has taken some small steps toward open-sourcing Java, they have not actually done so, and the decision to go open source or not will be a daunting one. From a rebel language that only a few years ago was focused on justifying its own existence to now the being the nonpropriety enterprise language of choice, Java has come a long way.