I wonder if this discussion might be more productive if someone offered a particular concrete situation they encountered that reflects the limitations that they feel are holding Java back. If the author reads this, I would call on him to supply a specific instance that would allow us to get away from opinions, and rhetoric. Discussing an example in this way (...more likely a series of examples) will probably help us get closer to understanding each others points. Given that understanding we could then expand back out to more broader discussion of he subject. Thoughts?

I personally kind of hate comparisons between XML and Java. They are two totally different kinds of tools that are used in totally different ways. To suggest that XML (a spectrum of markup languages) is somehow 'more flexible' than Java (an OO programming language) is meaningless. They are both flexible and both rigid in totally different ways. An xml schema can be just as rigid and inflexible in the face of future changes as a strongly typed java interface signature. The correlary to the deprecated method in a java api is a long list of 'optional' attributes in a tag definition. I would suggest that the vast majority of problems that 'limit flexibility' in Java (or any programming language) have more to do with design patterns - too many anti-patterns and not enough use of good patterns. That is not to say that there could not be differences in the language that would help encourage good patterns and discourage falling into bad patterns. But even without fundamental changes to the language, much of that could be achieved through better development tools that better guard the developer from falling into anti-pattern traps and encourage the use of good patterns that facilitate future flexibility.

The point about strong typing vs. weak typing is not trying to bring back memory spaces as a way of passing program arguments. Typing is good for the compiler, however it is too inflexible. XML has a benfit in its design which is that the data and the meaning of the data are combined, so it can be partially complete, pieces ignored by the receipient if it doesn't understand it. I'm not advocating XML as a transport between Java program modules, I just think we need something less brittle than what we have now that can't accomodate change. This is similar to how the Java XMLEncoder is superior to binary serialization, because it can deal with updates without everything breaking

Somehow, I doubt it is the fact that java is a strongly typed language will be the determining factor in its long-term success. There are much bigger issues, primarily whether java is percieved by customers to be an appropriate tool for the solution of their real world problems. The number of deprecated calls in some of the APIs does not in my experience jump to the forefront of things to consider when comparing development tools and languages. Yes, it is symptomatic of some things that are going on, but it isn't imho a deal breaker with most folks. The fact is, the quality of development tools available and the ubiquity of experience of developers with a language and the associated tools will tend to be the predominant factors. I'm not saying here that java is destined to succeed (whatever that means) or not with or without faster adoption of changes. I just don't think that strongly typed signatures are the major negative issue the article makes them out to be. There are, after all, real benefits to a strongly typed language.

I could not disagree more with this article. Maybe you have to be as old as I am to remember the days before strong type checking. The lack of strong type checking in assembler, c and basic lead to very slow development process due to bugs that would have easily been detected in java, Modula 2, or Pascal. To me, you are advocating a change back to the bad old days. Elimination of strong type checking will not solve the problem of brittleness of code. Even eliminating type checking completely will not solve the real issues that impede software state of the art. To reach the next state in software development, we have to be able to stand on previous developers shoulders - that means reusability and modularity. What is the biggest impediment to reusability? From my perspective, it is the proliferation of computer languages. How many times do we have to spend software $ rewriting quicksort when we could be using those dollars to write more intelligent database management systems?

You mention XML as a positive thing because it is text based. That would only be true if their were fewer character sets than there are representations for fundamental types like float and integer. Last time I looked, this was not the case. The number of character sets is roughly the number of standards (ascii, ebcdic, unicode, etc.) times the number of languages (english, japanese, etc.) - there is plenty of room for lack of compatibility and brittleness in the XML world. Also, XML only addresses a very limited universe: fundamental data types. It does not deal with: semantics, complex objects (e.g. multimedia), mobile code, security, inheritance etc and most of all, performance. I see XML as negative evolution because these problems are already solved in JVM byte code. To call XML portable would be to mis-characterize it - the standards and tooling are so complex, they really exist on only 2 platforms: Java and .net. Even those platforms are not exactly compatible.

At some point in the evolution of any API, you hear someone calling for a do-over. They say, "it's too hard to evolve," or "nobody uses those methods anymore." While there might be the best of reasons for changing, incompatible changes to any API cause much more confusion and strife than living within the boundaries of compatibility.

If such a thing occurred to Java, it would kill the language. All the new cool features would go into "Java 3", while the entire corporate world stayed on Java 2, and thus the tools would also remain on Java 2 (it's a chicken-and-egg problem that we've seen numerous times before). Eventually, anyone who got fed up with the lack of new features in Java 2 would decide to move to a new language, and the lack of tools for Java 3 would be a big strike against it.

No, painful as it sometimes is, compatibility must be preserved. Once you decide you can break compatibility, you fracture the entire user landscape, and Java can't afford that.

Certainly, maintaining interfaces that supply services to support many applications is not easy and they do evolve over time. This form of distributed computing has advantages that do not come without a cost, but I would contend that the problems the author of this article is citing are as much or more a problem with how we design the interfaces as they are with the implementation. I agree, that you'll never get the interface right the first time and yes it will most certainly change over time. What I am talking about is tight cohesion and loose coupling. Nothing magical, but not always practiced. My experience has been that as long as you keep the services loosely coupled and start with some flexible command pattern super classes, that you can deal with lots of these issues. Is it perfect, no. I still, however, remember the headaches and hard to find runtime bugs that can be introduced through a lack of strong data typing and the use of pointers to blocks of memory. I have programmed in everything from mainframe assembler to Java (...and .NET). I still would rather use it than other language/platform I've ever encountered.

>The class java.util.Date, for example, has more public deprecated methods than non-deprecated ones.

IMHO, Sun should come up with a light-footprint version of JVM that does not support deprecated version of methods. This JVM should not replace the existing one, but present an option to the recently developed Java applications

If Apple couldn't withstand the digital gottedamerung that is Microsoft, then what makes anyone think that Java can?

Rhe buzz at the last software conference I attended, a conference previously full of Java developers and talk, was all about Ruby. The web is filling up with talk about the R word. Yesterday an old friend of mine IM'ed me to say that he had a new job and his boss wants him to do this Rails thing and did I think it was for real.

>> Survival against change - whether in nature,
>> technology, or business - is determined by one's
>> ability to adapt. Java needs to move forward by
>> being more agile

Java's popularity might be declining, but the concept of its JVM sure isn't. And I'll be using Java for a considerable amount of time in the future anyway.

||| If Java is to remain at the forefront of technology for the next 10 years, it ... needs to find a way of decoupling API calls between internal code and external blocks, perhaps even introducing soft typing calls across program boundaries or having flexible message transport across modules. |||

It will be interesting to see what Sun itself has to say in response to this. I am assuming JDJ has invited a response?