The comments about the annoying ads on this site are right on!

This is among the worst obscenities ever placed before human people...

I find it ironic that this site is writing about development and web design and is so completely ugly and filled with moving advertisements that it gives me a headache to read. Stop the madness!! The ads are an irritation.

I agree that the Richard Batty (whoever the heck that is) quote is useless other than being purposely inflammatory. Ignoring whether or not Java has been / is /will be viable on the desktop for rich/thick applications, the point of Gosling's comparative allusion in his comment about 'ones that just generate web pages' is validated simply by noting that Java is used to implement back end systems. A least a significant percentage (if not a majority) of Java code used in 'web applications' does not have anything to do with rendering the actual html page, but is instead performing business logic of some sort.

Well, that very last quote is quite true or at least has been for a long time.
Trends like Rich Client Applications (based on Swing or SWT) bring Java to the desktop again. And there seems no serious competition other than partly from C# and VB.NET on Windows.

Ironic enough that Rich Client trend really went hotter than before after Eclipse launched their RCP not based on Swing or what Sun and Gosling have given to Java only...

Lisbeth,
some people like their compiler to pick up their mistakes at compile time rather than runtime, and take advantage of runtime non-virtual call optimisations. It can be _very_ costly if ae fault is not picked up until substantial sums of money get stuck, because the software misbehaved or broke, or runs very slow because the interpreter sucks. Been their done that, like strong typing much better, if only for the piece of mind and, yes, reduced maintenance cost over scripting language apps.

"Developer" Roy Batty?

Of all the so call comments left by JDJ the one picked is the most imflamatory and useless of them. Roy Batty (AKA Dick Ford) is well known to be a troll of many different weblogs. Including one here...

http://www.artima.com/forums/threaded.jsp?forum=106&thread=150515

What exactly is he backing? Nothing from any of his statements (only to incite and be an ass) and yet he has contributed NOTHING in the face of Gosling or Rossum...

For shame JDJ for highlighting a troll!

I watched Java getting popular 10 years ago and joined the fun early on (moving on from C++ to Java). Still like Java and also do quite some C# which is more or less the same.

Observing what is happening around Ruby (and Ruby On Rails in particular) I recognize the exact same momentum/enthusiasm and also reluctantly recognize that Java is getting stale.

I think Sun would do well to embrace Ruby on the Java JVM platform in order to keep Java (the platform) relevant 10 years from now. The weakness in Ruby is the Ruby "VM" which is the chance for the Java JVM technology. By combining Ruby the language with the great Java VM + Java libraries, Sun could produce the killer technology for the next 10 years (and easily outdistance C# and .NET).

P.S There is a partial ruby implementation called JRuby that runs on the Java VM. As such, this implementation has indirect support for native threads and unicode (which is lacking in the official Ruby "VM"). However, JRuby is not the "official" version nor currently complete enough to run "ruby on rails" or other main offerings (allthough the implementers are working hard on this). Also, JRuby is currently not very fast compared to Jython or similar scripting language implementations on top of the Java VM. Still, JRuby is very promising and worth a special notice.

The thing that Gosling - and a lot of other people in this debate - can't seem to see is that "strongly-typed languages" and "scripting languages" are not polar opposites.

Strongly-typed languages are generally characterized as being compiled into machine code (actually Java is compiled into p-code), and having data types established at compile time.

Scripting languages are characterized as being interpretted at run time with data types established as the code runs.

These two characteristics of when code is parsed and semanticly analyzed, and when data types are determined are not the same thing. It is perfectly possible to have an interpretted language with strong typing or to have a compiled language with weak typing.

In fact there are tremendous advantages to weakly-typed, compiled languages, especially where the interfaces are weakly-typed. How much recompiling is done because the data type of a property in a library has been changed? How much of the Windows DLL nightmare would be solved by it?

It's time for people to start thinking outside their current paradyms.

Too many people are missing Gosling's point. Specialized languages are great and often, if not usually, the best tool to solve a problem in a given problem domain. However, the world is full of many complex problems that cross multiple problem domains and there are many advantages to having a single language that will address all those domains. I'm not saying this is always preferable, but it definitely has proven by the market to be a preferable thing in a significant share.

Before you compare language / tool usage, you first need to scope whether your problem is restricted to a given problem domain or crosses multiple domains. You then need to evaluate whether you need a cross-domain tool or whether your particular scenario is better solved by using specialized tools at each point. If the former, then it is reasonable to compare Java to other tools that also cross those same domains and in this case few tools compare with well with Java. If the latter, then Java may still do well in a given specialized domain, but it may just as easily be eclipsed by a more specialized alternatives.

For example, if I want to only build a fast scientific computation, I still may be best served to use good ol' Fortran or C or, who knows, even assembler. Or I might use a scripting language like IDL that is loaded with powerful primitives designed specifically for scientific computations. I might get away with using Java but I doubt seriously I should begin to try to address that with a interpreted scripting language like Ruby that does not have the correct set of primitives for this problem domain. On the other hand, if I am building a web app, Ruby or PHP may be by far my best options, while Java is not unreasonable and Fortran drops right off the table of consideration. Now, suppose I have to build a web app that is backed up by my super fast scientific computation? That is where it gets interesting. Am I best served by creating the web app in Ruby and the back end in Fortran? Or is it better to do it all in Java? There are a wide variety of factors one has to consider here from talent resources to cost of integration to pure performance. There is no single answer. The market for the most part has leaned toward the "use Java in some form for all stages" answer so far, because Java is a reasonable solution in most all problem domains. However, there are clearly many cases where it is not sufficient or we need 'better than just reasonable' and that is why there is a ton of room for other tools and languages that are stronger in various specific domains.

Finally, it is the nature of programming languages to evolve and grow in scope. Who knows? In the future, maybe Ruby will encompase a broader scope of problem domains than Java. If that happens, I guarantee that this same conversation will be happening where someone is advocating the advantages of specialized tool over the 'beast' that Ruby will have become.

End note - a minor correction on someone else's comment that Java is also just an interpreted language. Java is not interpreted in the same way as many of the scripting languages we are talking about. Java is compiled to an optimized bytecode which is then in turn interpreted by the JVM (although in most cases large amounts of the bytecode are in turn compiled to native opcodes by either a JIT or Hotspot). This is not the same as a language that is line-interpreted from source. Drastically different.

I agree with James Gosling that he was shorted in the article. When I read the orginal article, I was disappointed with its lack of content. This time I went back with a ruler. The article amounted to only 8.4% of the entire page - 92% of the page was flashing, scrolling, blinking crap. That's a lot to wade through for a disappointing few sentences.

Of what was contained in the orginal article JDJ managed to get at least 17% of it not quite right. The first article quoted Gosling as saying C# was 'hopelessly focused' on one platform when Gosling actually said 'obviously focused'. This might seem trivial, but the misquoted version has potential implications not present in the real version.

Now in this article JDJ focuses on defending themselves.

Engineers want information - not this he-said/she-said soap-opera garbage; we have the National Enquirer for that.

I feel that James Gosling is trying really hard to avoid talking about Ruby and instead attacking PHP, PERL, and C#. He should look more into Ruby.