Welcome!

Java Authors: Yeshim Deniz, Elizabeth White, Pat Romanski, Liz McMillan, Jason Bloomberg

Related Topics: Java

Java: Article

What Does the Future Hold for the Java Language?

I think that it's possible to be a master of only one language

Before Java I was a Smalltalk guy. I remember switching from one language to the other and the tipping point that you reach when you’ve mastered the new language and how many months it takes, not to mention the years, to do really good design and know-how, which patterns to apply and how to avoid mistakes, understand performance issues, and so forth. I recently had to look at some Smalltalk code and realized that after spending a period away it was hard to figure out what to do – I definitely wouldn’t call myself a competent Smalltalk programmer anymore.

What’s my point? I think that it’s possible to be a master of only one language – either that or a Jack of all languages and a master of none. I like a nice simple flat world with only one language. Communication is easier and everyone can enjoy the shared experience it brings rather than having to constantly switch back and forth. When Java first came out, there was a rearguard action by the virtual machine guys to make Java run on Smalltalk virtual machines that, while they had pretty cool technology called the Universal Virtual Machine (UVM), was probably a sort of denial move to protect Smalltalk’s turf. It’s there in theory so people don’t have to switch syntax, but in practice it was a nightmare to code with JNI to bridge the two and a lot of horrible datatype conversion between primitive language types. To code Smalltalk running under Java or vice-versa you needed to be a master of both languages as well as have a strong head when it came to debugging virtual machine registries and heap stacks. Ugh.

The Microsoft guys, after initially bashing Java for years as being slow because it’s interpreted bytecodes rather than fully compiled, an accusation that’s mostly FUD because of JIT compilers and the fact that most Redmond languages compile to interpreted p-codes anyway, now preach the common language runtime (CLR) as the holy grail of programming. It’s not unlike the Smalltalk/Java hybrid UVM. In fact it’s exactly like it – it just happens to run Microsoft languages on it. It doesn’t have a lot of traction since it came out.

What bothers me now is that there seems to be a resurgence of the idea that virtual machines can do anything. Rather than focus on Java and what the language needs to move it forward, there is a lot of hoopla and fanfare about making JVMs to run Ruby, PHP, or other equally trendy languages, as well as technologies like Java FX, which itself abstracts programming to an even higher and utterly non-Java syntax. If this all occurs, what do we have left? We have a virtual machine that can run Java but can run other languages as well; we have languages that compile to Java but aren’t authored in Java; and we have something that has lost its value proposition and is now all but indistinguishable from its Redmond counterpart. In other words, we’ve lost the plot. For those of us who have to write code, I still maintain that having to be fluent and versant in lots of languages just isn’t that possible in practice and we’ll end up with lots of second-rate programmers writing poorly performing and badly designed code, not to mention the debugging nightmare as context and language switches occur all over the place at runtime. What’s it all for? Because the other languages are dynamic, or because the other languages are better for the web, or whatever? We should be fixing these in Java, not doing VM hacks and increasing complexity just to embrace these other languages that didn’t even exist a few years ago. There is nothing wrong with Java as a language that we can’t fix. I really believe this is where the focus of attention should be, rather than bloat and hack the JVM to become a Jack of all trades that will ultimately make the Java language suffer the same fate the Smalltalk language did.

History has awonderful habit of repeating itself, and if we don’t back Java as a language, rather than some kind of nebulous “Java technology” thing, we’re just dooming it to incognizance by diluting it with other languages and just increasing the entropy required to build good software.

More Stories By Joe Winchester

Joe Winchester, Editor-in-Chief of Java Developer's Journal, was formerly JDJ's longtime Desktop Technologies Editor and is a software developer working on development tools for IBM in Hursley, UK.

Comments (18) View Comments

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.


Most Recent Comments
Baruch Atta 07/25/08 01:20:33 PM EDT

My comment is that good coding shouldn't be cryptic, that is, hard to understand and write. Elegant code should be readable, and in essence, be easy to write. Unfortunately, the Java world is moving in the opposite direction, by adding complexity to the language and it's frameworks. It's a full time job just to stay abreast of the "new" developments in the language. And, for the most part, all this "new" stuff rarely adds anything that is not already available in other languages or frameworks. So, the Comp Sci department should look to the English department, and "simplify, clarify, and condense".

jelly 07/24/08 03:37:33 PM EDT

"It doesn’t have a lot of traction since it came out."

HAHAHA, now who's spreading FUD? Trashing .Net won't save Java, pal.

Ruben Martin 07/24/08 09:19:09 AM EDT

I completely agree with you. The Java language should be the center of the Java community. If we want the language to be more dynamic (whatever that means) we have to work to build that feature into the own language. What's the point in learning a new language over and over again just because it has a new cool feature?

On the other hand, the Java language has to react faster to the changes of the IT community. Maybe that could be achieved by having a faster language extension mechanism.

Good programming is difficult, expensive and requires a lot of experience. We cannot be changing from one language to the other to suffer the same old nightmare everytime.

Rajesh Kumar Raj 07/21/08 02:31:23 AM EDT

Hi All,

I Agree with the author .First we have to understand that here we are discussing about general purpose enterprise programming languages. So here we don't have to care about languages like HTML, XML, JavaScript and these languages are common for both java and .net platforms. In my experience designing and developing enterprise quality code in any platform requires at least 3 to 4 years of experience in that particular platform. So it’s better to stick to one language and invent something in it. Of course I agree that we can able to learn syntax of any language with in few hours. But then why corporate companies are calling for candidates with 4-5 years of experience in java or .net platform???

Tommy 07/19/08 07:23:57 AM EDT

I simply do not agree on many parts:
- .NET has a lot of traction
- you can certainly know well (and master) more than one language. If you cannot master more than one language, this could potentially be one of your limits.
- Java is not a perfect language
- It is easier to move to a new language than to change an existing one. (notice how features are added but rarely removed/changed!)
- What does the future hold for the Java language? You have simply not answered your subject. I think that if java wasn't as popular as it would really struggle. However, given how popular it is I am sure it will survive and get even more bloated.

George Birbilis 07/08/08 05:11:42 PM EDT

CLR doesn't have enough traction? Well, apart from the classic .NET on Windows checkout the "mono" project or Silverlight, XNA (on XBox and Zune) over .NET Compact Framework (also on Windows Mobile) etc.

Thierry Coq 07/03/08 03:36:11 AM EDT

Well, I don't sympathize at all with the author. We've had this junk all before:
- COBOL and FORTRAN started it all... then,
- C was THE language everybody could do things, then,
- Ada was THE language designed to do everything, then
- C++ was THE language for all those gurus outside to do everything with,
- now it's Java,
- and of course, UML is the design language to bring them in ;-)

There a variety of languages because it's needed: languages are designed for different purposes, they are like tools.
It's very common for an application to have 5 or 6 or more languages: UML, Ada, Java, SQL, PHP, XML and HTML, for example. Want to code everything in one language, Object COBOL maybe?

Come on, grow up. Many languages are good for the competition, and they address different needs. And a good designer benefits from knowing different languages and their styles. It broadens the available solutions. And sometimes, what people think good programming other people think bad programming (think pointer arithmetic in C/C++, for example!)

Best regards

Neil Crow 07/01/08 05:10:43 PM EDT

Interesting thoughts, I was contemplating similar points earlier on my way home from work. But I can't say that I reached a similar conclusion, the crux against the argument is that the language is a very difficult thing to change, adding API's is trivial in comparison.

One of the biggest gaping holes at the moment is the lack of first class closures in java.
This gap has been very neatly solved by the groovy guys (http://groovy.codehaus.org/), while managing to keep the syntax freindly to java developer types. Another of groovy's selling points is the immediate feedback obtained from running interpreted code; hot-code changes are simple to do in groovy when compared with java, although both are running on the JVM. This is something that I hear ex-smalltalkers complain about a lot, the tedious java deployment cycle, especially when developing webapps. I haven't seen a successful hot code deployment in a java IDE for quite a few years now. And if it does work, then you can be sure that the web-framework and the classloaders are going to trip over something that was cached.

If the development community were to wait for the language and the JVM to change to support this, that would be a very long wait.

Lastly, I agree with Craig that the framework bloat is a problem. This affects project ramp up time, while halfway though choosing your technology stack, the .net projects are already developing using what they have got because they've got less to choose from.
However if it weren't for this type of innovation, then we may still be left with EJB2, which when compared to Spring was butt-ugly. The two have now joined forces added annotations and provided something even better.

I would like to see java improved, but I shudder to imagine a world where the "CORE" controls the kinds of innovations that are allowed. Frankly I don't think that it can be improved fast enough to keep up with the demands of the development community; by its nature, the core language has to be behind the curve of innovation.

Cheers,
Neil.

Jason Rogers 07/01/08 11:26:21 AM EDT

What?!?! Structured Query LANGUAGE isn't a language? I agree markups aren't languages per se, but SQL?

FWIW: SQL _can_ loop, iterate, branch, etc. I wouldn't suggest that interfacing with the OS is a necessity of a language -- the toolkits on top of the language are a must, but not the language itself.

You're still off a bit -- but perhaps it's time to agree to disagree. :)

Craig 06/30/08 11:08:38 PM EDT

Joe, I agree with you -- and to make a parallel point, we're also getting overloaded with frameworks within Java. Dozens of them. This is where the .NET folks will really come in for the kill, because developers won't have to learn so many different, competing technologies to get their work done...

Joe Winchester 06/29/08 05:26:44 PM EDT

Wow, I had no idea that my piece would attract flame from the "Microsoft CLR Rules" guys. Glad to know you're still out there, take care.
However, I do agree with most of the feedback that I sort of missed making the point I was trying to, which is basically that Java as a language should be grown to do whatever is missing, rather than the JVM become a platform for lots of other languages that overlap. Sure, things like SQL and HTML and CSS and other stuff are great, I use them all the time, however folks they're not languages. Languages do things like iterate and loop and branch and interface with an operating system and do network I/O so forth. I think the only real split that should exist is whether you want a language that can manipulate pointers or one that can't. If you don't then there should be nothing between the rest, and I believe that the focus of the JCP and Java stewards and so forth should be to make Java do all that is missing and that other languages provide.

Jason Rogers 06/27/08 09:24:49 AM EDT

I see you've already been bashed here a lot, so I won't take too much time to do it again. However, I will say that the JVM-as-a-platform is not about being able to program in Java and some other language. It's about being able to program in some other (better) language and just leverage the work done in the JVM (as well as interoperate with common Java libraries like JDBC). I program in Java, Smalltalk and Ruby. I use the JVM for Java and Ruby. When I write in Java I don't have to debug JVM stack frames. Similarly, when I write in Ruby and deploy on the JVM I don't have to debug JVM stack frames. What in the world are you talking about?!?!

Cjeo 06/27/08 09:07:47 AM EDT

Try
I am fluent speaking English, Italian, Spanish and French already...is that imposible with computer languages? I do not think so.

Catch
As a newbie, on Java, your article scares me.

Finally
...you do not provide any info on the subject "What Does the Future Hold for the Java Language?"

jack g 06/27/08 07:33:54 AM EDT

Dude your totaly wrong!!! .Net has no traction, your an idiot! You should go back to school, by the way I was a smalltalk guy and I can work in 80' or digi or whatever no problem, I am still competent, WHY ARE YOU WRITING ARTICLES?

JulesLt 06/27/08 04:16:32 AM EDT

Well, some would say we do have plenty of second-rate developers developing systems without concern for performance already . . .

With my managerial hat on I really can see the advantages of a flat landscape - it makes people far more interchangeable between projects.

On the other hand, while I can see the advantages of having a lingua franca, I've never been convinced in the idea that Java is also the best language, or that the process around it selects the best options - i.e. EJBs, JDBC. There is also the fact that in a lot of cases to achieve the dynamic behaviour that is native to other languages, Java programmers typically have to resort to frameworks that utilise XML config files - for instance using Spring for dependency injection.

So you have this nice safe language and compiler - but an awful lot of your logic is in uncompiled text, from Struts action or Hibernate mappings to SQL statements in JDBC (see SQLJ or LINQ as alternative approaches), and only testable at runtime - just like dynamic languages.

Laurent Hasson 06/26/08 11:30:08 PM EDT

I agree that the focus should be on Java as a language and not Java as a general technology stack (it's very important, but shouldn't be the focus today in my opinion). However, i am surprised by your statement that good developers are generally fluent in a single language, or that it's only possible to be fluent in only one language. I and many others in my organization spend their days in 4 languages: Java, SQL, JavaScript and HTML. That's not counting CSS, XML and XSLT which i think would be a stretch to call languages, at least the way we use them, but they do require a brain context switch.

The key issue i think from your article is not about multiple languages from a syntax point of view, but from a conceptual point of view. One thing that the CLR has done well i think is to define clear conceptual language features right in the VM level. Lots of the CLR languages to feel very close and share a common conceptual base to some level. For Java however, the bytecode is more like assembly and quite low level, which to some extend may be a reason why some new language features are so hard to implement (i'll never argue that C++ templates are the most beautiful things on this earth, but Java Generics are a joke).

To put that in some example, having to write an application in Java that calls to a C++ DLL via JNI is a nightmare... But writing a Java app that calls a C++ engine through a web services stack for example is my bread and butter. So clearly, to me, the problem is not that within a day i need my brain to go back and forth between C++ and Java for example.

Dino Chiesa 06/20/08 08:24:47 PM EDT

Joe!

It must be fun to make stuff up all the time! You can make up words like Versant, and Incognizance.

And you can make up facts, too!
In describing the Microsoft CLR, you compared it to the hybrid UVM that hosted Java and Smalltalk: "In fact it’s exactly like [the UVM] – it just happens to run Microsoft languages on it. It doesn’t have a lot of traction since it came out."

huh?

Let's take the made up stuff one by one.
First, the No traction comment.
In a Forrester survey, 56% of corporate developers said they developed for .NET. That would imply the CLR. That smells like traction to me?

Ok, now the "Exactly like the UVM" comment.
I see what you mean - the similarity is that both the UVM and the CLR run multiple languages.

But, the CLR differs from the UVM in that it was designed for multi-language support from the outset. Neither the Smalltalk VM nor the JVM share that design point. In this regard, the CLR and its bytecode ("Intermediate language") are completely UNLIKE the JVM and the Smalltalk VM which became the UVM. This multi-language capability is what enables IronRuby, F#, VB.NET, C#, COBOL, and even languages like PHP to run on the CLR.

Last thing:
I don't remember the "Microsoft guys" bashing the JVM for years because it was interpreted. That would be a bit of a those-who-live-in-glass-houses thing, since VB at the time was an interpreted environment. You've heard of VB? It was the #1 language at the time, so to hear Microsoft saying "interpreted is bad" would be silly. Maybe you are thinking about C++ language zealots who decried interpreted environments? That I remember. Regardless, the fact is that Microsoft sold the most popular Java developer tool in the industry, Visual J++, as late as 1999. This was not "bashing" - this was endorsement and investment in Java and the JVM.

But hey, if you're going to make stuff up, you may as well "Go Big", right?

The puzzling part is that none of those made-up facts are supportive to your main point, which is that the JVM needs to focus on Java. That's a fair position, and worth consideration. I can imagine thoughtful discourse on either side of the issue. A rational examination of risks and benefits of such a strategy would be really valuable. But I didn't see that in your piece...

-Dino

Vijay 06/20/08 05:34:36 PM EDT

Well written article!!