Using Java 5 Features in Older JDKs

BlueVoodoo writes "Java 5 added a number of powerful language features: generics, enumerations, annotations, autoboxing, and the enhanced for loop. Even if you're stuck on JDK 1.4, you can still use generics. Use Java and theory to learn how."

Retroweaver

[Anonymous Coward]

Can't get to TFA for some reason, but isn't this the point of Retroweaver [sf.net]?

Re:Retroweaver

[Sciros]

Indeed. It is mentioned at the closing of the article.

Personally I'm a bit frustrated by this being a noteworthy topic... I'm a Java dev and I really wish accomodating pre-java 5 JVMs wasn't ever needed. Reminds me too much of web development.

Re:

[Anonymous Coward]

Uhmmmmmmmmm

If I understand correctly, Generics are backwards compatible because of how they get compiled. As long as your development machine supports generics, you don't have to worry about them on older JVMs!

This is about doing the development on older JVMs. This isn't a backwards compatibility issue in Generics themselves as much as a "If you already have backwards compatibility issues, but want X, here's how to do it."

But I'm inclined to say if you have problems with JVM versions, you're either doing so

Re:

[Sciros]

Class dependencies are different depending on the compiler, so there are a few situations where the JVM executing the code makes a difference. Java 5 isn't that new, so it's not even really bleeding-edge stuff.

Fun but not massively practical

[dintech]

I've worked in a an environment where 1.4 and lower is mandatory. This was down to management believing that 1.5 was too new and 'untested' even though Sun obviously thought it production worthy.

These kind of people would be very scared if you tried to write code using 1.5, recompile the byte code using some new-ish open source toolkits and then putting it in to production. Then again, you could do it without telling anyone. :)

Re:

[Dan Hayes]

You think that sucks, I have to code to the last MS JVM, which was 1.1.8 I think. A significant proportion of our userbase still uses it sadly - people still running Win98 most likely. It's very annoying... although I did manage to find a bug in JRE 6 within a day of its release which broke a small but not-insignificant proportion of websites serving applets, well done Sun for not spotting that one when testing your new caching system.

So to add mouse-wheel support to our applet for those actually using a

Re:

[Nimey]

Are you doing web applets? If not, call me crazy, but couldn't you distribute JRE 1.5.xx on your product's CD? I've seen a few commercial packages that do this.

Re:

[Dan Hayes]

Yeah, it's web applets, probably about as complex as they get - a small stub signed applet which acts as a class loader and downloads and caches class files which require updating from our site (the codebase is several megs), and then a set of applications that run on top of that. Some of the simplest applications are being ported to Flash, but I'd hate to try and do some of the more complex stuff we've got in ActionScript.

Re:

[alan_dershowitz]

Just because it's production-ready for Sun's purposes doesn't mean it's fully tested and supported by your J2EE application server (for example.) If it's not supported, you're on your own if you have a problem. That's why corporations don't switch right away. As far as code is concerned, if it ultimately ends up being targeted for the 1.4 JVM for example, your unit tests should still work, you're not running your server out of spec, and there shouldn't be any problem IMO.

Re:

[tlh1005]

This was down to management believing that 1.5 was too new and 'untested' even though Sun obviously thought it production worthy.

I love Java just as much as the next guy but just because Sun releases it into production doesn't mean there won't be regressions, new bugs, security alerts released saying move up to _xx to fix the problem, and then more of the same is found in the "fixed" release.

These kind of people would be very scared if you tried to write code using 1.5, recompile the byte code using some new-ish open source toolkits and then putting it in to production. Then again, you could do it without telling anyone. :)

I wouldn't want to work in an environment where someone, or some build system WOULDN'T notice this kind of change. Don't get me wrong, I'm one of the guys who wants to upgrade; I just know it needs to be done carefully, and I have also s

Re:

[oSand]

"Then again, you could do it without telling anyone"

Then you will be a convenient scapegoat for any particular calamity that occurs on the project. Performance problems, client goes out of business, server room on fire- you name it.

These have been around for years

[ghoti]

It's great that IBM is writing this, but these tools have been around for years. They basically came out the same time as the final release of Java 1.5 (or "5.0"), when many people realized that it would be tricky to deploy programs that were using the new features. By now, everybody should have updated to 1.5 anyway.

Re:

[Anonymous Coward]

*lol* People are still running production apps on 1.3.

Moving a large infrastructure to a new JDK, while a Really Good Idea, is non-trivial and can be very expensive.

Re:These have been around for years

[hansamurai]

If only everybody has updated to 1.5. The middle tier of our online product where I work is running on 1.4.2 with just a rumbling of upgrading to 1.5. That upgrade probably won't occur for another year for various reasons that I'm not involved in. I think the bigger the company and the bigger the product, the slower the upgrading process is. I think some architects may even be fearful of 1.5, as I just joined a new project that is running on 1.4.2.13, and they started developing that just last November!

Anyways, Java 5 has some great features but nothing that is absolutely required from my department's point of view. Autoboxing is a nice feature that helps clean up your code, but nothing we can't do now. Same for the new for-each loop. I could go on but this has been discussed to death already. I would rather we just upgrade so we can start taking advantage of the new features and supposed speed increases.

Re:

[RevWhite]

I think the bigger the company and the bigger the product, the slower the upgrading process is.

I agree with the former, at least. I work for a fairly large organization where nothing terribly critical or large is running on Java, but they still won't standardize, so some things run JVM, some run 1.3, some 1.4, and several apps need specific patch levels of 1.3 or 1.4.

Re:

[iabervon]

Generics are really the important feature. Replacing "Map" with "Map<String, List<MyDetail>>" is really significant for being able to tell what the code is supposed to be doing and making it work correctly and reliably. For that matter, I had a problem with a commercial J2EE implementation back in the 1.3 days that was returning a Map<String, String> when it was supposed to return a Map<String, String[]>, which was not obvious because it was just returning a Map.

Of course, this depen

Re:

[EsbenMoseHansen]

Generics are really the important feature. Replacing "Map" with "Map<String, List<MyDetail>>" is really significant for being able to tell what the code is supposed to be doing and making it work correctly and reliably.

Of course, this depends on figuring out how all your current code actually works, which can be a major undertaking.

Yep, this is a step in the right direction. Then we just need the rest of the C++ template features, and we might be approaching something that approaches being a nice language. Like not throwing out half the types during runtime... how silly is that? Getting rid of that terrible Collection framework and replacing it with something STL like (improving the STL interface with a Range class) would be a nice bonus. Oh, and I can have a lambda function, too? And operator overloading. And ruby like setters/gett

Re:

[clay_buster]

I've worked on Java teams at two large companies in the last year. Both companies had multiple projects some over 1M lines of Java code. None of the projects were on JDK 1.5 in production. Company 1. Too much code was already in production. Switching to JDK 1.5 involved migrations through all 4 different environments and a full regression test. It's often hard to justify that amount of work to bookkeepers and business users. Company 2. Moving from weblogic to WebSphere using IBM's JDK. Support wasn'

the bleeding obvious... pointed out...

[bitbucketeer]

Java 1.6 is out now. 1.5 is so, like, last year.

Re:the bleeding obvious... pointed out...

[Zonk (troll)]

Java 1.5 was released 09/29/2004, so it's over two years old. 1.6 was release 12/12/2006.

Re:

[Ilgaz]

It doesn't work that way in Java. I am sure there are still people running 1.3.x out there. OS X still doesn't have official (final/ public release) Java 1.6 for instance.

I am sure massively popular apps like Azureus, Limewire are already using such stuff mentioned in article.

Autoboxing

[mccrew]

Autoboxing, a bad idea - automated!

Re:

[nullchar]

What? Not having to: object.toPrimitiveValue() is great! Use the object where you allow NULLs and use the primitive otherwise.

Another option

[malraid]

This is another option used by the people that make GlazedLists, so that they can support Java 5.0 and 1.4 with the same code base: http://publicobject.com/glazedlists/wiki/index.php ?title=Declawer [publicobject.com]

jsr14 compiler target

[Anonymous Coward]

You can use the undocumented, unsupported 'jsr14' target in javac to compile 1.4-compatible bytecode from 1.5 source. No need for Retroweaver or other tools:
javac -source 1.5 -target jsr14

Re:

[caseih]

If you'll read the article, you'll find that this is mentioned. In fact Retroweaver was around before this undocumented switch, and supports the same things plus a few additional features like the for-each loop, autoboxing, string concatenation, and enumerations. So you'll get a bit more mileage out of Retroweaver than just plain old -target jsr14.

cldc1.0

[pjt33]

-target cldc1.0

allows you to produce 1.3-compatible bytecode. It has the CLDC preverifier attributes in, but if you're worried about class file size you can strip them with an optimiser.

Retroweaver has more features than reported

[xlv]

I'm the current maintainer for Retroweaver and the article does not mention all the Retroweaver features:

Annotations are supported, the concurrent backport is used for the concurrent packages, runtime classes can provide support for new features or replace classes entirely, ...

I suppose the article is based on the 1.2.5 version and not the beta version(s). I guess I followed the Google model of having a really long beta cycle with a stable product...

Seeing the possible confusion with the Beta tag, I just decided to release the official 2.0 version earlier today.

Xavier

Is it just me

[zappepcs]

or did anyone else read the title and see "Using Java 5 Features in Older *JoKes*"

Misinformation about Retroweaver

[rreyelts]

Disclosure: I'm the Retroweaver author.

The article seems to miss all of the features that Retroweaver has added over the past year. I think the author may not have been paying attention to the active releases on-going with Retroweaver. For example, Retroweaver supports every feature that the author purports is specific to Retrotranslator.

I have been spending less of my personal time on Retroweaver over the past year, but Xavier Le Vourch [sourceforge.net] has been doing an excellent job improving Retroweaver over that period.

Still stuck with old URLConnection

[Kenja]

Bah, still stuck with the old URLConnection in 1.4.2 that dosn't have a timeout. Ah well, got my hopes up there for a moment. On a sinde note, anyone know how to set the timeout of Sunlabs.brazil's HTTPConnect?

Java on Mac

[tomaasz]

While it's true that most businesses should have upgraded to 1.5 a long time ago, that's just the server side. Most Mac computers still run Java 1.4; I develop Java games for PC and Mac and I need them to run on 1.4.

Retroweaver is a great tool that allows me to use the language features of Java 1.5. I miss the improved 1.5 standard classes, ie. the improved string manipulation (String.format() with printf formatting for instance). At least I get a warning from Retroweaver..

Re:

[xlv]

Basic support for String.format was added recently in retroweaver. If you let me know what you need or submit a patch for enhanced support, we can make sure the features you need are incorporated in future versions of retroweaver.

Xavier