What versions of software are you on?

In a recent thread on TSS, 2 posters (Andrew Lombardi and Augustienje Bloem) got into it a little bit about when things were released, and how long it should be before you think "okay, you're on something too old to really consider any more, update already." It was in context of JSF 2.0, where Mr. Bloem was saying that 2.0 came out in June 2009, people should stop blaming JSF for the problems of 1.0.

Mr. Lombardi had this point:

How many companies still use JDK 1.4?

His implication was "a bunch."

Mr. Bloem responded with:

It's another debate really, but the short answer: surprisingly few. It's a bit of an urban myth that the majority of big enterprises are still on 1.4, while in reality 1.4 has been EOL'ed a LONG time ago and big enterprises don't like running on EOL stuff.

Now, that's something else. Every time I find another person who says they're on 1.4, it's a surprise and I think "that's too many people on 1.4." But it happens - enough that I'm always a little surprised. I don't think there's any reason anyone should ever even be on 1.5 any more, but they are, and for reasons I can't really say are wrong (usually, updates are scary, which for a big company is probably true.)

As for me, I'm on Java 1.6.0_23 (upgraded as soon as it came out, to try to see if Yakov Fain's problem was real or not, and it wasn't, he's an idiot sorry) ; I use Glassfish 3.0.1 as my Java EE container even though I usually just stick to the basic servlet API, I use Maven 3.0 (haven't seen a reason to upgrade to 3.0.1), IDEA X and Eclipse Helios, Hibernate 3.6, and Spring MVC 3.0.4. And a host of other things like it.

The versinos of most of the software isn't important, but the JVM and the Java EE versions are. So I created a survey to try to figure out which versions people use. It's just for information's sake, I'm not selling anything; I'll post the results here on TSS when I get to an interesting threshold, as long as people are interested.

I see there are 2 problems. 

1) Migration projects and their associated costs.

2) Upgrading existing projects from previous java version into newere versions.

We started to do project and delivered on JDK1.4, but when java5 came, when we were interested in making the changes, the client's were retrospective and they were unwilling to move forward. Part of the problem is that you have to be sure to do regresiion testing, application downtime, production movement and there is a huge cost involved in it.

For the newer projects we were able to satisfy the clients the benefits of going forward and we are on java 5 and above but there are ample number of projects where we are still at java 1.4

It's called now Java SE for Business and you can buy support.

My company still uses 1.4.2_19 because it just works.

Unfortunately, at work, we are stuck on Java 1.4/J2EE 1.4 due to the commercial application server we are using.  We use Struts 1.1/1.2, JDBC, and there is lots of scriptlet code inside of JSPs.  There are plans to upgrade to Java 5, but that is still a year or so away.  (I'm translating what is planned as a first half of 2011 initiative into actual time.)  Needless to say, without generics, EJB 3.0+, JPA, or other innovations in Java, and with Java code that was written in the dark ages, as an organization, I estimate we spend probably 100%+ more time fixing bugs and coding than would be required if we were using the newer technologies.  The newer technologies do not excuse the poor coding that was done here (nor will it fix it), but it would definitely be an incentive for those of us working on this code to upgrade and fix what we have.

without ... JPA, ...

You could easily use JDO in that case, and there are implementations runnable with JDK1.4. It's ORM definition was and still is as advanced as anything around

We generally use the highest version of the JDK/JVM supported by all our deployment targets.  In other words, it's often out of the hands of the development staff.  It's a decision made by people who don't know or care about the language features supported by different versions.

While it's certianly not a "majority of big enterprise" that are on Java 1.4 -which nobody is actually claiming, so that's a bit of a strawman argument in the other discussion- the fact that it is still used in quite a few places is obvious to anyone who spends time on Java forums such Sun's/Oracle's and JavaRanch/CodeRanch.

BTW, the link to the other discussion is broken - it's now at https://www.theserverside.com/discussions/thread.tss?thread_id=61492

My company still has a couple of WebSphere 5.1 (gosh!) installations, so JDK1.4 is still being used. However, recently a Tomcat-based (v4.1) application is successfully migrated to Tomcat 5.5 on JDK6. So application server usage is a factor whether migration is straightforward or tricky...and cost too (Tomcat free, WAS not).