Posted by enicholas
on May 26, 2006 at 3:58 PM PDT
In my recent client-side Java rant, I complete forgot about one major problem area: updating Java. Fortunately the Washington Post is here to remind me.
Scott Delap (of ClientJava.com fame) just sent me a gem of an article: a Washington Post blogger complaining about how hard it is to update Java . I have blogged previously about deployment issues, and it remains my single biggest complaint about the Java platform. So here it is from another perspective, a highly detailed and thoroughly sordid look at Java on the client side. The results aren't pretty. If a technically savvy user is having this much trouble just updating Java to the latest version, how are average-Joe users supposed to manage?
He touches on the Java version numbering confusion, which is something near and dear to my heart. Is the latest version of Java 1.5 or Java 5.0? Depends where and how you ask. I've been complaining (some might call it "bitching") about the schizophrenic version numbering since the Java 1.2 / "Java 2 Standard Edition" days. I have brought my complaints up with a number of Sun engineers, and every time I was at least hoping for something along the lines of a defeated sigh followed by "I know, it's stupid. It's marketing's fault." That at least would have made me feel better. Instead I have received a number of patient lectures about how, deep down, it actually does make sense and needs to be this way. Well, I don't agree. It's silly and needs to be fixed. When a question as simple as "What version of Java am I running?" needs an answer that begins with "Well, it depends..." something is deeply wrong. It is not okay for Java to refer to itself 5.0 in some places and 1.5.0_06-b05 in others. Pick one. I don't care which. But pick one and stick with it.
For the rest of the issues, you'll have to go read the original post . It's a good read, although it did sort of make me want to take a shower afterward.
In addition to Brian Krebs' excellent complaints, I have one more to add from the developer side of things. Why is there no easy way to say "I need Java 1.4.2 or higher" and actually get Java installed and updated properly? I want the basic Web Start functionality of updating the Java version based on version selector strings, like "1.4.2+". Except I want it to A) not require me to be using Web Start, B) not require Java to be currently installed at all, and C) function reliably. I basically just want a little utility, perhaps "updatejava", which will check to make sure the right version of Java is installed, download and install it otherwise (or run the installer from a path I specify), set JAVA_HOME to point to it, and return a code indicating whether or not it was successful. Such a utility would simplify all of my Java installers so much it isn't even funny, and I'm sure I'm not alone. Wouldn't it be fantastic to just be able to run "updatejava 1.4.2+", check for a successful return code, and then be able to trust that there was a working 1.4.2 or higher installation of Java pointed to by a correct JAVA_HOME? Does such a utility already exist? If not, how have eleven years gone by with nobody writing one?
(Yes, I know, I should shut up and write one myself... but hey, I'm already occupied . Besides, given the profusion of platforms, Java versions, and issues to work around, this feels to me like the sort of thing that would need Sun support, or at the very least significant resources behind it, in order to succeed.)