Java as we know it is over. Time to fork?

Yeah, this is "Armchair forking" -- happens all the time.

As you see in this blogpost it starts with the most important thing -- naming the new fork and then goes into some politics and maybe even release schedule.

Of course, the second the "Post blog" button has been pressed it's all forgotten as nobody wants to actually spend time working on this.

I was watching a video on TED the other day about this phenomenon. If you talk about your goals your brain releases the chemicals that are needed to make you feel that you have actually done something. You "feel good" about doing something even if you haven't actually done it. Having already felt good about your self you have little to no motivation to start doing it.

In contrast, if you don't talk about your goals then you really try to achieve them because you know at the end you will "feel good".

Isn't it generally the case that when people sit down and talk about x, it doesn't happen?

The bar is a lot higher for forking Java. It's not enough for a few guys to establish a track record of consistent high-quality releases. IBM and Red Hat will only give their support to an organization comparable to the one that controls Java now, which requires a lot of coordination and planning, which in turn requires some high-level decisions made inside organizations that have the personnel, money, and interest to make it happen. A single blog post doesn't accomplish much, but this is the kind of pitch that needs to be made in more detail to the right people inside IBM et al.

Isn't this just a special case of "someone oughta do X?" syndrome? If it's not important enough for you personally to make the effort to do it, why do you think anyone else will?

My personal anecdotal experience from using it on Ubuntu is that Iced Tea isn't as reliable or compatible with Java code as the Sun java. Maybe that's changed recently?

It also has multiple backends. I think a debian dev had coded a llvm backend for it. And a cambridge student coded an ARM backend.

Edit : grammar. Also forgot to mention that the ARM backend run on n900, my precious.

If you start with existing GPL code from Oracle, then you have an implicit patent license. But your code has to be under the GPL.

If you meet the "Technology Compatibility Kit", then you're an official Java and have a license. Unfortunately Sun was not in the habit of giving access to the TCK to people who were doing things they did not like with Java, and there is no reason to believe that Oracle will be more generous.

Google's problem was that they didn't want GPLed code (wouldn't work for the vendors who want to customize Android), and Sun wouldn't give them access to the TCK.

So anyone can create a GPLed fork, or someone who already has access to the TCK (eg IBM) can fork as long as they maintain complete compatibility with the TCK.

Still, having a separate markup/scripting language for the UI layout has become an established paradigm for anyone writing apps that need to run in a browser.

side note: this action I guess has prompted the creation of "visage" http://steveonjava.com/accouncing-visage/ as a JavaFX scripting alternative.

I'm actually glad to hear that. I really scratched my head over why on earth Sun invented a new language for scripting UIs even while simultaneously funding things like JRuby, or having options like Groovy or Jython they could have simply run with (albeit, with some slimming down). The mind share they would have gotten by tapping into any one of those communities would have blown the lid off JavaFX adoption. Heck, even just use Javascript / Rhino which already ships with the JVM (which would ironically be a fast way to port Flash / ActionScript developers over). But announcing yet another new language to learn that is not used anywhere else in the universe? Meh.

What's wrong is that Oracle won't (ever) allow Harmony to be called Java Compatible™, even though it is.

There are contributions from many large companies including IBM, Google and Intel.

I wonder how encumbered it is with Oracle contractual restrictions? Didn't IBM at one point license Sun code to build its own version of the JVM?

Lava = Libre Java?

Scala needs the JVM (well, at least it's main implementation does).

Why the downvotes? That's a perfectly reasonable suggestion.

Java is attractive because it's mostly sane, and it's backed by The Enterprise in a big way. Sure, if Google, RedHat, HP and IBM gets behind a fork, it could work - but I can't really see what RH and HP has to gain there. They both need to keep support for Oracle Java to stay relevant. IBMs primary interest in Java is consulting - and the consulting-money going to stay with Oracle Java. Google has a lot of stuff in Java, but they're by no means married to it, except for Android/Dalvik. I'd bet they're more keen to push something like Go (or even Scala for that matter).

But without the enterprise backing, Java is just a pretty dated language. Scala, on the other hand, is a very modern language by most measures, and it will integrate seamlessly with your Java legacy stuff.

Sure, it doesn't solve the problem of depending on the JVM, but if Oracle kills Dalvik, getting a good open source drop-in replacement to the JVM is not going to happen. If not, well, then we have Dalvik :)

On top of that Java has the brand.

Programmers may not feel positive about it but it's embedded in a lot of large organisations, many of whom will, at the management level, be perfectly happy with the Oracle acquisition. They won't see the need or the appeal to move to something else until there is a strong incentive to do so.

Against that sort of entrenchment programmers will struggle to influence things a whole load in a lot of Java's key markets.