Works for me.

Cas

See i have a question about that,  eh sick and it's hard to keep my thoughts straight so please bare(?) with me..

i was hoping to make small java games in the future, in the form of applets and maybe expand to phones after that or something :/ well i have a rough idea, but are applets really the way to go? I keep hearing about something called webstart? if i got that right. I'd imagine people would like applets more since it doesn't involve downloading something.

What performance Handicap?

Okay, there is an issue for cases when optimized MMX/SSE/SSE2 etc. is the right thing to do.  Like with many signal processing algorithms.  But for general code there is usually no performance issue.  At my company we are moving more and more towards Java code for as much as we can - and we are in the (video) signal processing business  ;-)
We just use native code where it matters.  For games the core performance areas are usually handled by hardware via OpenGL, so native code isn't needed *at all*, or such native code is already in a library like LWJGL.  Coding the bulk of your stuff in Java will just be easier for other reasons (much better support in tools, better standard libraries, etc.)

The Java language may not always be the way to go...  Scala or something may be better depending on the application, but running on the JVM, using the Java Platform is still a reasonable choice.

Bear in mind that the Unreal engine and similar engines are for triple-A games. Part of the reason they use C++ is because Java is only very fast in recent years. Back when Unreal Engine 1 was created C++ was clearly the best choice. If you have a very large set of libraries and ecosystem built around one language in mind, then it makes no sense at all to switch to Java. For example John Carmack was once asked about the idea of building Quake 3 in Java and he said it would be perfectly possible, but what would be the point?

Secondly if you want to build your own low-level locking systems (which the Source engine does) then again C or C++ is the best option. The Linux kernel still advocates using C over C++ because of the extra cruft C++ compilers generate. C gives them ultimate control. So if you want to be a part of building the very extreme and low-level parts of a game engine, such as writing lock-free algorithms and advanced smoke systems using the lattest in DX11 tesselation, then yes Java is a bad choice. Otherwise it's fine and for the web Java applets are still FAAAAAAR faster then anything else.

The main issue I have with Java is that when you compare it to Silverlight, Flash, JavaScript or XNA then there is more you need to do. The ecosystem is much smaller (less forums to join and less tutorials for you to follow). Secondly you also have to write more of the cruft yourself. An example I often give is if your user doesn't have the runtimes you use, with Flash this is already solved for you (they are provided with an update or told that they need Flash). With Java there are solutions (Oracle provides a JavaScript script to help) but you have to research and bind this together yourself.

But ultimately it's all about the game. An amazing game is amazing, a bad game is bad. That has nothing to do with the language.

Also, Java will earn you a mint in the general programming industry

Cas

Mint = Shedloads of dosh!

As a New Zealander, I am surprised by how much we copy of the English rather than the Americans when it comes to slang.

The slang "make a mint" comes from the fact that we call the place that prints money (or stamps too) the mint. ie the Royal Mint.

Use Java, always. It can be even 15% faster than C++  Update your knowledge, forget your prejudices about Java, there are enough Java haters in this world.

Java performance handicap is essentially a prejudice like C++ superiority in term of performance. Memory allocation and method calls are faster in Java than in C++ (according to Brian Goetz, author of "Java Concurrency in Practice") and dynamic optimizations allow optimizations at runtime not possible in compiled languages (according to IBM researchers on the HPC project). If you really want to do the next big game, Java can do the job, look at Poisonville.

I didn't know that, it is a very bad piece of news

OMG everyone need to look at poisonville. Here is a good video : http://mmohuts.com/browser-games/poisonville

Yeah I read it... It was kinda funny when the person testing the game was not able to enter in the police car. Seriously, I don't know why they feel it was necessary to tell everyone it was made in Java. I have no idea in what language the other game I play are made and it is never mentionned built with c++ in the review I read.

Well you are probably right kappa that the game was just bad managed. They failed to do a ''simple'' fps (so many glitch, gameplay buggy) but they tried to make a MMOFPS. It's really stupid. It was more to show that that I post the video. The game seems not bad overall and if it was made by a couple of indie it would be great but spending 2 million on that...

All that theory and research is nice but then you go and open NetBeans, JEdit, or one of the other many Java built applications and you find quite a different reality.

When you look at the non-Java versions of those applications (such as Visual Studio instead of NetBeans and Eclipse, and Notepad++ instead of JEdit) then you find them more responsive and use less memory. The same is also for plenty of other Java apps I've used over the years.

It does not prove that Java is slow. Anyone can find some counter-examples. I disagree with you about Visual Studio, I used it only a few weeks, it doesn't have as much features as Eclipse and I found VS less stable.