We should make one language extension to end all language extensions. That is a sublanguage feature, a syntactic contruction to let you embed regions of other languages into java and specify a compiler extension and an editor syntax coloring extension, debug facilities extensions, code completion extensions that seamlessly let you develop in the sublanguage. Examples of possible sublanguages: multiline strings, jsp, new loop contructs, domain specific languages, JavaFX, ... some are very simple some extremely complex. When the competing language Cx develops a new language, the java community could develop a similar extension and developer tool support within weeks, because the extension is optional and severel versions could coexist there would be no need to take it through a commttee. Think of what the simple language extension mecanism annotations have done. What would be possible with this hyper generic extension mecanism?

should be avoided. what's the point of adding new language (not library) feature when the previous one has still its, let's say, side effects? it will just add more enthropy. but if those of you that are using java for living feel that closures are the right thing, then i would go the joshua way.

I think java should follow direction of increased productivity but without changing to scripting like language. If we have generics in java 1.0 there could be generics that behave always predictably without interfering with non generics old code. I have some doubts on closures, maybe because of my little knowledge in that but it they can increase productivity without sacrificing validation and readability of code I am on it.

I agree that "frozen" is too hard for this poll. I'm not in favour of "freezing" in the sense that anyhting is changed, but I only hope in minor changes. For what concerns generics, I agree that they are indeed a good addition to the language, that can lead to more robust constructs and for some aspects is more readable (returning a Collection is more self-documenting than returning a Collection). It's just that they could have been designed in a simpler fashion, as Howard Lovatt recently suggested at Artima. Sun probably tried to do "too much" with them, and this is precisely what I fear from possible further additions. It's better to have minor effective changes, for most dramatic improvements go with a brand new language.

Is this only about the syntax of the Java language or the APIs, etc. as well? It is not very obvious...

I also think that "freeze" is quite a hard word - but I would like Sun to avoid any large, invasive changes like Generics. Who cares about Strings in switch statements or synthatik sugar like that, but e.g. I would avoid closures completly.

Freeze is too strong of a word. But no invasive changes that aren't really compelling. Some reasons for change: better support for parallelism. But in general adding things just because other languages have them is dumb. Generics added little value for example. I fear closures will add little value as well.. but maybe they're but syntactic sugar to anonymous classes?

...anyone that's spent time chasing runtime exceptions (class casts et al) rather than use a type safe construct knows the value of generics. The beauty of them is that you pick when to use them, you're not forced to. I think that they are readable much more so than looking at some non-typed list and having to "guess" at what can go into the list without causing problems. Simplicity comes at a cost, maintainability in many cases. As for closures, why?

IMO, the poll needs to split the 'no' into two versions:
-No, but take greater care in adding new features -No, let it evolve as much as possible.

I would like the language not to be frozen, but i think that we need to think more on the useability of the language.
We need to keep java being a clean language, easy to read from the blink of an eye by any java-able person.
Java learned from the mistakes of the others (and missed some, but this is an other story). Let's follow this road, looking at how those "so necessary" features go in other languages then adapt, if they are successful in ... 4 or 5 years...
In the meantime, not only everyone will be prepared to the ideas developed (how many persons really understand closures, in all their forms? come on ) but eventually much better solutions will come out which will please everyone. Eventually those ideas will grow from discussion and/or from analyzing other languages failures and successes.
Too rapid evolution is something .net people bitch about. Great things take time.

One thing i learned from my mistakes is:
"better is the ennemy of good"

The main evolution of Java should be in the APIs and the platform (e.g.. multimedia, platform integration, mathematics APIs ...) . In the language itself should be relatively few, wisely chosen changes and they should focus on simplifications (and removing idiosyncrasies from generics). And the additions to the core language should always have a good sense of proportion between the gain of power and the complexity. The BGGA closures especially the control invocation syntax is a good example how not to do it. I am for simplifying the usage of inner classes like the CICE proposal. It gives you most of the benefits of closures at reasonable complexity

Seriously, the only new language feature in Java 5 that is useful is enums. Generics are a disaster. The idea was a mess in C++, and a mess in C#. What made anyone think that Java would be different? The "enhanced" for loop and autoboxing serve ... what purpose, exactly? So you can save ten keystrokes when you build a loop? Hint: It doesn't hurt to type with both hands, people.