Java 7 : More concurrency | Hacker News Reader

Java 7 : More concurrency

baptiste-wicht.com

47 pointsfogus16 comments

Still no composable concurrency primitives. Maybe Java 8 will catch up to GHC from 5 years ago.

For the benefit of the philistines, what composable concurrency primitives does Haskell ca. 2005 have, and which would map well to an imperative OO language?

fwiw - there is the Kilim framework focusing on the actor model. (http://www.ibm.com/developerworks/java/library/j-javadev2-7....)

That looks related to the actor-model for parallel processing.

How so?

No, it looks related to divide and conquer, as the article says.

Duke Nukem Forever: More weapons! More enemies!

You can play with most improvements coming in JDK7 right now ... http://download.java.net/jdk7/binaries/

You can probably find prototype versions of DNF out there too if you look hard enough. I fully realize that "Java is dead" posts are heavy down-mod territory right now because there are so many emotional and financial ties to the platform, but wait a few years.

DNF was never released. Java 7 is being released.

The downmods are probably coming not because people have emotional or financial ties to the platform, but because you made a bad comparison between two unrelated things in a way that adds nothing. In other words, your comment is indistinguishable from trolling.

What do you think the odds are that it will be 2011 and still with no clear concrete release date planed for Java 7? I think very good. The plural of tomorrow is "never", here.

People are optimistic about Java as an evolving unified platform when there is no such indication. The important Java pioneers have abandoned the ship. Google is hedging their bets on Android by divorcing itself from the JVM and the core libraries. History shows that such things (open or not) require strong unified leadership, and Java has none. Even so, what indications of Java 7 being agreed on and adopted are there that should make someone more optimistic than about say Python 3 or Perl 6? The fact is that Java 6 is good enough for the spaces where it is currently used.

Java 7 will have no place in the mobile world. (Though a variant of Java, taken in proprietary directions by Google, might.) In the server space it has lost and is rapidly losing more ground to C# (by nearly all accounts now a superior platform which is evolving to be better), and other languages built on the JVM.

People have saying Java was in severe decline for many years, but never have the signs been so consistent and clearly aligned as they are now.

> People are optimistic about Java as an evolving unified platform when there is no such indication

You're not asking the right people ... http://groups.google.com/group/jvm-languages

> Google is hedging their bets on Android

That's because Dalvik is a register-based VM (versus stack-based) optimized for efficiency on a mobile device that's using solid-state memory (which has sloooow access times). The bytecode itself lends to better interpretation (and it wouldn't be useful on a general purpose PC where you have the resources to JIT).

> what indications of Java 7 being agreed on and adopted are there that should make someone more optimistic than about say Python 3 or Perl 6?

InvokeDynamic.

Dynamic languages on top of the JVM are already fast. Jython / JRuby are already ahead of the reference interpreters in many benchmarks. And in real-world projects it is much nicer to drop to Java for bottlenecks than to C/C++.

In JDK7 InvokeDynamic provides a way to specify a handler for a call-site the JVM doesn't understand (and I think the caching is at least monomorphic). Thus it is possible for dynamic calls (with an unknown interface) to be as fast as virtual methods invocations (blazingly fast on the JVM).

And for JDK6 there's even a backport (using Java agents) for backward compatibility (also with speed improvements).

I haven't seen any real project (or any significant open-source project other than Parrot / Perl6 itself) that's based on Perl 6. And it was started when? 10 years ago?

And many people would argue that Python 3 only brings a couple of syntactic improvements. Only Unladen Swallow makes it exciting for me.

Until then, some reasons why it's not exciting for me ... GIL (no parallel threads), reference-counted (huge memory usage on long running processes), still slow as fuck for expensive computations.

Python is my favorite language, but I'm moving gradually to using Jython more and more.

> losing more ground to C# (by nearly all accounts now a superior platform

No, not really. Both have strengths and weaknesses.

I haven't seen any real project (or any significant open-source project other than Parrot / Perl 6 itself) that's based on Perl 6. And it was started when? 10 years ago?

That sounds a bit like the No True Scotsman fallacy. (What makes a project "real" or "significant"?) Regardless, why should that matter? How many real projects do you know of based on Java 7, or even Python 3? (Python 3 is older than Perl 6.)

Sorry, that came out wrong.

I actually took a look at Parrot ... great tools for language designers, flexible type-system (oh god, the type system in the JVM is horrible).

I really hope you guys can pull it through.

I was judging from my POV ... I want to implement a language than I can use as soon as the parser is finished (I'm not enough of a man to implement my own VM or to deal with non-mainstream choices, although that may prove to be better long-term).

Our biggest lack right now is of documentation for the compiler tools. I hope to address that in the next few months.

> "Java is dead"

Saying that is childish, although I am interested to know why you think so.

IMHO, I can answer why not ... generational state of the art GC, JITting with inlining of virtual methods, meta-circular hosting researchy projects (Jikes RVM, Maxine VM), ecosystem of open-source libraries rivaling only with CPAN, JRuby, Clojure, open-source license.

If your favorite language/platform doesn't have these right now, you can always wait for a few years.

Reply on news.ycombinator.com