Slashdot Mirror
Have a Nice Steaming Cup of Java 5
wap writes "The language/VM/religion that everyone loves to hate is now serving another cup: Java 1.5 is ready for download. The new features of 1.5 have been discussed here before. I, for one, welcome our new virtual machine overlord. I have been using the release candidate, and startup times are noticeably faster, as is overall performance, and the new features like typesafe collections and static imports are great to have. Let the Java flames begin!"
Microsoft was right to be afraid, developing in Java is a delight.
Sam
After they took all that time to rewrite it with the latest API they claim they can closely track Sun releases. This will be the first big thing since then, so it will be a test of Apple to get it out quickly.
Brennan Stehling - http://brennan.offwhite.net/blog/
I look longingly at typed collections to save yet another ClassCastException on anonymous iterators. *sigh* oh well, maybe 6 years from now...
[% slash_sig_val.text %]
Well, some may call the 1.5 as being increasingly bloatware, but, why in some aspects this may be true, I think all by all there are considerable improvements over the former releases, especially 1.4.2.
JVM 1.4.2 (at least some sub-versions) were riddled with bugs, which, for instance, become apparent when people use systems that rely on it in a special way, as with Freenet. It comes as no surprise, that there were numerous reports of some errors on OSX and BSD, as well as on linux, when running JVM 1.4.2. For some time, we had to say "If you experience any difficulties, please try/revert to JVM 1.4.1 or 1.5.x and see if that solves the problem."
It is crazy to recommend reverting, but the main devls of java were unwilling to remedy the bugs in 1.4.2, claiming it was a Freenet-problem, while our devls said it was a JVM problem. Though it must be said some within freenet claim their is little to no problem with it (probably windows-users, or maybe some sub-versions that worked on specific linux-distributions). Anyway, my advice has always been, and will be (certainly in the light of the stable 1.5 release), to NOT use the 1.4.2, especially when using OSX or another 'nix based OS.
And also; be sure to get the JRE, and not the full SDK, unless you plan to develop Java software.
--- "To pee or not to pee, that is the question." ---
Actually no, it is an app problem. Sun has introduced weak references way back in 1.3 to cope with that problem. And I have to admit, I never had any of these problems since 1.3
But it also could be that your program simply needs a tad more ram.
Following, check out the -Xmx and -Xms parameters of your application startup file and add more ram (java fetches 32MB in without any params and fills it up before it starts the GC for the first time) That might help.
But never count memory leaks out, they are very rare, but can happen if somebody has tangling references, pretty much the only case where the Garbage Collector can do basically nothing.
The VM itself is not flakey, I have a few servers running here, with an uptime of a year already.
In an interview given just last night, the spec lead for 5.0 is asked what in his view the coolest new feature of the language is. Calvin Austin replies: If I just restrict myself to the language it would be metadata (JSR 175). We've only scratched the surface of its potential. For the platform, it's a bytecode insertion for profiling (JSR 163).
- Generic Types (AKA templates)
- Enumerated Types
- Static Import (usage of this looks quite similar to the #define method)
- Formatted Output / Input (printf/scanf style)
- Varargs
I like this trend. These, among other changes in version 5, are all steps towards reducing the awkwardness in the Java syntax that many people complain about. Java is a young, evolving language that still has a lot of potential.
It is a bit funny, though, that this evolution takes the form of borrowing stuff from an ancient language. Maybe C just got things right in the first place, huh?
/me puts on a flameproof suit
I hear there's rumors on the Slashdots
I work on a minor, highly targetted Linux distribution. I'd love to include Java, and I actually get a lot of requests for it. But, here's an excerpt from the license agreement you'll find if you look to download the software:
(Yes, it really does just end abruptly without finishing the sentence. That trailing "and" there doesn't lead into the next section; it's just not done. Obviously I'm the only one who bothers to read these things -- *including* the people at Sun. Anyway....)
My wish to give the software to my users fails almost every test....
i: I don't want to distribute it for the "sole purpose of running [my] Programs". I want to distribute it so people can run other people's programs, including possibly their own.ii: uh, well, actually, my "Programs" don't add any functionality to Java, let alone "significant and primary".
iii: oops, I include gcc, which has a Java compiler. And I'm definitely not going to drop that.
iv: ahh, now this one I can agree with -- fine keep your copyright notices, etc.
v: my distribution as a whole is under the GPL. I'd have to run this by our lawyers, but this looks like it'd conflict by requring additional restrictions (even if I could get special dispensation to deal with the other issues).
vi: I don't really have the resources to defend and indemnify Sun "from and", even if I wanted to, thanks.
And frankly, that's why I wish people would stop writing things in Java. It's a pain to deal with. I want to make everything as slick, integrated, and as easy as possible for my end users. Sun makes that impossible for Java applications. If you want your code to be easily integrated and made available to users like mine -- and really, that's users of any Linux distro targetted more broadly than the super-geek sector -- please don't use Java. If you must, at least design it to work with gcj instead of Sun's virtual machine.
Unless Sun changes the license terms, their Java can never fill the "write once, run anywhere" goal -- but cleanly written source in an open language can.
As much as I might hate Java, I repeatedly have to throw this one thing into the fray: I had exactly zero former experience with network programming, and still, I was able to produce a basic telnet-like application in under 30 minutes, using only the Java API doc and some logical thinking. And I only had a very brief introduction into the language. Transferring this to C (although the basic structure is exactly the same with BSD and Java sockets) took me about a week, with all those damn low-level error conditions.
Fight hunger. Filet a politician and send him to a 3rd world country of your choice.
I was actually looking forward to templates in Java but was disappointed once I saw the implementation of generics.
Every cast operation in Java is expensive because the VM invokes security code to verify that the cast is legal. Java generics don't allow you to avoid the implicit cast so you still get the performance hit. Generics only make the code look cleaner.
Template meta-programming is also a very important part of the modern C++ libraries and is also something that generics cannot do.
SYS 49152
I was surprised to find that the ObjectOutputStream has a static HandleTable inside it that creates an entry for each HashMap that I put through, and it keeps a reference to each HashMap. I searched around, and this is a common problem that is not mentioned at all in the javadocs. You're supposed to reset the ObjectOutputStream periodically to free up the HandleTable. I had assumed that reset was like InputStream's reset and never would have guessed that it had to do with Object caching in the stream.