It's amazing that Sun's (or any other non-M$) VM could approach the M$ VM perf considering the proprietary hooks and low-level tricks M$ used to get that kind of performance.
Let's ignore for a moment the fact that Sun's 1.1 JVM had comparable performance with Microsoft's - how do you explain then that the JVMs Sun makes for Linux (open source, no hidden hooks or APIs) and Solaris (their own platform) aren't any faster? How do you explain the fact that IBM has consistently released JVMs that run faster than Sun's? How do you explain bugs like Integer division and modulo operations are 10 times slower on Hotspot?
Anyone out there who developed an application in J++ and discovered to their surprise after it was finished that it only ran on Windows?
You'd be surprised how many "newbie" to "average" level developers don't understand the difference between the standard Java libraries and the additional libraries that usually accompany the "big" IDEs (Visual Cafe, J++, JBuilder). They just think that whatever the class tree shows them, is there to be used. Java newsgroups, message boards and mailing lists are full of questions like "Why does this applet run in Visual Cafe but not in IE or Netscape?" with a stacktrace like:
That is true, Microsoft VM is a very stable (unlike Netscape or Kaffe) Java 1.1 JVM. Another advantage is that it comes with windows - no need for the user to go and install it himself. Could this be considered yet another "bundling with the OS" problem? Would OEM's preinstall Sun's VM if Microsoft didn't provide one?
MS tried to screw people in adding uncompatible java calls (non-JNI) without labelling them properly. They were trying to break Java.
They were trying to break Java, but adding your own libraries is perfectly ok (Sun VMs come with their libraries, which you can use if you want, but of course you lose portability). The problem was that they added nonstandard methods to the standard Java libraries and fooled J++ developers into using them. This made their programs run only on windows.
Another "bad" thing they did was not support RMI or JNI in their JVM and claiming it was completely 1.1 compatible (compatibility with 1.1 requires both RMI and JNI to be implemented). This is what Sun sued them for.
Sun VMs have taken a long time to match MS VMs in perfs.
That is not true. Sun's 1.1 JVM has similar performance to Microsoft VM. The problem was that the 1.2 JVM was much slower on the client side because of Java2D, which significantly slowed rendering. In fact, even the latest (both Sun's and others') VMs are still slower on rendering than the 1.1 JVMs.
Sun hasn't done much good in client-side support. Java applications are memory-hungry and just slow.
Java is somewhat memory hungry, but fast enough for practically any client side application. What is true (and unfortunate) is that Java makes it all too easy to write slow code. If you know what you're doing, you can make Java code only marginally slower than equivalent C/C++ code. And since you only need "fast enough for the user not to notice" code on the client side (unlike "as fast as possible" on the server), Java is actually a very good platform for client side apps.
See a clean, fast Java client for chessclub.com I wrote at http://www.hightemplar.com/jin/. Ironically, I use Microsoft VM by default on windows:-)
Chances are that MS would have done better.
Given Microsoft's record in developing good languages, I find that statement amusing at the very least.
Many people posted saying that since we don't know what AOL/TW's plans are yet, Alan should wait and only when AOL/TW's plans are known make a decision. Don't you people think he understands that? If you give him some credit, you'd realize that being one of the important people at RedHat, he probably knows about AOL/TW's plans a bit more than us.
We should make conclusions from what Alan says, not question his decisions.
I have had the misfortune to have to implement a complete UI toolkit based on the limited capabilities of the JDK 1.1 AWT.
As far as the high level design is concerned, I can recommend using a model similar to Swing, simplifying where possible. Swing, although implemented horribly, has a fairly clean design and it supports practically anything you'd want from a UI toolkit. Here are some: Model View Controller pattern makes it very modular and therefore easily customizeable, pluggable look and feel which can change at runtime, the event model is clean although sometimes requires you to write a lot of classes just to support a single event.
Actually, JDK1.2 was significantly slower than 1.1 on the client side because of Java2D which requires a lot of overhead even when not used.
I'm always up for microsoft bashing, but despite all the horrible things they tried (mostly unsuccessful) to do to Java, they did one good thing - the infamous Microsoft VM (the one comes with IE) is still (4 years later) the fastest one at executing bytecode. Try it, run a simple loop which adds, multiplies and puts the result into an array. Microsoft's VM will do it at least 1.5 faster than Sun's 1.3 Hotspot. Hell, it even makes Swing usable.
For a clean, fast application that runs on microsoft's VM and uses swing, see Jin - a client for a chess server.
Yes, it's old. No, it's not useless, and you can do a LOT in PersonalJava, and I intend to do a *lot*.
I, too, hope that it will be upgradeable to J2ME/CDC/PDA profile when that is released, but you can't sit around waiting for Sun, and PersonalJava is the best you get right now (except for SavaJe XE, but then you lose the OS).
This is exactly the sort of thing I *don't* want an editor to do. This is of course my personal preferences, but I absolutely hate editors which try to guess things for you or do something they think they know how.
Generally, I think software should try to guess as little as possible about what the user wants and just present him with an easy way to tell the software explicitly what he wants.
>computers can only beat the superstar
>grandmasters because of opening books... and
>computers are much worse at blitz chess then
>regular chess...
It's true that computers benefit from vast opening books (they also benefit from vast endgame databases), but actually, they're much better, not worse in blitz than regular chess (relative to humans of course) because in blitz, tactics, the area where computers are best, is a much more important factor than positional (strategical) thinking (the area where humans are best).
Slashdot Mirror
It's amazing that Sun's (or any other non-M$) VM could approach the M$ VM perf considering the proprietary hooks and low-level tricks M$ used to get that kind of performance.
Let's ignore for a moment the fact that Sun's 1.1 JVM had comparable performance with Microsoft's - how do you explain then that the JVMs Sun makes for Linux (open source, no hidden hooks or APIs) and Solaris (their own platform) aren't any faster? How do you explain the fact that IBM has consistently released JVMs that run faster than Sun's? How do you explain bugs like Integer division and modulo operations are 10 times slower on Hotspot?
Anyone out there who developed an application in J++ and discovered to their surprise after it was finished that it only ran on Windows?
You'd be surprised how many "newbie" to "average" level developers don't understand the difference between the standard Java libraries and the additional libraries that usually accompany the "big" IDEs (Visual Cafe, J++, JBuilder). They just think that whatever the class tree shows them, is there to be used. Java newsgroups, message boards and mailing lists are full of questions like "Why does this applet run in Visual Cafe but not in IE or Netscape?" with a stacktrace like:
java.lang.NoClassDefFoundError: com.symantec.vc.SomeNonStandardUsefulClass
MS has offered a decent VM from the start.
That is true, Microsoft VM is a very stable (unlike Netscape or Kaffe) Java 1.1 JVM. Another advantage is that it comes with windows - no need for the user to go and install it himself. Could this be considered yet another "bundling with the OS" problem? Would OEM's preinstall Sun's VM if Microsoft didn't provide one?
MS tried to screw people in adding uncompatible java calls (non-JNI) without labelling them properly. They were trying to break Java.
They were trying to break Java, but adding your own libraries is perfectly ok (Sun VMs come with their libraries, which you can use if you want, but of course you lose portability). The problem was that they added nonstandard methods to the standard Java libraries and fooled J++ developers into using them. This made their programs run only on windows.
Another "bad" thing they did was not support RMI or JNI in their JVM and claiming it was completely 1.1 compatible (compatibility with 1.1 requires both RMI and JNI to be implemented). This is what Sun sued them for.
Sun VMs have taken a long time to match MS VMs in perfs.
That is not true. Sun's 1.1 JVM has similar performance to Microsoft VM. The problem was that the 1.2 JVM was much slower on the client side because of Java2D, which significantly slowed rendering. In fact, even the latest (both Sun's and others') VMs are still slower on rendering than the 1.1 JVMs.
Sun hasn't done much good in client-side support. Java applications are memory-hungry and just slow.
Java is somewhat memory hungry, but fast enough for practically any client side application. What is true (and unfortunate) is that Java makes it all too easy to write slow code. If you know what you're doing, you can make Java code only marginally slower than equivalent C/C++ code. And since you only need "fast enough for the user not to notice" code on the client side (unlike "as fast as possible" on the server), Java is actually a very good platform for client side apps.
See a clean, fast Java client for chessclub.com I wrote at http://www.hightemplar.com/jin/. Ironically, I use Microsoft VM by default on windows :-)
Chances are that MS would have done better.
Given Microsoft's record in developing good languages, I find that statement amusing at the very least.
The new Morpheus Preview edition makes you accept the following license:
GNU GENERAL PUBLIC LICENSE
Version 2, June 1991
..
..
1 and 2 can even be combined.
Many people posted saying that since we don't know what AOL/TW's plans are yet, Alan should wait and only when AOL/TW's plans are known make a decision. Don't you people think he understands that? If you give him some credit, you'd realize that being one of the important people at RedHat, he probably knows about AOL/TW's plans a bit more than us.
We should make conclusions from what Alan says, not question his decisions.
Here are 2 I tried, I liked WebWindow a bit more, but IceSoft's supports more features.
I have had the misfortune to have to implement a complete UI toolkit based on the limited capabilities of the JDK 1.1 AWT.
As far as the high level design is concerned, I can recommend using a model similar to Swing, simplifying where possible. Swing, although implemented horribly, has a fairly clean design and it supports practically anything you'd want from a UI toolkit. Here are some: Model View Controller pattern makes it very modular and therefore easily customizeable, pluggable look and feel which can change at runtime, the event model is clean although sometimes requires you to write a lot of classes just to support a single event.
Read the swing tutorial or the documentation
Actually, JDK1.2 was significantly slower than 1.1 on the client side because of Java2D which requires a lot of overhead even when not used.
I'm always up for microsoft bashing, but despite all the horrible things they tried (mostly unsuccessful) to do to Java, they did one good thing - the infamous Microsoft VM (the one comes with IE) is still (4 years later) the fastest one at executing bytecode. Try it, run a simple loop which adds, multiplies and puts the result into an array. Microsoft's VM will do it at least 1.5 faster than Sun's 1.3 Hotspot. Hell, it even makes Swing usable. For a clean, fast application that runs on microsoft's VM and uses swing, see Jin - a client for a chess server.
Yes, it's old. No, it's not useless, and you can do a LOT in PersonalJava, and I intend to do a *lot*.
I, too, hope that it will be upgradeable to J2ME/CDC/PDA profile when that is released, but you can't sit around waiting for Sun, and PersonalJava is the best you get right now (except for SavaJe XE, but then you lose the OS).
This is exactly the sort of thing I *don't* want an editor to do. This is of course my personal preferences, but I absolutely hate editors which try to guess things for you or do something they think they know how. Generally, I think software should try to guess as little as possible about what the user wants and just present him with an easy way to tell the software explicitly what he wants.
>computers can only beat the superstar
>grandmasters because of opening books... and
>computers are much worse at blitz chess then
>regular chess...
It's true that computers benefit from vast opening books (they also benefit from vast endgame databases), but actually, they're much better, not worse in blitz than regular chess (relative to humans of course) because in blitz, tactics, the area where computers are best, is a much more important factor than positional (strategical) thinking (the area where humans are best).