Well written, a very nice article!

Whatever forum software you choose, please pick one with the ability to read and post via NNTP and open up that port.
I prefer to read discusions in my newsreader, where I have a nice and usable interface and a good editor instead of this clunky browser based interface.

In case you are still wondering why this is so:

An URL consists of <protocol>://<hostname>/<pathspec>.
An URL of the form file:/// has an empty hostname part, meaning localhost, that's all.

Mee too. The JSDT API was one source of inspiration for my last network layer. Nice too see the inner workings now.

I'm glad I could help  

You might be interested in this /proc related bug, which has been fixed in 1.4.2_04 (and 1.5.0):

http://developer.java.sun.com/developer/bugParade/bugs/4902977.html

for a pre 1.4.2_04 JDK you could mount /proc (possibly with -bind) into your virtual filesystem.

C'mon, even on our 100 mbit network VNC is hardly responsive enough to edit a file in a text editor, whereas with X you can work like it was a local display.

So the only things you can do with VNC are some small admistrative tasks,  chores you'd had done quicker with a secure shell login, if the windows servers had one and were administrable with a text interface.

I cannot understand all this X bashing. I absolutly love it. Even if it's quite a resource hog for running emacs, a browser and some xterms. But hey, memory is cheap and I even boot KDE just to have a pager and the taskbar.

Of course, the client server architecture and network transparency of X comes with a price, but it's the best feature of X. Imagine you are forced to use some crappy GUI like windows at work. Horror! Gone is your productivity! But you just need an Xserver for win32 and a remote login to a unix box to have your familiar desktop and all the apps you are used to back :-)

Because no fancy IDE matches its power up to this very day!
Eclipse with it's open architecture has the potential to come close one day,  but today it's still a very specialized program, constraining its users unecessary and good for nothing but java editing.

You don't need to memorize 300 keystrokes.
Emacs has its documentation built in, to get started you just need one keystroke: M-x  and a few function names like apropos, describe-function describe-variable and describe-mode.
The shortcuts come naturally after a time, Emacs will tell you the key binding if you invoke a function.

Lazy ones can use the menus as well, of course.

I could do an pserver cvs checkout on friday evening. After getting  EOF from the server serveral times, of course, but that's the usual sourceforge problem. ATM I am behind a firewall, so I can't try.

BTW, the public sf cvs servers are always 24 hours behind.
I heard if you if you use the subdomain cvs, eg:
cvs -d :pserver:anonymous@cvs.drts.sourceforge.net:/cvsroot/drts
it works more you have no lag and it works without EOF errors.
I have not tried, though.

With the public sourceforge cvs, one gets your initial cvs import.
I had a quick glance at it over the weekend. Especially at headquarter. Nice work. Some concepts are pretty similar to the game object replication library I wrote at my time at massive development. It was based on UDP, but had a similar property handling. But you have a nice, more generalized approach. I will do some experimentation. The code could surely use some comments at the class level, explaining what each class does. I had a hard time getting an overview at first.

The library I mentioned unfortunally got scrapped before the final release. Despite working well, it was deemed as 'too complicated'. With the bosses, I had the reputation of writing bloated and complicated code because I used templates in C++ and even had heretic ideas of using STL containers instead of  plain arrays, like a game coder should do

That's really great! Thanks Herkules.
I will have a look at it over the weekend.

So it seems there are only two groups of developers: the ones that have found the tool(s) they are comfortable with, and the ones that are still searching.

... or to read java-gaming.org. It renders so bad in emacs-w3m.

GNU Emacs with JDEE[1], Emacs Code Browser[2], Xrefactory[3] and various other modes: Reading Javadoc with emacs-w3m, code-folding with outline-mode and cvs with pcvs-mode, etc.
Of course as well: Mail & News with Gnus  and IRC with erc. No need to leave Emacs, never  

Footnotes:
1. http://jdee.sunsite.dk
2. http://ecb.sf.net
3. http://www.xref-tech.com

Ok, I should have taken a look at the thread title :-)

OK here we go, requirements for running joal on linux:

Apart from joal.jar and libjoal.so you need a native openal library. It must be named libopenal.so and has to be available on the path java looks for shared libraries. This can be accomplished in quite a few ways:

• call java with Djava.library.path=/dir/where./libopenal.so/lives
  
• export LD_LIBRARY_PATH=/dir/where/libopenal.so/lives
  
• copy libopenal.so into $JAVA_HOME/jre/lib/i386
  

Sadly openal for Linux seems to be in a state of constant (but unalien) flux. I had success with a fairly recent cvs version of openal, but the newest one gave me trouble: the sound is distorted.

The version from the gentoo portage ist too old. It does not work.

Regarding the state of he Linux port: It could be improved. ALC doesn't seem to work at the moment and there is no EAX support, afaik.

Sorry Jens, you need of course the native openal.dll in your path (assuming you are on windows). See the project page of joal for instructions how to get it, or go directly to http://openal.org.

I tried it yesterday on my Redhat 9.0 Athlon box, with J2SE 1.4.2_02, SB Live (emu10k driver), and openAL from CVS:

Lesson 1 to 3 ran, after some Linux specific tweaking of the build
files and sources: You need to set java.library.path via a jvmarg tag and the keyboard input loop in Lesson 1 does not work with Linux. Propably some issues with buffered stdin stream, I havn't investigated but simply hardcoded the al* functions calls seperated by Thread.sleep().

Lesson 5 hangs in after failed ALC initialization. Same thing GKW sees.

I have some pictures on the web as well:



2nd one is part of my resume :-)

Some news about Java 1.5 from the inside:

One the concurrency-interest list Doug Lea stated the following in a message he wrote about one week ago:

I'm really looking forward to 1.5. If there now was (good) news about the fate of the isolation API...

Where's the news?  

If I read you right and you want to get rif of the whole subdirectory, just tell everybody to check in, then go to the cvs repository and delete it with 'rm -r'. The people who had this dir checked out need to get a fresh copy of the parent directory, or delete the corresponding line in CVS/Entries manually.

Hi Greg,

look into $JOAL_HOME/src/native/build.xml

Ole

Please feel free to improve on joal. I did the initial port to Linux and I don't think it's beeing worked on at the moment.
All I did was some tweaking of the build files and some fixes in the native part, so it compiled and ran the tests on my Redhat 9 box with OpenAL from the cvs. But I do not have the time currently to do more than that.

Most of the native code is still a bit of a mess. There is at least one memory leak I know of in the initialization code. So your fixes and contributions will be very welcome.

This simple game is fun and reminds me of the times I visited university. We had these Sun Sparcstations and xpilot was very popular...

I had no problems on Linux, btw. Runs smooth as silk.

I think you missed at least some of Williams subtle irony. Who is going to pay the licenses for open source projects?

But the IE/plugin patent case shows another interesting aspect:The crosslicensing practice won't protect the big businesses from small specialized companies whose business case it is to generate patents and then sue the ones they can get the most money from. These shops are not interestet in cross licensing, they only want the money.

This is actually quite a common method to do it.
Google for 'aura nimbus sphere'.

Even that isn't necessarilly true. A (big) company with a large patent portfolio can just take the risk and violate the patent. If the small business dares to sue them, they can always find a patent of their own the small company does violate, so the y will settle with some cross licensing.

In fact, many big companies have just cross-licensed all their patents to the competition. This is very comfortable for them: It gives the big players the necessary freedom in development while keeping new businesses, without an patent portfolio, out of the market.

Patents are nowadays cleary used to erect market barriers, thereby making the free market a little less free. But every company dreams of it's own little monopoly.

When patents where invented sometime in the 18th century, they should originally assure that new inventions will become part of the human heritage. They encourage the inventor to tell it's secret to all, but give him a limited time to exclusively exploit this knowledge in exchange.

If this ever worked is the question, I do not know anyone who ever learned something by reading patent applications. But even if it worked, the problem is the 'limited time'. The timespan of now 20 years is a long time in computing. And this timespan tends to grow. In the related topic of copyrights, the original time the creator of a work retained all rights was 20 years. Now it's 95 years after the death of the creator, at least in the USA, thanks to some lobbying of Disney and MGM.

The main point the patent advocates are missing is IMHO:
The patent law (and this applies to copyright law as well) has  not been made to subsidice some individuals or corporations, however important to society they may be. It has been crafted to assure the knowlegde ends in the public domain, so everyone can benefit from it. The aim is not for the inventor  to monopolize his knowledge, the aim is to increase the common knowledge of our society and to add to the human heritage.

Ahh, ok, mine runs at 24 bits color depth. It's probably the old problem, X reports a depth of 24 but really has a 32 bit framebuffer.

Btw, Markus: Wurm did once run on my Linux box, but now it terminates just after opening it's main window with: "Error starting Wurm Online: Error making context current".
I'm trying to start it in a window with OpenGL check enabled.
I have a Nvidia card with nvidia OpenGL and JDK 1.4.2.

There are a lot of informational files in the /proc directory.
For example try this:

If you have an nvidia card:
Explore the other files in /proc by yourself. You also can see a list of loaded modules (drivers) by calling: