JOGL 1.1.1 released

[wiederke]

Hi,

in the previous jogl versions the GLEventListener.reshape() method was called, if the display mode changed. I used this event to reconfigure glu.gluPerspective() accordingly. Since jogl 1.1 b09 no event is fired, neither reshape() nor displayChanged(). What is the reason for this change?


Best regards

Torsten

[Ken Russell]

in the previous jogl versions the GLEventListener.reshape() method was called, if the display mode changed. I used this event to reconfigure glu.gluPerspective() accordingly. Since jogl 1.1 b09 no event is fired, neither reshape() nor displayChanged(). What is the reason for this change?

How are you changing the display mode, using the Java fullscreen APIs or in the Control Panel (assuming Windows)? I don't remember changing anything in JOGL that would have affected this. Did you also upgrade your JDK or switch your setting of -Dsun.java2d.noddraw=true?

There are problems with the Java fullscreen APIs on Windows (when -Dsun.java2d.noddraw=true is set, which it needs to be for JOGL compatibility) and Mac OS X where the window and GLCanvas don't get resized properly. This can be worked around by putting this at the top of your GLEventListener.init() method:

1   2   3   4   5   6   7   8   9   10   11   12   13

public void init(GLDrawable drawable) {   // FIXME: workaround for Windows full-screen bug when  // sun.java2d.noddraw=true and similar bug on Mac OS X  if (fullScreen) {     final GLDrawable tmpDrawable = drawable;     EventQueue.invokeLater(new Runnable() {         public void run() {           frame.setBounds(0, 0, initWidth, initHeight);           tmpDrawable.setSize(initWidth, initHeight);           frame.toFront();         }       });   }

The only thing that really appears to be needed out of those three lines is the call to tmpDrawable.setSize().

[kati]

What OS and graphics card are you using?

Linux, Nvidia Geforce 2, not the most recent driver (03.2004)

Maybe, I found out the problem. It crashes when I start the animator again, although it is already working. But when I start the animator again after the rendering was done at least one time, no crash appears. (I stop the animator when the mouse is leaving the panel and start it when the mouse is entering. I have to start the animator for the first rendering to see anything and when the mouse is outside this area and enters, the animator is started a second time and the api crashed, when no display was finished in between.)

Furthermore it crashes when I stop the animator and some milliseconds later I start it again and when I call the method gl.glSelectBuffer(bufferSize, pickBuffer) in mouseMoved(). I don't know why!!!!!

Games published by our own members! Check 'em out!

[wiederke]

Hi Ken,

removing -Dsun.java2d.noddraw=true as VM argument fixed the problem.

Thx for help.

[Ken Russell]

Linux, Nvidia Geforce 2, not the most recent driver (03.2004)

Maybe, I found out the problem. It crashes when I start the animator again, although it is already working. But when I start the animator again after the rendering was done at least one time, no crash appears. (I stop the animator when the mouse is leaving the panel and start it when the mouse is entering. I have to start the animator for the first rendering to see anything and when the mouse is outside this area and enters, the animator is started a second time and the api crashed, when no display was finished in between.)

Furthermore it crashes when I stop the animator and some milliseconds later I start it again and when I call the method gl.glSelectBuffer(bufferSize, pickBuffer) in mouseMoved(). I don't know why!!!!!

It is only valid to make OpenGL calls from your GLEventListener's callbacks. This is probably the root cause of your problem. You need to buffer up mouse motion and other events and process them in your next display() callback. Additionally, you probably shouldn't need to start and stop the Animator as often as you are. Calling repaint() or just letting the AWT / Swing framework paint your GLCanvas or GLJPanel should work fine for most interactive applications.

[c_lilian]

Hello,

I'm having 2 problems with JOGL 1.1b8 and b9 on my PC :

1) Running a personnal jogl app from command line, I have a JVM crash on exit (I can live with that) :

#
# An unexpected error has been detected by HotSpot Virtual Machine:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x695838c4, pid=3236, tid=2588
#
# Java VM: Java HotSpot(TM) Client VM (1.5.0-b64 mixed mode, sharing)
# Problematic frame:
# C  [nvoglnt.dll+0x838c4]
#
# An error report file with more information is saved as hs_err_pid3236.log
#
# If you would like to submit a bug report, please visit:
#   http://java.sun.com/webapps/bugreport/crash.jsp
#

Extracted form hs_errXX_pid.log
[error occurred during error reporting, step 270, id 0xc0000005]
Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j  net.java.games.jogl.impl.windows.WGL.wglDeleteContext(J)Z+0
j  net.java.games.jogl.impl.windows.WindowsGLContext.destroyImpl()V+13
j  net.java.games.jogl.impl.GLContext.destroy()V+33
j  net.java.games.jogl.GLCanvas.removeNotify()V+4
j  java.awt.Container.removeNotify()V+38
j  java.awt.Container.removeNotify()V+38
j  java.awt.Frame.removeNotify()V+51
j  java.awt.Window$1DisposeAction.run()V+105
j  java.awt.Window.doDispose()V+16
j  java.awt.Window.dispose()V+1
j  shoot.ShootWindow.backToSystem()V+38
j  shoot.ShootCanvasControl$4.run()V+7
j  java.awt.event.InvocationEvent.dispatch()V+47
j  java.awt.EventQueue.dispatchEvent(Ljava/awt/AWTEvent;)V+26
j  java.awt.EventDispatchThread.pumpOneEventForHierarchy(ILjava/awt/Component;)Z+200
j  java.awt.EventDispatchThread.pumpEventsForHierarchy(ILjava/awt/Conditional;Ljava/awt/Component;)V+26
j  java.awt.EventDispatchThread.pumpEvents(ILjava/awt/Conditional;)V+4
j  java.awt.EventDispatchThread.pumpEvents(Ljava/awt/Conditional;)V+3
j  java.awt.EventDispatchThread.run()V+9
v  ~StubRoutines::call_stub



2) And most problematic, running (fine) the jogl demos makes my PC reboot on exit ...


Configuration :

Pentium -IV, Windows XP, nvidia 6600, latest drivers.
Tried with jogl 1.1.b8 and b9.

Tried with -dsun.java2d.noddraw=true or false (same results).

Any idea ?

Lilian

[kati]

It is only valid to make OpenGL calls from your GLEventListener's callbacks.

Hello Ken,

thanks a lot. Now I understand. It is not allowed to call any OpenGL method outside from display, init, ....
I changed my code according to this hind and now no crash at the moment, but there are still some problems with the GLJPanel. When I use this panel, text and buttons in JDialogs are not shown.

Kati

[Ken Russell]

1) Running a personnal jogl app from command line, I have a JVM crash on exit (I can live with that) :

Does this happen with the JOGL demos or only your application? If the JOGL demos terminate cleanly, look at how the runExit() method is implemented. It starts up a new thread to shut down the Animator and call System.exit().

Also, please run with the system property -DJOGL_SINGLE_THREADED_WORKAROUND=true and post again indicating whether this fixes the behavior.

2) And most problematic, running (fine) the jogl demos makes my PC reboot on exit ...

This has come up before with 6800 cards but I haven't been able to reproduce it in house. Could you please run with -DJOGL_SINGLE_THREADED_WORKAROUND=true and b09 and tell us whether that works around the problem?

[Ken Russell]

I changed my code according to this hind and now no crash at the moment, but there are still some problems with the GLJPanel. When I use this panel, text and buttons in JDialogs are not shown.

You probably need to specify the system property -Dsun.java2d.noddraw=true . Due to the fact that Java2D uses DirectDraw internally on Windows and that there are driver-level incompatibilities between OpenGL and DirectDraw you basically need to specify that system property for all JOGL applications.

[c_lilian]

Does this happen with the JOGL demos or only your application? If the JOGL demos terminate cleanly, look at how the runExit() method is implemented. It starts up a new thread to shut down the Animator and call System.exit().

Also, please run with the system property -DJOGL_SINGLE_THREADED_WORKAROUND=true and post again indicating whether this fixes the behavior.


This has come up before with 6800 cards but I haven't been able to reproduce it in house. Could you please run with -DJOGL_SINGLE_THREADED_WORKAROUND=true and b09 and tell us whether that works around the problem?

Well, I've changed my app, and now my PC crashes as well... I'll try to isolate where the problem comes from (as i prefer a jvm death instead of a pc reboot ;-) )

I've also downloaded the jogl demos, and ran them from command line (tried only Gears). It crashes my PC either with or without the -DJOGL_... arg.

Seems to be a rather serious problem...

Lilian

EDITED :

- I have the same behaviour in full screen exclusive and windowed mode.

- My computer does'nt reboot if I dispose() the frame and System.exit() immediately after (but JVM dies), withour removing the glcanvas.

- It reboots immedialtely if I "frame.remove(canvas)" of if I "sleep(3000);System.exit()" after dispose() , in another thread.

Hope it helps.

Games published by our own members! Check 'em out!

[Ken Russell]

I've also downloaded the jogl demos, and ran them from command line (tried only Gears). It crashes my PC either with or without the -DJOGL_... arg.

Seems to be a rather serious problem...

I completely agree that this is a serious problem. We were able to reproduce the crash upon exit of the Gears demo on a colleague's Windows machine with a 6800 card. Specifying -DJOGL_SINGLE_THREADED_WORKAROUND=true works around the driver bug. I think you may have misspelled it. Try it again and specify -Djogl.verbose as well; you should see a confirmation message printed. ("Using single-threaded workaround of dispatching display() on event thread")

Based on this and a couple of other bugs on Linux and Mac OS X that are worked around with this flag, we're going to make it the default in the next JOGL release. We'll continue to push the various graphics card vendors to improve the multithreading support in their OpenGL drivers. Based on my tests, I don't think that there will be a significant performance impact from changing the default value of this flag, but even if there are, we believe that stability must be the first concern.

[c_lilian]

You were right, I misspelled the -DJOGL.... now It doesn't crash my PC anymore (great !).

I still have a problem with my app (JVM death), but it is not really important and i may have something wrong with my threads (private testing stuff).  

By the way, is there a preferred thread to create the glcanvas ? (I think i've done it on the event dispatch thread).

Many thanks Ken.

Lilian

[Ken Russell]

I still have a problem with my app (JVM death), but it is not really important and i may have something wrong with my threads (private testing stuff).

Check to make sure you aren't trying to perform OpenGL rendering at the time your application exits. The JOGL demos stop any Animators before exiting for this purpose.

By the way, is there a preferred thread to create the glcanvas ? (I think i've done it on the event dispatch thread).

No, any thread should work, although you should follow the AWT's rules. Specifically, once your component hierarchy is realized, the only modifications to it should be performed from the event dispatch thread.

[Ken Russell]

JOGL 1.1 b10 has been released on February 27, 2005. Please see the following thread for details:

JOGL release information

Please post comments about the new release here.

[rexguo]

Hi Ken,

I noticed a small increase in the
file size of 'jogl_cg.dll' and wonder
what new goodies went into it since
there's no mention of it in the release
info? Thanks!

.rex

[Ken Russell]

I forgot to mention that because of a bug report from one of the members of Agency9, the JOGL Cg binding now has NIO entry points along with the entry points taking Java arrays.

[Markus_Persson]

I'm still getting a weird error when trying to sign jogl.jar

1   2

C:\dev\wurm\lib_bin>jarsigner -keystore mykeystore -storepass ***** -keypass ***** jogl.jar mycert jarsigner: unable to sign jar: java.util.zip.ZipException: invalid entry compressed size (expected 11218 but got 11181 bytes)

Manually rejaring the file makes the problem go away.

[Ken Russell]

I'm still getting a weird error when trying to sign jogl.jar

1   2

C:\dev\wurm\lib_bin>jarsigner -keystore mykeystore -storepass ***** -keypass ***** jogl.jar mycert jarsigner: unable to sign jar: java.util.zip.ZipException: invalid entry compressed size (expected 11218 but got 11181 bytes)

Manually rejaring the file makes the problem go away.

This appears to be a bug in the tool we are using to sign the jogl.jar file, which isn't jarsigner but an old Netscape tool. I think there are alternative mechanisms available to us to store our certificates and sign our jars at this point and I'll try to change over to those. In the meantime please file a bug on this so we can track it. Thanks.

[Ken Russell]

JOGL 1.1 b11 has been released on May 11, 2005. Please see the following thread for details:

JOGL release information

Please post comments about the new release here.

[Ken Russell]

JOGL 1.1 b12 has been released on May 27, 2005. Please see the following thread for details:

JOGL release information

Please post comments about the new release here.

[Rob Nugent]

Ken,

Good stuff - thanks!

I suppose I'm going to have to buy an fully OGL 2.0 compliant card now...

Rob

[Wizumwalt]

I'm trying to get the 1.1b12 running so I can play with the gleem package and try again to see if I can get some nice zooming, panning, and rotating w/ the mouse implemented (which btw, anyone happen to know if gleem now works w/ orthographic projections)?

Anyways, when I start demos.gears.Gears, I get the following errors:
---
net.java.games.jogl.GLException: java.lang.reflect.InvocationTargetException
       at net.java.games.jogl.GLCanvas.maybeDoSingleThreadedWorkaround(GLCanvas.java:233)
       at net.java.games.jogl.GLCanvas.display(GLCanvas.java:77)
       at net.java.games.jogl.Animator$1.run(Animator.java:107)
       at java.lang.Thread.run(Thread.java:534)
Caused by: java.lang.reflect.InvocationTargetException
       at java.awt.EventQueue.invokeAndWait(EventQueue.java:834)
       at net.java.games.jogl.GLCanvas.maybeDoSingleThreadedWorkaround(GLCanvas.java:228)
       ... 3 more
Caused by: net.java.games.jogl.GLException: Error making context current
       at net.java.games.jogl.impl.x11.X11GLContext.makeCurrent(X11GLContext.java:172)
       at net.java.games.jogl.impl.x11.X11OnscreenGLContext.makeCurrent(X11OnscreenGLContext.java:118)
       at net.java.games.jogl.impl.GLContext.invokeGL(GLContext.java:246)
       at net.java.games.jogl.GLCanvas$DisplayOnEventDispatchThreadAction.run(GLCanvas.java:266)
       at java.awt.event.InvocationEvent.dispatch(InvocationEvent.java:171)
       at java.awt.EventQueue.dispatchEvent(EventQueue.java:454)
       at java.awt.EventDispatchThread.pumpOneEventForHierarchy(EventDispatchThread.java:201)
       at java.awt.EventDispatchThread.pumpEventsForHierarchy(EventDispatchThread.java:151)
       at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:145)        at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:137)        at java.awt.EventDispatchThread.run(EventDispatchThread.java:100)

---

I'm running on 2.6.5-gentoo-r1 with the ati-drivers 8.8.25-r3.

Any help much appreciated.

[Ken Russell]

I'm trying to get the 1.1b12 running so I can play with the gleem package and try again to see if I can get some nice zooming, panning, and rotating w/ the mouse implemented (which btw, anyone happen to know if gleem now works w/ orthographic projections)?

There haven't been any enhancements to gleem recently so you'll still have to extend the package to add support for this.

Anyways, when I start demos.gears.Gears, I get the following errors:

Caused by: net.java.games.jogl.GLException: Error making context current

I'm running on 2.6.5-gentoo-r1 with the ati-drivers 8.8.25-r3.

Is this a  new failure in 1.1 b12 or was it present with earlier versions? I can't think of anything that changed in JOGL's OpenGL context handling on X11. Are there any other exception causes further up the stack trace? Have you deleted all old jogl.jar and jogl-demos-*.jar files? ATI's Linux drivers don't work on my machine any more, but the current release ran fine on several other Linux and Solaris machines.

[Ken Russell]

The final JOGL 1.1 build has been released on June 24, 2005. Please see the following thread for details:

JOGL Release Information

Please post comments about the new release here.

[zhaoyoubing]

Congratulations!

[zhaoyoubing]

Another small question to Ken, where can I find JSR 231 APIs ?? Thx .

[jill]

apologies in advance if this doesn't qualify as comments about the new release. it probably does though but feel free to remove if this should go into a different thread:
for some reason since installing the 1.1 release, my system doesn't pick up the proper driver at 16-bit resolution.

Everything works just as it used to with b12 (i.e., fine) when 32bit resolution is active, but when the display is set to 16bit, only the GDI generic renderer is available and so all the ARB extensions are unavailable.

I haven't changed the drivers (latest Intel 82855 drivers for IBM laptops), and haven't changed the code. Putting back the b12 libraries/dlls fixes the problem. Running on WinXP.

Any ideas?

[Markus_Persson]

Wooo, nice.

* Markus_Persson mutters something about a milestone, and looks happy

[Ken Russell]

apologies in advance if this doesn't qualify as comments about the new release. it probably does though but feel free to remove if this should go into a different thread:

This is the right place for your comment.

for some reason since installing the 1.1 release, my system doesn't pick up the proper driver at 16-bit resolution.

Everything works just as it used to with b12 (i.e., fine) when 32bit resolution is active, but when the display is set to 16bit, only the GDI generic renderer is available and so all the ARB extensions are unavailable.

I haven't changed the drivers (latest Intel 82855 drivers for IBM laptops), and haven't changed the code. Putting back the b12 libraries/dlls fixes the problem. Running on WinXP.

Any ideas?

Between 1.1 b12 and 1.1 final we changed the Windows pixel format selection code to prefer ChoosePixelFormat over wglChoosePixelFormatARB to work around some serious memory leaks with older ATI drivers. I'm pretty sure this is why you're seeing different behavior on your Intel chipset.

Have you tried writing your own GLCapabilitiesChooser and/or trying Mojang's ScoreCapabilitiesChooser which is attached to one of the bugs in the JOGL Issue Tracker?

In general I would recommend running your laptop at 32-bit depth if you're doing OpenGL work. If you don't need any of the other bug fixes in the 1.1 final release then stick with b12. If you could file a bug describing the change in behavior beteween 1.1 b12 and 1.1 final I'd appreciate it.

[guo]

When could jogl move to package javax.media? As part of Mustang?
I can't wait to use it in applet~