Pls test new JOGL aggregated build (pre RC11) - no XInitThreads() ..

Aggregated build:
  <http://jogamp.org/deployment/archive/master/gluegen_594-joal_362-jogl_819-jocl_683/>

Signed Aggregated build:
  <http://jogamp.org/deployment/archive/master/gluegen_594-joal_362-jogl_819-jocl_683-signed/>

Jenkins Build 819:
  <https://jogamp.org/chuck/job/jogl/819/>

Prev. Jenkins Build 803:
  <https://jogamp.org/chuck/job/jogl/803/>

Covering changes from
  <http://jogamp.org/git/?p=jogl.git;a=commit;h=2ea3f82baa32ff7f25856c7e3210fb70664b49e6>, to
  <http://jogamp.org/git/?p=jogl.git;a=commit;h=43891be36e8485353ac74f329fd2f7438303a846>

+++

Disclaimer: ATI driver issues are known, i.e. sporadic XCB assertions w/ concurrent access.
            Pls report them anyways if they bother you.

+++

43891be Cache XineramaIsEnabled(dpy) per display-name in X11Util minimize triggering ATI driver bug of sporadic XCB errors and to reduce server roundtrips.
e0904a6 Misc X11: Add GLX_MESA_swap_control; GLXExtensions (private); X11GLXContext commented out GLX_MESA_swap_control; native test of Mesa context-retarget bug, cannot re
18ae645 GLContext.makeCurrent(): Clean API doc
4f8432b GLDrawableHelper.init(..): Avoid double reshape if subsequent display issues reshape
27fe889 Adding Mesa Quirk 'NoSetSwapIntervalPostRetarget': SIGSEGV on setSwapInterval() after changing the context's drawable w/ 'Mesa 8.0.4' dri2SetSwapInterval/DRI2 (soft
30d6d5e X11 NativeWindow: Fix Error and IOError handler, i.e. NPE checks and no JVM dependency (signal is async); X11Util: XSync w/ discard reusable Display when queue-in.
52b4990 Texture: Clarifiy API doc of getTextureObject(GL)
e4176f4 NativeWindow/X11 + NEWT/X11: Cache 'isXineramaEnabled()' to reduce X11 server roundtrips.
54f79e4 NEWT X11 ScreenMode: Ignore invalid rotation event
4c24011 GLXUtil: Lock X11 Device
285bd9b FBObject: Clarify reset(..), resetSamplingSink(..) ; Rename syncFramebuffer(..) -> syncSamplingSink(..)
fbe331f Fix Bug 616: X11: Remove XInitThreads() dependency while cleaning up device locking, resulting in a native-lock-free impl.
ba846a4 Fix Bug 621, JVM Crashes During X11 Shutdown while closing all unclosed display connections.
495a1fb FBObject: Small cleanup of freeColorbufferImpl(gl, i) -> freeAllColorbufferImpl(gl)
459af1b MacOSXCGLContext: Fix missing initialization of 'needsSetContextPBuffer' if not offscreen-layer.
5b378cb MacOSXCGLDrawableFactory: Cannot use shared context to retrieve quirks since it's transient, store it.
48f033d NativeWindowFactory/NEWT: Use relative sub-package names in NativeWindowFactory's TYPE_* strings, not NEWT's; NEWTFactory: Use default NEWT package name if rel. 'pa
923d9dd Add GLRendererQuirks; Fix shared EGL/ES resources;
0f531ec FBO tests: Add singleBuffer cmd-line option for offscreen-layered tests; FBOMRT: Allow test w/ only one attachment number (sure not MRT then)
60c63ac OSX CALayer / MacOSXCGLContext: Remove pbuffer swap garbage; Split pbuffer/FBO revalidation/redraw; Don't sync @ needsDisplay
7da8654 GLFBODrawableImpl: Fix garbage @ swap, add [disabled] reset-quirk for testing
8be507a FBObject cleanup / review: strings, simplified re-attachment, removed redundancy, adding quirk.
55e2ff7 GLReadBufferUtil, ShaderUtil: Catch GLException (and dump if thrown), to increase robustness $ glReadPixels and Get NUM_SHADER_BINARY_FORMATS
5684b94 Make GLBufferStateTracker more robust: Catch GLError on Get(), causes INVALID_ENUM w/ pname GL_PIXEL_UNPACK_BUFFER_BINDING on GL 3.0 Mesa 8.0.4
3c09f42 Fix Bug 612: glVertexAttribIPointer/glVertexAttribLPointer are BufferObjectKind (VBO) and needs a RangeCheck as well
3f5aac3 Fix non NIO direct GL semantics: Restrict certain GL function to NIODirectOnly (again)

+++

~Sven

Prev. aggregated build removed.

+++

Aggregated build:
  <http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_824-jocl_686/>

Signed Aggregated build:
  <http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_824-jocl_686-signed/>

Jenkins Build 824:
  <https://jogamp.org/chuck/job/jogl/824/>

Prev. Jenkins Build 819:
  <https://jogamp.org/chuck/job/jogl/819/>

Covering changes from
  <http://jogamp.org/git/?p=jogl.git;a=commit;h=8a1e50dc33e14e9423e2f4bc066b32a835bb33fb>, to
  <http://jogamp.org/git/?p=jogl.git;a=commit;h=bb8454d735c511c6d80798d3b6258e1ed355579e>

+++

bb8454d Refine a3cb6bb14f410f67fccf5ccd4cd7ecc66f448389, fix deadlock (regression)
67c2ab9 EGL/X11 DummySurfaceHook: Lock display device at creation/destruction
bab6160 Refine ed7d5f2e3fc696d47b10c8d62a071643bf385588, set and restore FBO caps @ setRealized(true/false) already, refine at initialize(true)
ed7d5f2 FBO/Offscreen: Use most simple GLCapabilities for GLFBODrawableImpl's underlying dummy surface/drawable
f541f34 Fix FBObject DEPTH_STENCIL renderbuffer usage
93c557c UITestCase.snapshot: Add Depth and Stencil bits to filename
7422822 Use helper.isExternalAnimatorAnimating() instead of helper.isAnimatorAnimating() for decision whether to display() now; Minor API comments.
a3cb6bb Fix Bug 572 (2nd time): GLCanvas.validateGLDrawable() @ display() and reshape() ; GLCanvas.reshape() only if drawble valid ; GLCanvas.validateGLDrawable() also test
8f6233f Fix regression of commit fbe331f013608eb31ff0d8675f4e4c9881c9c48b: X11 DisplayDriver dispatchMessagesNative() aDevice NPE at finally
84632ca Fix SWTEDTUtil bug, where it simply doesn't start by an implicit 'invoke()' - No NewtCanvasSWT resize, nor input event delivery.
a44c2df OSX Context: Fix caps determination for on-/offscreen since pformat's NSOpenGLPFAPixelBuffer is ambiguous
7278244 Footprint Stats ; Fix Bug 624 - Compile and generate android JAR files (jogl-all-android.jar, jogl-test-android.jar) if android SDK JAR files are available. ; Split
4c9083b Fix regression of commit a644d779ab19cb1d200ae4ba567b9c042c34b337, cannot compile FixedFuncHook due to removed 'isDirty()'
65a94fb FixedFuncPipeline: Cleanup warnings ..
3ae26ed ShaderState: Adding Class API doc and removing attached object w/ 'int' key, similar to GLContext change 6d241fc2a46413ee478985d676d2481c5a7ed119
a644d77 Update PMVMatrix/GLMatrixFunc API doc and refine PMVMatrix update / get-Mvi/Mvit-Matrix operation. (Minor API change)
f7ca4df Fix regression of fbe331f013608eb31ff0d8675f4e4c9881c9c48b [Remove XInitThreads()]
a7ab7a7 Misc: tests.sh
9c96be7 NativeWindowFactory: Remove 'remedy' of Bug 613 Commit 92398025abdabb2fdef0d78edd41e730991a6f94 GlobalToolkitLock for create/destroy
541dc2c Misc: Add commented-out NEWT X11/XCB-Event and X11ScreenRandR1[13] compilation ; ...
eb7ca05 GLRendererQuirks.NoDoubleBufferedPBuffer: Set if ( mesa || gallium ) && hwAccel
98d5ae6 FBObject/GLFBODrawable: Debug / glError detection
21b85e6 Fix commit fbe331f013608eb31ff0d8675f4e4c9881c9c48b (Bug 616 - Remove XInitThreads())
7c333e3 Relax Bug 613 workaround of commit 92398025abdabb2fdef0d78edd41e730991a6f94
0790108 JAWTUtil ToolkitLock: Add lock counter to make lock validation recursive (Fix regression of commit 92398025abdabb2fdef0d78edd41e730991a6f94)
c69d62a test script: remove deleted unit test
7302b10 NullToolkitLock.validateLocker(): Throw exception if locking is required (detecting impl. bugs)
e5c1e60 X11Util: Detailed closing information only in DEBUG mode, just print summary
ef4990e Merge remote-tracking branch 'origin/master'
2d35fc5 Bug 623: X11Util adds property 'nativewindow.debug.X11Util.ATI_HAS_NO_MULTITHREADING_BUG' to disable GlobalToolkitLock workaround for testing.
9239802 Workaround for Bug 623: Sporadic XCB assertion failures w/ ATI proprietary driver and w/o native X11 locking
20a2a4f Merge pull request #52 from xranby/bcm.vc.iv-autodetect
8351d51 Add braces and fix indentation to conform with jogl coding standards.
fa55b54 ".bcm.vc.iv" is enough now -> NEWT agnostic relative package path.

Known issues with the pre RC11:
http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_824-jocl_686-signed/

- TODO/ enhancement  : It is not yet possible to run the web browser applets or AWT on the Raspberry Pi using the newt bcm.vc.iv driver.
https://jogamp.org/bugzilla/show_bug.cgi?id=626

- The java doc in pre rc11 is not re-generated and do not match src:
[ ] jogl-javadoc.7z 24-Aug-2012 09:47 2.1M
expected a jogl-javadoc.7z from 5 Oct 2012.
example the PMVMatrix javadoc do not contain the new enhancements:
http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_824-jocl_686-signed/javadoc/jogl/javadoc/com/jogamp/opengl/util/PMVMatrix.html
Expected this doc to contain https://github.com/sgothel/jogl/commit/a644d779ab19cb1d200ae4ba567b9c042c34b337

On 10/06/2012 03:23 AM, Xerxes Rånby [via jogamp] wrote:
> The java doc in pre rc11 is not re-generated and do not match src:
> example the PMVMatrix javadoc do not contain the recent enhancements found in
> the source:
> http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_824-jocl_686-signed/javadoc/jogl/javadoc/com/jogamp/opengl/util/PMVMatrix.html
>

Thats true (javadoc generation takes a lot of time).
However, the source zip files for attachment (IDE) are up-to-date,
so OK for developers.

javadoc will be updated for RC11.

~Sven

On 10/06/2012 02:51 AM, Xerxes Rånby [via jogamp] wrote:
> Known issues with:
> http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_824-jocl_686-signed/
> - TODO/ enhancement  : It is not yet possible to run the web browser applets
> on the Raspberry Pi using the newt bcm.vc.iv driver.
> https://jogamp.org/bugzilla/show_bug.cgi?id=626

As described in the bug 626 entry,
do you like to solve it ?

~Sven

Maybe you could update the javadoc at least just before the RC11.

It works fine, I have no crash when leaving (Nvidia NVS 285).

Prev. aggregated build removed.

API doc updated (thx to Xerxes & Julien of reminding me).

+++

Aggregated build:
  <http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_825-jocl_687/>

Signed Aggregated build:
  <http://jogamp.org/deployment/archive/master/gluegen_598-joal_364-jogl_825-jocl_687-signed/>

Jenkins Build 825:
  <https://jogamp.org/chuck/job/jogl/825/>

Prev. Jenkins Build 824:
  <https://jogamp.org/chuck/job/jogl/824/>

Covering changes from
  <http://jogamp.org/git/?p=jogl.git;a=commit;h=d3033e42faa909f6c1055f5ba3c7925766a3e583>, to
  <http://jogamp.org/git/?p=jogl.git;a=commit;h=630d43be7a8cccf28632fc1f7542bb484382d775>

+++

630d43b Fix Bug 617 (part2): Workaround buggy GPU driver where FBO reattachment of Colorbuffer w/ diff size leads to GL_FRAMEBUFFER_UNSUPPORTED.
d3033e4 Adding unit test for Bug 605 - WORKSFORME, i.e. no unwanted y-flip, so I can just assume it's the SW renderer, flipping.

Summary of OSX 10.8.1 issues on twitter:

1h ago Jan Ekholm Jan Ekholm ‏@jan_ekholm
JOGL on OSX doesn’t seem to be too well supported yet. Can’t get any context that actually can render without getting “invalid context”.

40m Xerxes Rånby Xerxes Rånby ‏@xranby
@jan_ekholm can you test with the new pre RC11? It include a lot of OSX fixes. http://forum.jogamp.org/Pls-test-new-JOGL-aggregated-build-pre-RC11-no-XInitThreads-tp4026312p4026410.html …

38m Jan Ekholm Jan Ekholm ‏@jan_ekholm
@xranby I downloaded the newest nightly build, from the 6th I think. The code is ok on Linux, so there can’t be something big I do wrong.

28m Xerxes Rånby Xerxes Rånby ‏@xranby
@jan_ekholm Please report your findings in the thread. Which version of OSX are you using? OSX is the hardest platform for #JOGL to support.

16m Jan Ekholm Jan Ekholm ‏@jan_ekholm
@xranby 10.8.1, i.e the newest. Could be something I do wrong too. new GLCapabilities( GLProfile.get(GLProfile.GL3) );

2m Xerxes Rånby ‏@xranby
@jan_ekholm I have added your report to the forum thread, I will keep you informed if we can A) reproduce the bug and B) find a fix for it.

Xerxes, has jan_ekholm tried NEWT? Sorry to avoid replying on Twitter, I refuse feeding this monster.

I'm the Twitter user referred to above.

OSX is 10.8.1.
JOGL is jogl-2.0-b825-20121006-macosx-universal (daily build).

My canvas is set up as follows:

        GLProfile glprofile = GLProfile.get(GLProfile.GL3);
        GLCapabilities glcapabilities = new GLCapabilities( glprofile );
        glcapabilities.setHardwareAccelerated( true );
        final Canvas canvas = new Canvas( glcapabilities );

Some added checks removed (such as a check for glprofile.isGL3()). So far no errors. Later there is
code like this to do some sanity checks:

    public void initialize(GL3 gl) {
        System.out.println( "Scene.initialize" );

        System.out.println( "Scene.initialize: INIT GL IS:  " + gl.getClass().getName() );
        System.out.println( "Scene.initialize: GL_VENDOR:   " + gl.glGetString( GL.GL_VENDOR ) );
        System.out.println( "Scene.initialize: GL_RENDERER: " + gl.glGetString( GL.GL_RENDERER ) );
        System.out.println( "Scene.initialize: GL_VERSION:  " + gl.glGetString( GL.GL_VERSION ) );

        if ( !gl.isFunctionAvailable( "glGenBuffers" )
                || !gl.isFunctionAvailable( "glBindBuffer" )
                || !gl.isFunctionAvailable( "glBufferData" )
                || !gl.isFunctionAvailable( "glDeleteBuffers" ) ) {
            System.exit( 1 );
        }
        ...
}

This gives:

Scene.initialize: INIT GL IS:  jogamp.opengl.gl4.GL4bcImpl
Scene.initialize: GL_VENDOR:   NVIDIA Corporation
Scene.initialize: GL_RENDERER: NVIDIA GeForce GT 650M OpenGL Engine
Scene.initialize: GL_VERSION:  3.2 NVIDIA-8.0.61

When my shaders are compiled and linked I see this error:

createAndCompileShader: Pre GL Error: 0x500

No error in the info log and everything completes nicely and seems to be ok apart from that printed error.

Later when rendering I see:

2012-10-08 14:58:10.779 java[3950:707] invalid share context
2012-10-08 14:58:10.779 java[3950:707] invalid context

Nothing is shown.

gouessej wrote

Xerxes, has jan_ekholm tried NEWT? Sorry to avoid replying on Twitter, I refuse feeding this monster.

gouessej please explain, what is monsterish/trollish about discussing JOGL issues on Twitter? Is Twitter the monster? I am assuming you refer to the Twitter service and not Jan in person.

Jan Ekholm wrote

I'm the Twitter user referred to above.

Welcome to the discussion and thank you for reporting issues with jogl.

Jan Ekholm wrote

When my shaders are compiled and linked I see this error:

createAndCompileShader: Pre GL Error: 0x500

No error in the info log and everything completes nicely and seems to be ok apart from that printed error.

Error 0x500 indicate that you have a GL_INVALID_ENUM at some place.
Try look into why you get this first.

Jan Ekholm wrote

My canvas is set up as follows:

        GLProfile glprofile = GLProfile.get(GLProfile.GL3);
        GLCapabilities glcapabilities = new GLCapabilities( glprofile );
        glcapabilities.setHardwareAccelerated( true );
        final Canvas canvas = new Canvas( glcapabilities );

I am a bit puzzled by your code, are you using a SWT or AWT Canvas or something else?
com.jogamp.opengl.awt.GLCanvas
com.jogamp.opengl.swt.GLCanvas
com.jogamp.newt.awt.NewtCanvasAWT

In order to isolate the bug that may be isolated in one of the above Canvas can you try use a NEWT GLWindow instead of a Canvas?

Try replace the Canvas implementation
final Canvas canvas = new Canvas( glcapabilities );
with a NEWT GLWindow.
NEWT is the most platform independent way we have in JOGL to open a drawable.
http://jogamp.org/jogl/doc/NEWT-Overview.html

import com.jogamp.newt.opengl.GLWindow;

        GLWindow glWindow = GLWindow.create(caps);

        glWindow.setTitle("foo");
        glWindow.setSize(width,height);
        glWindow.setUndecorated(false);
        glWindow.setPointerVisible(true);
        glWindow.setVisible(true);

        // Finally we connect the GLEventListener application code to the NEWT GLWindow.
        // GLWindow will call the GLEventListener init, reshape, display and dispose
        // functions when needed.
        glWindow.addGLEventListener(new RawGL2ES2demo() /* GLEventListener */);
        Animator animator = new Animator(glWindow);
        animator.add(glWindow);
        animator.start();

The canvas is a javax.media.opengl.awt.GLCanvas, the class used there is my own minimal wrapper class.

The GL_INVALID_ENUM comes from a:

        gl.glEnable( GL.GL_TEXTURE_2D );

This was not triggered on the Linux side, but it is perfectly ok as that enum really is deprecated for a core 3.x profile.

Now there are not other errors anymore apart from the invalid context that comes when rendering for the first time. It is
only shown once and never shows up after that.

I have not tried NEWT, I have all my event handling done for AWT and I'm reluctant to rip it apart again. Not too good with this Java UI stuff yet. Java's UI frameworks really are clusterfscks of the form "if in doubt, add more levels of abstraction"...

AWT is one of the first GUI frameworks for Java unfortunately it is single-threaded and gets in the way for fluid OpenGL rendering all the time. AWT got a event dispatch thread that is also in charge of GFX updates.
AWT only work on Desktop devices with a JavaSE and some JavaME stack. thus not really suitable for mobile where you want portability and have to do graphics updates using OpenGL ES.

NEWT got created inside the JOGL project in order to be able to do event handling in parallel to OpenGL rendering.
NEWT work on AWT-less java implementations where we can cut all the legacy AWT like on for example Android phones and tables. We can also use NEWT to get hardware acceleration on exotic frame-buffer EGL implementations like the Raspberry Pi also suitable for running on devices like MeeGo/Harmattan Nokia N9 phones.
If you want to remove layers of abstraction then NEWT is the tool, it will give you the lowest level access to OpenGL/ES. Using NEWT the Window is the EGLNativeWindowType.
http://jogamp.org/jogl/doc/NEWT-Overview.html

Also we have less obscure bugs using NEWT so we love it.

On 10/08/2012 02:03 PM, Jan Ekholm [via jogamp] wrote:
> I'm the Twitter user referred to above.
>

Thank you for joining here.

> OSX is 10.8.1.
> JOGL is jogl-2.0-b825-20121006-macosx-universal (daily build).
>
> My canvas is set up as follows:
>
>         GLProfile glprofile = GLProfile.get(GLProfile.GL3);
>         GLCapabilities glcapabilities = new GLCapabilities( glprofile );
>         glcapabilities.setHardwareAccelerated( true );
>         final Canvas canvas = new Canvas( glcapabilities );

Can you pls do the following:

- see our wiki/faq/bugreport

- post a complete [unit] test here which fails for you ?
  - this can be one of our existing (preferable),
    which you may have modified.
  - run with -Djogl.debug=all -Dnewt.debug=all
    and post log file here / bug-report

- debugging a JOGL app.
  - find general error:
    - -Djogl.debug=all -Dnewt.debug=all

  - find opengl error
    - -Djogl.debug.DebugGL -Djogl.debug.TraceGL, or just
    - -Djogl.debug.DebugGL
Always check the extension availability 1st,
function may be available even though not implement.

~Sven

--
health & wealth
mailto:[hidden email] ; http://jausoft.com
land : +49 (471) 4707742 ; fax : +49 (471) 4707741
Timezone CET: PST+9, EST+6, UTC+1

Not sure I can use NEWT here. This is not really for embedded use and will need to integrate with an older and much larger legacy desktop code base as a 3D visualization module. I'll see what can be done, but if NEWT is not easily integrated into such an environment I guess it's ruled out for now.

My little app is already somewhat complex, so it's not too easy to extract a standalone thing which I can post here. I started with this on Linux where everything worked fine and only later moved to OSX (due to my OSX laptop having more memory...). I'm such a n00b when it comes to Java and OpenGL in general, so whipping up a test app will take me some days. If all fails I'll just abandon using OSX for the OpenGL parts of this project.

I'm sorry, I'm quite useless at helping you help me with this issue. So, don't worry about it.

On 10/09/2012 09:33 AM, Jan Ekholm [via jogamp] wrote:
> Not sure I can use NEWT here. This is not really for embedded use and will
> need to integrate with an older and much larger legacy desktop code base as a
> 3D visualization module. I'll see what can be done, but if NEWT is not easily
> integrated into such an environment I guess it's ruled out for now.
Even though we prefer NEWT, I didn't said anything to push you to it.
Or did I even mention NEWT in my last email, but the debug flags ?
Sorry if I have mislead you.
Just send your crash with all debug flags enabled from the commandline:
  java -Djogl.debug=all -Dnativewindow.debug=all -cp <your-classpath> YourApp >&  YourApp.log

Then you can send us your log file.

shouldn't be too hard ..

~Sven

Oh, no worries about that NEWT thing, I just got the impression it's the only supported platform.

I tried the debug flags but execution seems to barf out when I load a texture, apparently there's some
problem with mipmap generation that causes an exception when running with those flags but which seems to
work ok when running normally. Will try to solve that and then get a proper log file.

On 10/09/2012 01:31 PM, Jan Ekholm [via jogamp] wrote:
> Oh, no worries about that NEWT thing, I just got the impression it's the only
> supported platform.
>
> I tried the debug flags but execution seems to barf out when I load a texture,
> apparently there's some
> problem with mipmap generation that causes an exception when running with
> those flags but which seems to
> work ok when running normally. Will try to solve that and then get a proper
> log file.

another hint:

-Djogl.debug.DebugGL  <- enabled debug pipeline, i.e. test GL error and throws exception
-Djogl.debug.TraceGL  <- enabled trace pipeline, i.e. dumps all GL commands to stderr

~Sven