jogamp › jogl

New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Classic

List

Threaded

20 messages Options

Sven Gothel

New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

Find it here:
http://jogamp.org/deployment/archive/master/gluegen_476-joal_269-jogl_623-jocl_529/

Please test this version w/ Mac OSX (10.5.* - 10.7.*),
a report would be greatly appreciated!

Report from other machines are welcome as well, ofc.

~Sven

gouessej

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

Does it fix the cursor problem on Mac?

Julien Gouesse | Personal blog | Website

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

On Saturday, January 14, 2012 12:10:57 PM gouessej [via jogamp] wrote:
>
> Does it fix the cursor problem on Mac?

No - sorry, I haven't checked that bug yet.
All I can say the methods are impl. and worked on OS X when I tested them manually.
I will redo this soon.

This pre-release shall fix crashes/freezes/low-performance on OS X 10.5 - 10.7 ..
So, whoever experienced such instability - please test and lets try to solve
such remaining problems to get rid of them once and for all.

As you can see here, we are -10 failures:
https://jogamp.org/chuck/view/fwd/job/jogl/lastSuccessfulBuild/label=macosx-10_6-x86_64-nvidia/testReport/

Here you can see the performance fix, 23min -> 12min for all tests - ie. back to 'normal':
https://jogamp.org/chuck/view/fwd/job/jogl/lastSuccessfulBuild/label=macosx-10_6-x86_64-nvidia/testReport/history/?

~Sven

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by Sven Gothel

I got access to an old Windows XP 32-bit machine and basic tests (NEWT/AWT)
shows no crash, GL2 and GL2ES2 profiles are working (ie. w/ GLSL).

A junit test run is underway ..

+++
Thinkpad TP60 (?)
Windows XP Version 2002, SP 3
Intel T2400 @1.83 Ghz

Platform: WINDOWS / Windows XP 5.1 (os), x86 (arch) 2 cores
Platform: Java Version: 1.6.0_30, VM: Java HotSpot(TM) Client VM, Runtime: Java(TM) SE Runtime Envir..

GL Profile GLProfile[GL2/GL2]
CTX VERSION 2.1 (compatibility profile, any, new) - 2.1.8544 Release
GL jogamp.opengl.gl4.GL4bcImpl@b2fd8f
GL_VENDOR ATI Technologies Inc.
GL_VERSION 2.1.8544 Release
GL_RENDERER: ATI Mobility Radeon X1300

ATI driver (lenovo/mobile) 8.593.100.7-090929a from 2009-11-13

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

On Saturday, January 14, 2012 02:50:10 PM Sven Gothel [via jogamp] wrote:
>
> I got access to an old Windows XP 32-bit machine and basic tests (NEWT/AWT)
> shows no crash, GL2 and GL2ES2 profiles are working (ie. w/ GLSL).
>
> A junit test run is underway ..

- 1 UI/Focus failure
- 1 GLSL Error w/ ElektronenMultiplizierer (heavy GLSL)

All other tests passed - looks good then.

~Sven

>
> +++
> Thinkpad TP60 (?)
> Windows XP Version 2002, SP 3
> Intel T2400 @1.83 Ghz
>
> Platform: WINDOWS / Windows XP 5.1 (os), x86 (arch) 2 cores
> Platform: Java Version: 1.6.0_30, VM: Java HotSpot(TM) Client VM, Runtime:
> Java(TM) SE Runtime Envir..
>
> GL Profile GLProfile[GL2/GL2]
> CTX VERSION 2.1 (compatibility profile, any, new) - 2.1.8544 Release
> GL jogamp.opengl.gl4.GL4bcImpl@b2fd8f
> GL_VENDOR ATI Technologies Inc.
> GL_VERSION 2.1.8544 Release
> GL_RENDERER: ATI Mobility Radeon X1300
>
> ATI driver (lenovo/mobile) 8.593.100.7-090929a from 2009-11-13

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by gouessej

On Saturday, January 14, 2012 01:48:43 PM Sven Gothel wrote:
> On Saturday, January 14, 2012 12:10:57 PM gouessej [via jogamp] wrote:
> >
> > Does it fix the cursor problem on Mac?
>
> No - sorry, I haven't checked that bug yet.
> All I can say the methods are impl. and worked on OS X when I tested them manually.
> I will redo this soon.

I could replicate the mouse invisible bug on OS X 10.6.8 - good.
It works on OS X 10.7.2 though.

~Sven

jouvieje

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

I've just updated from jogl 565 to 623 and my NewtCanvasAWT code is broken: nothing is rendered anymore to the screen. The render loop is running as if everything was fine, but nothing is rendered to the screen. I can only see a white area in place of the NewtCanvasAWT component.

If I use GLCanvas or GLWindow/Window, everything is fine.

Is there some recent changes on NewtCanvasAWT that I missed ? (I'm running windows 7 x64)

Jérôme

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

On Sunday, January 15, 2012 12:26:51 PM jouvieje [via jogamp] wrote:

>
> I've just updated from jogl 565 to 623 and my NewtCanvasAWT code is broken:
> nothing is rendered anymore to the screen. The render loop is running as if
> everything was fine, but nothing is rendered to the screen. I can only see a
> white area in place of the NewtCanvasAWT component.
>
> If I use GLCanvas or GLWindow/Window, everything is fine.
>
> Is there some recent changes on NewtCanvasAWT that I missed ? (I'm running
> windows 7 x64)

Changes happend in the makeCurrent/release code of context and drawable helper
but not really within NewtCanvasAWT in particular AFAIK.

Is it reproducable w/ one of our unit tests
or can you send a small example please ?

Looking forward to fix this regression.

~Sven

jouvieje

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Is there an archive with junit test sources ?
I've tried to look in jogamp*pltform*.7z but there's only the classes and jogl-demos.jar does not have the unit tests too.

Jérôme

jouvieje

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

This post was updated on .

I've modified an old junit test and I can reproduce the problem.

While writting the junit test, I've found the NewtCanvasAWT was rendered but with some kind of offset in its position. First screenshot shows a white area where the NewtCanvasAWT is supposed to be, and after resizing I can see the NewtCanvasAWT rendered in wrong place. This also appear in the unit test (second picture) where the gears are also rendered in wrong place:

bug-1.png
bug-2.png

and junit test:
NewtCanvasAWT_bug.java

Let me know if there's something wrong in the unit test.

Jérôme

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

On Sunday, January 15, 2012 03:52:44 PM jouvieje [via jogamp] wrote:

>
> I've modified an old junit test and I can reproduce the problem.
>
> While writting the junit, I've found the NewtCanvasAWT was rendered but with
> an offset. First screenshot shows a white area where the NewtCanvasAWT is
> supposed to be and after resizing I can see my app with a huge position
> offset. This also appear in the unit test (second picture) where the gears
> are also rendered in wrong place:
>
> http://jerome.jouvie.free.fr/jogamp/bug-1.png bug-1.png
> http://jerome.jouvie.free.fr/jogamp/bug-2.png bug-2.png
>
> and junit test:
> http://jerome.jouvie.free.fr/jogamp/NewtCanvasAWT_bug.java
> NewtCanvasAWT_bug.java
>
> Let me know if there's something wrong in the unit test.
>

Thank you, I will look into it.

~Sven

gouessej

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by Sven Gothel

@Sven Ok it is less important than I thought, it is only broken on Mac OS X 10.6.8. If it is too difficult to fix, should we use a transparent cursor?

Julien Gouesse | Personal blog | Website

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

On Monday, January 16, 2012 10:23:20 AM gouessej [via jogamp] wrote:
>
> @Sven Ok it is less important than I thought, it is only broken on Mac OS X
> 10.6.8. If it is too difficult to fix, should we use a transparent cursor?

http://jogamp.org/git/?p=jogl.git;a=commit;h=f5c8cdad92687782184122fecc341258a6283d89

+++

NEWT/OSX Pointer Invisible Fix: 10.6.* responder declarations & test focus/isInside

On OS X 10.6.8 the lack of responder method declarations (mouseEntered, ..)
lead to ignore the impl. callbacks.

'isMouseInside' cannot rely on 'mouseExit'/'mouseEntered' when
setMouseInvisible() is being called, deduce it manually.

focusLost == mouseExit (OS X behavior), hence focusGained needs
to set mouse invisible/visible in case it's requested.

+++

The 10.6.8 part was 'funny' .. dunno why the OS version makes a difference here.

The other 2 changes fixes the 'invisible' behavior in general on OS X.

In short: Good bug report - this needed to get fixed - thx.

~Sven

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by jouvieje

On Sunday, January 15, 2012 03:52:44 PM jouvieje [via jogamp] wrote:

Nope, it's a regression indeed!

Thank you for reporting. Since I didn't manually verified the behavior
visually, it slipped through our unit tests.

For some reason the x/y offset exists (not corrected)
probably while allowing the negative x/y coordinate patch for 'big-desktop'.

I look into this now.

~Sven

gouessej

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

This post was updated on .

In reply to this post by Sven Gothel

Thank you so much. I will update my demo with it in about 10 hours and I will ask Renanse to give it a try. I really appreciate your efforts and you know subtle details about OS X that are necessary to build a native windowing system like NEWT, good riddance :)

Edit.: I'll wait for the next autobuild containing these fixes to update JogAmp JARs in my project.

Julien Gouesse | Personal blog | Website

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by Sven Gothel

New aggregated builds http://jogamp.org/deployment/archive/master/gluegen_476-joal_269-jogl_627-jocl_533/

Thx to all the bug reports .. 4 days of bug fixes have passed.

https://jogamp.org/chuck/job/jogl/627/

Includes fixes:
https://jogamp.org/chuck/job/jogl/626/
https://jogamp.org/chuck/job/jogl/625/
https://jogamp.org/chuck/job/jogl/624/

andrade

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Hi,
I tested this last build (jogl-627 gluegen-476) on a OSX Leopard (10.5.8) and I'm still getting the same crash that I reported on a previous post:

Exception in thread "main" javax.media.opengl.GLException: Profile GL_DEFAULT is not available on null, but: []
at javax.media.opengl.GLProfile.get(GLProfile.java:655)
at javax.media.opengl.GLProfile.getDefault(GLProfile.java:478)
at javax.media.opengl.GLProfile.getDefault(GLProfile.java:486)
at com.jogamp.newt.opengl.GLWindow.main(GLWindow.java:904)

See the log files attached.
Thanks !
test.log
test_dbg.log

--
Rodrigo Andrade de Almeida

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

On Thursday, January 19, 2012 01:47:12 PM andrade [via jogamp] wrote:

>
> Hi,
> I tested this last build (jogl-627 gluegen-476) on a OSX Leopard (10.5.8)
> and I'm still getting the same crash that I reported on a previous post:
>
> Exception in thread "main" javax.media.opengl.GLException: Profile
> GL_DEFAULT is not available on null, but: []
> at javax.media.opengl.GLProfile.get(GLProfile.java:655)
> at javax.media.opengl.GLProfile.getDefault(GLProfile.java:478)
> at javax.media.opengl.GLProfile.getDefault(GLProfile.java:486)
> at com.jogamp.newt.opengl.GLWindow.main(GLWindow.java:904)
>
> See the log files attached.
> Thanks !
> http://forum.jogamp.org/file/n3672231/test.log test.log
> http://forum.jogamp.org/file/n3672231/test_dbg.log test_dbg.log
>

Thank you Rodrigo - so it's still the same error:

java.lang.UnsatisfiedLinkError: .. libjogl_desktop.jnilib: Symbol not found: _OBJC_CLASS_$_NSOpenGLLayer

too bad, the weak linkage didn't help

now I need to find at least ssh access to a 10.5.* os x machine to fix and test it,
if anybody volunteers .. ?

~Sven

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by andrade

On Friday, January 20, 2012 01:55:09 AM Sven Gothel wrote:

> On Thursday, January 19, 2012 01:47:12 PM andrade [via jogamp] wrote:
> >
> > Hi,
> > I tested this last build (jogl-627 gluegen-476) on a OSX Leopard (10.5.8)
> > and I'm still getting the same crash that I reported on a previous post:
> >
> > Exception in thread "main" javax.media.opengl.GLException: Profile
> > GL_DEFAULT is not available on null, but: []
> > at javax.media.opengl.GLProfile.get(GLProfile.java:655)
> > at javax.media.opengl.GLProfile.getDefault(GLProfile.java:478)
> > at javax.media.opengl.GLProfile.getDefault(GLProfile.java:486)
> > at com.jogamp.newt.opengl.GLWindow.main(GLWindow.java:904)
> >
> > See the log files attached.
> > Thanks !
> > http://forum.jogamp.org/file/n3672231/test.log test.log
> > http://forum.jogamp.org/file/n3672231/test_dbg.log test_dbg.log
> >
>
> Thank you Rodrigo - so it's still the same error:
>
> java.lang.UnsatisfiedLinkError: .. libjogl_desktop.jnilib: Symbol not found: _OBJC_CLASS_$_NSOpenGLLayer
>
> too bad, the weak linkage didn't help
>

Ok, I have fixed the linkage bug on OSX 10.5.8 w/ Java6 (64bit version - the Java Updates).
Note: Java5 is no more supported by any JogAmp project!

http://jogamp.org/git/?p=jogl.git;a=commit;h=d19cd1049d55f93580c37f6a9599f7583e160258

Build will be ready soon (4-14hrs, depending on current cleanup).

> ~Sven
>

Sven Gothel

Re: New pre-release: Unsigned gluegen_476-joal_269-jogl_623-jocl_529

Administrator

In reply to this post by andrade

>
> Ok, I have fixed the linkage bug on OSX 10.5.8 w/ Java6 (64bit version - the Java Updates).
> Note: Java5 is no more supported by any JogAmp project!
>
> http://jogamp.org/git/?p=jogl.git;a=commit;h=d19cd1049d55f93580c37f6a9599f7583e160258
>
A local 64bit only unit test resultet in 7 failures [ 4 failures, and 3 errors ].

~Sven