Created attachment 29060 [details]
c test file
On Linux 188.8.131.52-217.2.16.fc11.i686.PAE i686 i386 GNU/Linux
OpenGL vendor string: Tungsten Graphics, Inc
OpenGL renderer string: Mesa DRI Mobile Intel® GM45 Express Chipset GEM 20090114 x86/MMX/SSE2
OpenGL version string: 2.1 Mesa 7.5-devel
OpenGL shading language version string: 1.20
And >> lsmod | grep i915
i915 147340 2
drm 166512 2 i915
i2c_algo_bit 4804 1 i915
i2c_core 18016 4 i2c_i801,i915,drm,i2c_algo_bit
video 17344 1 i915
Using glDrawBuffer(GL_FRONT) with JOGl or with C++/OpenGl result to a drawing on the upper-left part of the screen instead of the GL window.
Created attachment 29061 [details]
screenshot of a C test program
Created attachment 29062 [details]
screenshot of a scilab witch use JOGL AWTcanvas.
When Scilab use AWT GLCanvas, the zoom rectangle is draw out of the window.
Rem : when scilab use SWIG GLJPanel no bugs append, but on Intel Graphics, GLJPanel don't work must of the time, it's why we prefer to use AWT canvas.
The SwapBuffers swaps the back on top of the frontbuffer rendering you did. Remove it, and your demo app works.
Off course .. but.
My problem is not about seeing the cube.
My problem is that the GL frame isn't draw in the app window but on the upper-left corner off the screen.
See the screen-shot :
Created attachment 29089 [details]
A screenshot of the test program without glutSwapBuffers
The difference is that the window is not draw at all (no random colours, just the back off the window)
Created attachment 29090 [details]
Single Buffered test program.
Here a single Buffered test program.
glutInitDisplayMode(GLUT_RGB | GLUT_SINGLE | GLUT_DEPTH);
The result is the same :
The window program is empty and the GL frame is rendered to the upper-left corner of the screen (see the previews screen shot).
What version of the 2D driver (xf86-video-intel) and xserver do you have? The front buffer rendering fixes require updated versions of all three packages. I believe xf86-video-intel 2.8.1 and xserver 1.6.2 are the minimum versions required.
On my system with GIT master installed from all three, the single buffered test program produces the correct results.
I also see from the version string that you're using a pre-release of Mesa 7.5 (the "7.5-devel" part). Have you tried the actual 7.5 release?
Two weeks with no update: closing.