next »

[Patheros]

I'm just posting this before I commit to see what everyone thinks.

I've created a CanvasPeerSwingImpl and a RenderPeerSwingImpl that uses it.

The only differences being that it uses a GLJPanel rather than a GLCanvas. Other than that they are exactly the same as the AWT versions. The only other difference is I had to comment out line 116 of the Canvas

Code:

this.gl = glJPanel.getGL();

Since it was throwing a null pointer exception and the gl appears to be set in the init method.

Should I commit these into xith3d or into xith-tk? (bear in mind I have no idea what I have commit access to)

[Marvin Fröhlich]

Cool. Please commit it to the xith3d (core) project. I gave you dev-access to both xith3d and xith-tk on subversion (well, at least I hope, I did. I added a user named patheros).

Marvin

[Patheros]

I seems to have worked (On the second try). I'm sure someone out there can break it . Enjoy.

[Marvin Fröhlich]

I fixed some things and comleted the Swing integration (added OpenGLLayer.JOGL_SWING, added Canvas3DJPanel, etc.).

Unfortunately there seems to be something, that needs to be done differently. The Swing rendering is dead slow.

Marvin

[Patheros]

I see what you mean. I'm running the benchmarks and getting 1/2 to a 1/3rd the FPS.
GLJPanel is suppose to be inherently slower than the GLCanvas because the it has to draw to a background buffer and copy it over to the screen. On a side note GLJPanel runs smother than the GLCanvas but I think thats more of a problem with the GLCanvas, which I'm about to post on in a different thread.

[Marvin Fröhlich]

I see what you mean. I'm running the benchmarks and getting 1/2 to a 1/3rd the FPS.
GLJPanel is suppose to be inherently slower than the GLCanvas because the it has to draw to a background buffer and copy it over to the screen. On a side note GLJPanel runs smother than the GLCanvas but I think thats more of a problem with the GLCanvas, which I'm about to post on in a different thread.

How can it run smoother when it runs at a fraction of GLCanvas' FPS . Well, looking forward to see your other thread...

Marvin

[Patheros]

By smother I mean the duration of time between each frame deviates less. That is to say with AWT the duration between each frame is either very small or very large. But with the Swing its always smaller, but larger than average for AWT.

Anyways, I'll take a closer look at this performance issue after the other issue, Jumpy Benchmarks.