| Just the BACKING_STORE alone is fine. Use it just like DOUBLE_BUFFER attribute.
-Charlie On Nov 7, 2007, at 3:24 PM, Dario Accornero wrote:
Charlie,
Thank you very much. I'll go ahead and repent publicly for (previously) relying on undefined behaviour.
Since I cannot find any sample code (only the documentation snippet you provided) for using this flag, I'll ask the obvious question -- the docs state AGL_BACKING_STORE is a Boolean. Does this imply that its presence along in the AGL attributes list is considered a valid request, or should I follow the flag with a GL_TRUE in the attributes list?
Thanks and ciao, Dario On 7 Nov 2007, at 23:48, Charlie Lao wrote: Hi Dario,
After a buffer swap, the back buffer content is undefined. You were relying on the undefined behavior in tiger and that actually gave you the result you want. There is a change in leopard that changes this undefined behavior. Your application will be affected if it is build on leopard. Use the following bit in your pixel format selection code. If you want to keep back buffer content valid. The backing store attribute (NSOpenGLPFABackingStore, AGL_BACKING_STORE, andkCGLPFABackingStore) specifies only renderers that have a back color buffer that matches the full size of the drawable object and that guarantee that the back buffer contents are valid after a buffer flushing operation (such as a call to the flushBuffermethod of the NSOpenGLContext class).
-Charlie
On Nov 7, 2007, at 2:10 PM, Dario Accornero wrote: Hello,
I have just upgraded a dual G4 machine (with a Radeon 9800 card) from 10.4.10 to 10.5, and was eager to try out one of our GL products (which requires Tiger). This is a Carbon, windowed AGL application: it runs fine in Tiger, on Intel, ATi and nVidia cards.
Our display methodology requires the backbuffer to be intact between subsequent calls to SwapBuffers (we only redraw the parts that have actually changed): we've had no problems with this approach so far (the Windows port works the same way, but it sets a specific WGL flag for Vista).
However, it looks like this scheme is no longer valid in Leopard (at least on ATi -- we haven't upgraded the other machines yet): apparently the backbuffer does change between calls to SwapBuffers.
Can anyone confirm this? If so, is there any AGL/CGL setting to use the pre-Leopard behaviour?
Thank you,
Dario
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Mac-opengl mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/mac-opengl/email@hidden
This email sent to email@hidden
|
