> Not to irritate, but I've run stuff where I accessed 32 bit depth on gpu's > that don't support it, without writing a fallback routine, and not had > crashes. That's not to say its the right way to do it, or that it may not > crash with some OS/gpu combo. I'd suspect it may be one of the other things > referred to.
He (Mark) wasn't reporting crashes, but pixel format creation failure. The latter doesn't necessarily imply the former. Additionally, there are some extra attributes which can be included in a pixel format that say "If you can't do this, pick something lower" or "pick something higher" and some other variations (See NSOpenGLPFA*Policy - there's Closest, Minimum, and Maximum). My reply was mostly in terms of trying to encourage best practices for GL pixel format creation generally, not to necessarily solve this particular problem specifically (though removing the fullscreen attribute may help in this particular case - depth size may also be a factor since it isn't universally supported); sorry for the ambiguity. It's possible to try and make 32 bit depth attachments for FBOs after a context has already been successfully created -- this might be what you did (it's unclear what "accessed 32 bit depth" means exactly), and isn't proscribed by the context's pixel format. -- Christopher Wright christopher_wri...@apple.com
_______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
