[MPlayer-dev-eng] [PATCH] Fix GL_SDL detection on OS X

[Alexander Strange]

On Jun 30, 2010, at 10:37 AM, Reimar Döffinger wrote:

> On Mon, Jun 28, 2010 at 02:09:10AM -0700, Alexander Strange wrote:
>> The configure test was failing because SDL's main() override comes with a prototype that didn't match the function.
> 
> Strange, never happened to me.
> 
>> GL+sdl seems usable after a quick test though a native one would still be better.
> 
> Yes, I just couldn't think of a good way.
> Basically the options would be
> 1) using Cocoa - lot of code, C and Objective-C interoperability annoyance, should be
>                 combined with cleanup of macosx vo

vo_corevideo's window display code should work. I don't know how to keep the shared buffer feature on 10.5, but on 10.6 it could use IOSurface which should be faster than the current method with mmap.

> 2) using Qt - needs C++...
> 3) using Gtk - not exactly the perfect choice for OSX
> 
>> It also doesn't detect texture_non_power_of_two:
>> [gl] Settings after autodetection: ati-hack = 0, force-pbo = 0, rectangle = 0, yuv = 2
>> Or is that intentional?
> 
> You misunderstand that, it does not list the detected supported features, it lists
> the settings is chose automatically.
> texture_non_power_of_two is less compatible and (with one exception) not faster, sometimes
> even slower.
> That one exception (well, that I know of) is the OpenSource R200 drivers, for which it
> is enabled automatically.

After some benchmarks on nvidia they all seem to be about the same speed, so I guess it's fine.