[MPlayer-dev-eng] Mac OS X native audio output

[Dan Christiansen]

On Thursday, May 15, 2003, at 08:47 PM, Alex Beregszaszi wrote:

> Hi,
>
> Good work!
> But don't forget to fill out the audio_resume() function!
I have filled it out ;) But maybe I'll move to using it instead of the 
started boolean.

> And please avoid using much global variables, or at least put the
> 'static' prefix there.
libao2 (the xiph one) originally used a struct, I'll move to using that 
so it'll fit better into the g2 structure.

> Anyway, you should forward this mail to the -dev-eng!
I don't understand, isn't that where I sent it?


On Thursday, May 15, 2003, at 08:45 PM, Arpi wrote:

> if your audio system/layer doesn't have such limitation, set it to 
> some nice
> value, 4096 or 8192 is a good start.

OK, I'll set it to the buffer size CoreAudio requests (4096 on my 
machine), or maybe twice just in case. (I have seen many buffer 
underruns but no buffer overruns.) I don't know if this is known 
behaviour, but setting it to zero doesn't seem to cause any problems.

Anyway, Arpi, could you consider licensing the ring buffer under the 
LGPL? It would be really neat to be able to use this in FFmpeg and/or 
XMMS. Xiph.org libao is GPLed too, but those parts would be easier to 
replace. (I'm beginning to understand the ring buffer, but it's a 
pretty complex machinery.)

Another thing, is there I way I can tell mplayer that I don't support 
volume? There is no direct support for it in the layers of CoreAudio I 
use. If af_volume can do it anyway, I don't see a reason for me writing 
yet another hive for bugs ;)