The aout can exist even if the input thread is stopped. This makes
audio-related functions work in case where they would previously fail.

Adjust the Android audio outputs to the latest aout core changes.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This reverts commit 346f385b.

Doesn't seem to work in all cases straight of the box

(instead of waiting for play)

If the playlist is activated, then input resources are almost certainly
going to be used. This is a bit simpler.

The stream might not exist.

Regression from e364643f.

In other words, keep the module loaded whenever the aout exists.

Signed-off-by: Rafaël Carré <funman@videolan.org>

This keeps track of available output devices. Again, without core
support yet, this is not yet useful in any way.

There is no core support for this as yet, so the crash cannot happen.

This concerns ALSA, AudioQueue, JACK and KAI.

This would enable the input manager to hold an audio output plugin at
all times. Then it would be possible to keep track of output devices,
volume, policy and (JACK) routing continuously. For this to work,
outputs will need to be patched so that device selection and volume
handling works even when playing back.

Note though that this commit only introduces the change in the
audio_output_t interface; it does not implement the features above.

it is not clear how glew(Active)ClientTexture should be acceeded,
do not use the broken glew 'shortcut' for these functions

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Seems to be used for RDP on Windows 8/2012 according to
http://msdn.microsoft.com/en-us/library/dd341376%28prot.20%29.aspx