The input resource creates the aout instance with the input manager as
its parent object.

If this fail, the aout input should not be created at all. This removes
the fallback to an aout instance with the decoder object as parent.
This would potentially crash as the decoder is shorter-lived than the
aout instance.

This enables use of PulseAudio flat volume, if enabled in PulseAudio.

Note that, in that case, VLC will multiply its own volume by the output
device ("sink" in PulseAudio parliance) base volume. That effectively
maps 100% volume in VLC to +0dB in PulseAudio. To me, that seems more
intuitive and backward compatible with older VLC versions.

However, PulseAudio people seem to think that 100% should be mapped to
the maximum hardware amplification without software amplification,
that is to say PA_VOLUME_NORM.

Known limitations:

* The initial volume is determined by PulseAudio. However VLC shows 100%
regardless, which is not necessarily correct.

* If the VLC stream/sink input volume is changed outside of VLC (by a
mixer application), the VLC user interface will not see the updated
value.

No functional changes. Some more can't be replaced yet, as we still support 10.5

300 743 v1.2.1 mentionned '1111 1111', but 1.3.1 mentions '11 1111' only.
I believe it is a typo, (maybe, maybe not) but the spec is out, so people
will write encoders following it.
So, we do not look at the 2 bits of strong weight.

Thanks for the hint.

there may be to be investigated problems with non-ASCII keys

This provides more stable names especially for hot-plugged devices,
and more evocative descriptions for the user. As an added bonus, the
code is more compact.

Note however that this is not thread-safe. This is a common problem for
configuration items with update callbacks.

Signed-off-by: Laurent Aimar <fenrir@videolan.org>

This should fix practically glitches. The resulting A/V synchronization
is not too bad for me inspite of the long buffer (1/2 seconds).
Your mileage may vary.

Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

Implementation of fork() is imperfect on OS/2.

Pointed-out-by: KO Myung-Hun

There is no implementation of clock_gettime() on OS/2.
Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

CMSG_NXTHDR requires this.
Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

Closes #4646

Audio and video decoding works again. And now, it should not anymore
crash with --sout-keep. However, due to the current limitations of
input_DecoderCreate(), subpictures, statistics and input events are
broken. Note that hotkeys, the Qt audio menu and the Qt video menu do
 depend on input events :/

Since input_clock_t cannot be created/used from plugins, the extra
functionality of input_DecoderNew() from the new input_DecodeCreate()
is not interesting in plugins.

(Yes, I know I suck as function names)

This still requires input resources. But that should not be a problem
as the needed functions are already exposed (for libvlc).

Statistics, subpictures and input events do not work when a decoder is
created in this manner.