Tests show that current docs are wrong.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Audio output should be more efficient on FPU-less devices.
Firstly the mixer should be faster with 16-bits instead of 32.
Secondly FPU-less systems usually output at 16-bits, so one sample
format conversion between the mixer and the output is now avoided.
Thirdly conversion from input format to the mixer format should
similarly be completely avoided or at least accelerated.

With only one input per output, the lock is only useful:
 - to restart the audio instance (needed 3 out of 4 locks anyway),
 - to propagate volume change to the mixer or output plugin,
 - to access the FIFO from the output plugin thread (if applicable).

So 4 fine-grained was over-engineering. Most importantly, the locking
scheme was overly complicated and generally misunderstood/misused. Also
to avoid lock inversion, some unlocking/relocking sequences were
introduced; they broke atomicity.

We could certainly reduce the scope of the remaining lock. Since we
have one only input per output, most of the code is only ever run from
the decoder thread. Thus reentrancy is not anymore needed in some
places. But first aout_Restart() needs to be fixed and simplified.

This is required for PulseAudio synchronization and to pause/resume
with low-latency. This should also be useful for other buffered
 audio outputs such as ALSA or CoreAudio.

Fixes #5026
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

 - do not assume output pitch equals (double) pixel width
 - improve function prototypes
 - hand zero-width or zero-height corner cases in ASM (totally useless)
 - use ARM conditon flag (opS) as appropriate

This closes #5000
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Ref: #5000
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Since set_text can be called on a label, it seems to be an editable
label.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Some tools seem to generate multi-parts rar where only the first chunk
contains the actual file size.

If the event is ignored, the Qt4-X11 will reject the quit ICCM request.
This annoyingly caused VLC to inhibit logging out.

As we need to build contrib before running configure, this remains a
little bit awkward:
 # cd builddir
 # mkdir contrib
 # cd contrib
 # ../../contrib/bootstrap
 # make

Also get rid of "hosts/" in the install prefix. Because I can.

'make clean' removes those, so there is no point in creating them from
bootstrap. $(RECONF) takes care of this better already for aclocal.
Otherwise, install targets should deal with this.

curl fits Makefile dependencies better, and supports FTP through HTTP
proxies.