Other (audio or video) demux/codecs than real demux can use it

...as I found a case where it makes the output almost unreadable, and
cannot find another test case where it makes it better, which I'm
pretty sure I had when I did this.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Pointed-out-by: Alexis Ballier

This is consistent with Qt4 behaviour. This also fixes a bug when using
the configuration; the item is called "vcd", not "vcd://".

This avoids pushing and checking the mode parameter at each recursion.
Also, FindParent is now iterative.

HAVE_{LIBCDIO,LIBVCDINFO} and have_{libcdio,libvcdinfo} were nowhere
used in the source except when --enable-vcdx was given, hence only probe
for them when vcdx is to be enabled.
Since have_libvcdinfo (which is a requirement for vcdx) can only be
enabled if have_libcdio was defined, I simply put it in the enable_vcdx
block. And because vcdx needs to be explicitly enabled, we can safely
AC_MSG_ERROR on every missing library.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

This reverts commit 09049343.

Sorry, I forgot to push that part

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

Add a helper to get the right values from the RTSP code

Calls to aout_PrepareFormat where missing.

 It fixes segfaults in src/input/stream.c (at least) when demuxers
overflow int64_t seek position, and avoid testing for negative values
everywhere.
 stream_Tell() and stream_Size() still returns signed values as too much code
depend on it.

They were dead long enough.

This codec wasn't good.
It didn't work on SIPR correctly in all cases.
The code had lots of warnings.
The philosophy of it sucks (opening dlls...)
It didn't work on non-standard architecture.