The biggest landmark of thread-unsafety.

With the introduction, initially for LUA SD, of the chain syntax to
the services discovery, there is no way to canonicalize the name of
a services discovery object. Thus, they cannot be compared reliably,
and therefore, playlist_IsServicesDiscoveryLoaded() does not operate
correctly. I won't remove it today because too many interfaces call it.
But you should know that this function does _not_ actually work, and I
see no way to fix it (other than remove LUA SD).

Futhermore, this function is fundamentally prone to a race condition,
when it is used to toggle an SD (this is admittedly a minor problem).

This patch removes the existing interface and replaces it with a cleaner more
user-friendly interface.

Screenshots are available here: http://www.archmageinc.com/vlcSigned-off-by: Jean-Baptiste Kempf <jb@videolan.org>

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

And the corresponding ui theme
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

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

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

Seriously, this is WTF-code

It is a bit ridiculous to have 34 translations needed for such a small
module...

This ensures we use the installed libvlc(core) run-time, and might
fix some obscure crashes.

Fix #4822 and #4823.

There was only release() before.
Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

(See also previous commit)

As far as I understand, this breaks C++ object scoping. qtu() creates
a temporary anonymous QByteArray from its QString parameter. That array
is destroyed at the end of the expression, thus the return value of
qtu() becomes a dangling pointer.