Manual forward-port of [41c69be0]

We don't want that anymore anyway

We parse the additional HTML5 section too, the only place where the
WebM formats are listed. If the URL contains "&html5=True", the WebM
formats are preferred, when available.

Normally, the window will go on top of the stack of normal windows
anyway. If not, we may need to force this, which is different from
always-on-top in any case.

Known issue: toggling video on top while in fullscreen is not
implemented yet.

This also saves one HTTP redirect

This saves a few mutexes, but it does not really solve gcrypt
reentrancy. gcry_control() should only be called once for all per copy
of gcrypt in the memory. This is hard to conciliate with dlopen().

The Fnordlicht-Class implements only parts of the fnordlicht-protocol:
sync, stop, start_bootloader, boot_enter_application and fade_rgb
Have a look at http://github.com/fd0/fnordlicht/raw/master/doc/PROTOCOL
for more details.
Signed-off-by: André Weber <atmo@videolan.org>

Now epg-content should update automaticly when user changes channel etc.

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

Compiling qt on win32 is still long and painful

Nnot every options are implemented right now, only the most wanted.

This extension is in French only.
A german version may appear soon.

It will be hard to convince all Mozilla plugins-capable browsers to
call XInitThreads() early enough, if it ever happens. In the mean time,
make sure we use only pure XCB and avoid Xlib in VLC plugins.

Note: I did not check that the web plugins itself was not using Xlib
outside the main Mozilla browser thread.

This should fix #3662. We loose GLX and PulseAudio inside Mozilla.

Certain LibVLC applications fail to call XInitThreads() before they
call XOpenDisplay(). Then VLC calls XInitThreads(). Then the
applications call XCloseDisplay(), which raises a segmentation fault.
In this case, Xlib tries acquire lock that was never created as the
Display ppinter was created before threaded Xlib mode was enabled.

These applications can now pass --no-xlib to libvlc_new(). This will
prevent any VLC Xlib-based plugin from being used. Currently, this
affects interfaces (not really an issue), PulseAudio (until Colin's
patch is merged upstream) and GLX (Xlib-based by design).

This will be necessary to address #3662.

Description of the bug:

 This freeze happened only on Windows (tested on XP & Wine), but not on
 Linux. If I only moved up the "killInstance()" line, then the freeze
 would happen only on Linux but not on Windows.
 In these cases the QDialog's destructor never returned.

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

We could use QPixmapCache here too, but I don't think it speeds up it
any noticeable way, just makes code littlebit longer.