Also, don't reset pts interpolator at discontinuity
Signed-off-by: David Flynn <davidf@rd.bbc.co.uk>
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

timestamp will be interpolated at output of decoder instead.
Signed-off-by: David Flynn <davidf@rd.bbc.co.uk>
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: David Flynn <davidf@rd.bbc.co.uk>
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: David Flynn <davidf@rd.bbc.co.uk>
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>

Noticed by fenrir.

Close blocker #2723

It seems that ffmpeg care about VP30 != VP31 fourcc.

Audioscrobbler: activate libre.fm capacity since last.fm is geing more and more on the dark side... :D

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

This was a bit sloppy. In some cases, we had the config fiel lock,
but that that does not protect against config_Put*(). In some cases,
the lock was only taken for strings but not float/integers.

Another VLC instance could get a corrupt configuration otherwise.

This leaves the choice for bindings to use it or not.

I wish we did the same for hotkeys and inhibit but it would break
backward libvlc API compatibility.

It will avoid memory corruptions in case of a broken decoder.

It probably closes #2719.