Steve Lhomme authored Mar 10, 2017 and Jean-Baptiste Kempf committed Mar 10, 2017

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

Steve Lhomme authored Mar 10, 2017 and Jean-Baptiste Kempf committed Mar 10, 2017

The modules are always loaded with an absolute path. So we should not
need anything else to load modules properly with their system libraries.

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

Correct misleading comments and help messages

Avoid mismatch of signature descrambling call

Just for the sake of consistency (with Android).

Popup is not needed, because macOS does not have libav-
based HW accels anymore.

Also redo some autolayout rules.

fixes #18095

In theory, it should not affect us.

And remove a left over libvlc.dll.manifest being packaged

They are now correctly embeded

RT_MANIFEST is a macro that wasn't being substituted
2 was the magic number that corresponds to
ISOLATIONAWARE_MANIFEST_RESOURCE_ID, which should be used for DLLs.
Instead, use CREATEPROCESS_MANIFEST_RESOURCE_ID as we are creating a
process

There is no way to know if a device supports a specific codec using the
ENCODING_IEC61937 audio format.

There is an API for that, starting Android 21. But unfortunately, you need a
Context and to implement an interface. This is not possible via JNI (except if
we depend on a proxy Java class).

Therefore, we let the Android LibVLC listening to audio devices events and
report new supported encodings via libvlc_audio_output_device_set().

"pcm" should be used instead of "stereo" starting a recent Android Version
(like 21).