vlc.git
3 years agoRemove Direct2D module
Jean-Baptiste Kempf [Tue, 22 Nov 2016 20:50:01 +0000 (21:50 +0100)]
Remove Direct2D module

3 years agovdpau: remove invalidated comment
Rémi Denis-Courmont [Tue, 22 Nov 2016 21:04:38 +0000 (23:04 +0200)]
vdpau: remove invalidated comment

3 years agodirect3d11: deduce the ID3D11Device from ID3D11DeviceContext on winstore builds
Steve Lhomme [Tue, 22 Nov 2016 12:23:19 +0000 (13:23 +0100)]
direct3d11: deduce the ID3D11Device from ID3D11DeviceContext on winstore builds

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: the flat Vertex Shader doesn't need any constants
Steve Lhomme [Tue, 22 Nov 2016 10:59:44 +0000 (11:59 +0100)]
direct3d11: the flat Vertex Shader doesn't need any constants

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: use the projection Vertex Shader for non-flat sources
Steve Lhomme [Tue, 22 Nov 2016 10:59:43 +0000 (11:59 +0100)]
direct3d11: use the projection Vertex Shader for non-flat sources

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: update the projection constants when the viewpoint changes
Steve Lhomme [Tue, 22 Nov 2016 10:59:42 +0000 (11:59 +0100)]
direct3d11: update the projection constants when the viewpoint changes

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: add a Vertex Shader using a projection
Steve Lhomme [Tue, 22 Nov 2016 10:59:41 +0000 (11:59 +0100)]
direct3d11: add a Vertex Shader using a projection

The projection matrices are passed via a constant buffer.

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: display equirectangle sources in a sphere
Steve Lhomme [Tue, 22 Nov 2016 10:59:40 +0000 (11:59 +0100)]
direct3d11: display equirectangle sources in a sphere

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: split the code to create the vertex/index buffers for each quad
Steve Lhomme [Tue, 22 Nov 2016 10:59:39 +0000 (11:59 +0100)]
direct3d11: split the code to create the vertex/index buffers for each quad

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agovdpau/avcodec: remove backward compatibility code
Rémi Denis-Courmont [Tue, 22 Nov 2016 19:56:31 +0000 (21:56 +0200)]
vdpau/avcodec: remove backward compatibility code

VLC requires a newer libavcodec version anyway.

3 years agodemux/mp4: adjust to changes introduced by 8f9b76e
Filip Roséen [Thu, 17 Nov 2016 04:21:09 +0000 (05:21 +0100)]
demux/mp4: adjust to changes introduced by 8f9b76e

ExtractIntlStrings is only applicable for a limited range of boxes,
though with the recent changes following its introduction it might be
invoked for entities it is currently not protected from misdiagnosing.

These changes make sure that we only interpret a box inside the
function if:

 - the atom is known as a metadata box
 - the box has been parsed
 - the box has a father, and it is ATOM_udta

The check so that we do not parse boxes that contain a "data"-box has
been removed given that this check is more suitable at where the call
to the function is made.

Signed-off-by: Francois Cartegnie <fcvlcdev@free.fr>
3 years agodemux/mp4: fixup of f695cf0
Filip Roséen [Thu, 17 Nov 2016 04:21:10 +0000 (05:21 +0100)]
demux/mp4: fixup of f695cf0

The mentioned commit inaccurately assumed that every path within
SetupmdirMeta requires the "metadata" box to have a child of type
"data", which is not the case.

These changes simply changes the behavior so that we invoke the
correct parsing mechanism when it is applicable.

Signed-off-by: Francois Cartegnie <fcvlcdev@free.fr>
3 years agodemux/mp4: make sure we have enough data remaining in ExtractIntlStrings
Filip Roséen [Thu, 17 Nov 2016 04:21:11 +0000 (05:21 +0100)]
demux/mp4: make sure we have enough data remaining in ExtractIntlStrings

The previous implementation would assume that we would always have at
least 4 bytes of data left in the pending buffer, which can cause us
to read more bytes than available inside the loop.

This change make sure that we do not continue parsing if this happens.

Signed-off-by: Francois Cartegnie <fcvlcdev@free.fr>
3 years agodemux/mp4: fix handling of com.apple.quicktime metadata
Filip Roséen [Thu, 17 Nov 2016 04:21:13 +0000 (05:21 +0100)]
demux/mp4: fix handling of com.apple.quicktime metadata

The previous implementation would ignore metadata keys unless there
was a mapping to both AppleNameToMeta, and AtomXA9ToMeta, which is of
course erroneous.

Signed-off-by: Francois Cartegnie <fcvlcdev@free.fr>
3 years agovdpau/display: fix subpicture leak on error
Rémi Denis-Courmont [Tue, 22 Nov 2016 17:37:27 +0000 (19:37 +0200)]
vdpau/display: fix subpicture leak on error

3 years agovdpau: fix double free on error
Rémi Denis-Courmont [Tue, 22 Nov 2016 17:09:36 +0000 (19:09 +0200)]
vdpau: fix double free on error

3 years agocodec: gstreamer: fix build
Thomas Guillem [Tue, 22 Nov 2016 09:01:54 +0000 (10:01 +0100)]
codec: gstreamer: fix build

3 years agodvb: downgrade error messages
Rémi Denis-Courmont [Mon, 21 Nov 2016 21:24:53 +0000 (23:24 +0200)]
dvb: downgrade error messages

The old DVB plugin generate syntax error messages on valid MRLs that
the normal DTV plugin just fails to open (e.g. due to missing tuner).
This has proven very confusing.

3 years agocodec: svcdsub: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:50:21 +0000 (20:50 +0100)]
codec: svcdsub: use video_format_Init/Clean

3 years agocodec: subsusf: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:48:33 +0000 (20:48 +0100)]
codec: subsusf: use video_format_Init/Clean

3 years agocodec: spudec: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:40:58 +0000 (20:40 +0100)]
codec: spudec: use video_format_Init/Clean

3 years agocodec: kate: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:37:29 +0000 (20:37 +0100)]
codec: kate:  use video_format_Init/Clean

3 years agocodec: dvdsub: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:33:43 +0000 (20:33 +0100)]
codec: dvdsub: use video_format_Init/Clean

3 years agocodec: cvdsub: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:32:56 +0000 (20:32 +0100)]
codec: cvdsub: use video_format_Init/Clean

3 years agoaccess: vnc: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:31:19 +0000 (20:31 +0100)]
access: vnc: use video_format_Init/Clean

3 years agovfilter: mosaic: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:29:06 +0000 (20:29 +0100)]
vfilter: mosaic: use video_format_Init/Clean

3 years agovfilter: erase: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:22:42 +0000 (20:22 +0100)]
vfilter: erase: use video_format_Init/Clean

3 years agovfilter: alphamask: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:21:22 +0000 (20:21 +0100)]
vfilter: alphamask: use video_format_Init/Clean

3 years agotext_renderer: svg: use video_format_Init/Clean
Francois Cartegnie [Mon, 21 Nov 2016 19:19:05 +0000 (20:19 +0100)]
text_renderer: svg: use video_format_Init/Clean

3 years agocodec: gstreamer: fix unsafe copy of video_format_t
Francois Cartegnie [Mon, 21 Nov 2016 17:52:58 +0000 (18:52 +0100)]
codec: gstreamer: fix unsafe copy of video_format_t

3 years agoM3U export: fix URLs with non-ASCII characters (fixes #15455)
Rémi Denis-Courmont [Mon, 21 Nov 2016 20:14:23 +0000 (22:14 +0200)]
M3U export: fix URLs with non-ASCII characters (fixes #15455)

3 years agoinput: remove input_item_node_t.pf_compar
Rémi Denis-Courmont [Mon, 21 Nov 2016 19:17:37 +0000 (21:17 +0200)]
input: remove input_item_node_t.pf_compar

3 years agoinput: remove constant input_item_node_t.b_can_loop
Rémi Denis-Courmont [Mon, 21 Nov 2016 19:13:25 +0000 (21:13 +0200)]
input: remove constant input_item_node_t.b_can_loop

3 years agoVideoToolbox: fix wrong function type check
Jean-Baptiste Kempf [Mon, 21 Nov 2016 16:48:41 +0000 (17:48 +0100)]
VideoToolbox: fix wrong function type check

3 years agopacketizer: h264: fix fps computation (fields/s vs fps)
Francois Cartegnie [Mon, 21 Nov 2016 15:15:48 +0000 (16:15 +0100)]
packetizer: h264: fix fps computation (fields/s vs fps)

3 years agovout: decklink: fix mode fps debug
Francois Cartegnie [Mon, 21 Nov 2016 14:32:18 +0000 (15:32 +0100)]
vout: decklink: fix mode fps debug

3 years agovout: decklink: debug field dominance
Francois Cartegnie [Mon, 21 Nov 2016 14:07:07 +0000 (15:07 +0100)]
vout: decklink: debug field dominance

3 years agovout: decklink: use 10bits output as default
Francois Cartegnie [Sat, 19 Nov 2016 19:30:55 +0000 (20:30 +0100)]
vout: decklink: use 10bits output as default

3 years agovout: decklink: fix format auto selection
Francois Cartegnie [Thu, 17 Nov 2016 19:08:23 +0000 (20:08 +0100)]
vout: decklink: fix format auto selection

3 years agodecklink output: use variable
Rafaël Carré [Fri, 11 Apr 2014 08:44:49 +0000 (10:44 +0200)]
decklink output: use variable

Signed-off-by: Francois Cartegnie <fcvlcdev@free.fr>
3 years agodirect3d11: compile the vertex shader only once
Steve Lhomme [Mon, 21 Nov 2016 09:03:05 +0000 (10:03 +0100)]
direct3d11: compile the vertex shader only once

also set the vertex shader layout only once

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: rename globVertexShaderDefault to globVertexShaderFlat
Steve Lhomme [Mon, 21 Nov 2016 09:03:04 +0000 (10:03 +0100)]
direct3d11: rename globVertexShaderDefault to globVertexShaderFlat

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: pass the opacity value as a constant to the pixel shader
Steve Lhomme [Mon, 21 Nov 2016 09:03:03 +0000 (10:03 +0100)]
direct3d11: pass the opacity value as a constant to the pixel shader

it is not needed in the vertex shader

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agodirect3d11: use a different vertex shader per quad
Steve Lhomme [Mon, 21 Nov 2016 09:03:02 +0000 (10:03 +0100)]
direct3d11: use a different vertex shader per quad

The SPU need the flat vertex shader but others may need different shaders.

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agovideotoolbox: assure that the vout is configured before creating a pic
Thomas Guillem [Mon, 21 Nov 2016 16:27:44 +0000 (17:27 +0100)]
videotoolbox: assure that the vout is configured before creating a pic

This fixes a regression from 47467b9628e4776164164ab9c6caadf3b36c2408

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agoes_format: remove useless init
Thomas Guillem [Mon, 21 Nov 2016 15:24:14 +0000 (16:24 +0100)]
es_format: remove useless init

p_palette is already initialized from the memset

3 years agoes_format: init video_format from es_format_Init
Thomas Guillem [Mon, 21 Nov 2016 15:20:32 +0000 (16:20 +0100)]
es_format: init video_format from es_format_Init

3 years agoftp: handle preliminary (1xx) reply systematically
Rémi Denis-Courmont [Sun, 20 Nov 2016 17:28:21 +0000 (19:28 +0200)]
ftp: handle preliminary (1xx) reply systematically

3 years agoftp: factor common line receive code
Rémi Denis-Courmont [Sun, 20 Nov 2016 17:17:08 +0000 (19:17 +0200)]
ftp: factor common line receive code

3 years agoQt: remove dead SD mapper and menu code
Rémi Denis-Courmont [Sun, 20 Nov 2016 14:44:46 +0000 (16:44 +0200)]
Qt: remove dead SD mapper and menu code

3 years agolua: fix playlist locking
Rémi Denis-Courmont [Sun, 20 Nov 2016 14:31:48 +0000 (16:31 +0200)]
lua: fix playlist locking

3 years agosd: split generic and playlist code to different modules
Rémi Denis-Courmont [Sun, 20 Nov 2016 14:31:27 +0000 (16:31 +0200)]
sd: split generic and playlist code to different modules

3 years agowin32: do not pretend to implement ftrylockfile()
Rémi Denis-Courmont [Sun, 20 Nov 2016 09:48:16 +0000 (11:48 +0200)]
win32: do not pretend to implement ftrylockfile()

This is not currently used anywhere in the code base anyway.

3 years agoWin32: remove outdated change-contribs-directory.sh script
Jean-Baptiste Kempf [Sat, 19 Nov 2016 22:45:35 +0000 (23:45 +0100)]
Win32: remove outdated change-contribs-directory.sh script

3 years agoWin32: manifest can be 0644
Jean-Baptiste Kempf [Sat, 19 Nov 2016 22:44:28 +0000 (23:44 +0100)]
Win32: manifest can be 0644

3 years agoMakefile: Move win32 and MacOS specific files in their folder
Jean-Baptiste Kempf [Sat, 19 Nov 2016 22:35:58 +0000 (23:35 +0100)]
Makefile: Move win32 and MacOS specific files in their folder

3 years agoRemove outdated altlinux spec file
Jean-Baptiste Kempf [Sat, 19 Nov 2016 22:28:43 +0000 (23:28 +0100)]
Remove outdated altlinux spec file

3 years agoios2 vout: fix the inverted video orientation
Adrien Maglo [Sat, 19 Nov 2016 22:07:40 +0000 (23:07 +0100)]
ios2 vout: fix the inverted video orientation

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agoiOS build: forbid daemon and fork calls
Jean-Baptiste Kempf [Sat, 19 Nov 2016 21:58:48 +0000 (22:58 +0100)]
iOS build: forbid daemon and fork calls

3 years agocpu check: use HAVE_FORK where applicable
Jean-Baptiste Kempf [Sat, 19 Nov 2016 21:29:29 +0000 (22:29 +0100)]
cpu check: use HAVE_FORK where applicable

3 years agoncurses: fix root node tracking
Rémi Denis-Courmont [Sat, 19 Nov 2016 21:14:33 +0000 (23:14 +0200)]
ncurses: fix root node tracking

3 years agoncurses: remove nonsensical pl_lock
Rémi Denis-Courmont [Sat, 19 Nov 2016 21:07:16 +0000 (23:07 +0200)]
ncurses: remove nonsensical pl_lock

This lock "protects" two things:
 - The need-update boolean which can more easily be protected by the
   playlist lock.
 - The "node", whose lifetime is not correctly handled anyway.

3 years agoncurses: fix exit and clean-up
Rémi Denis-Courmont [Sat, 19 Nov 2016 17:20:31 +0000 (19:20 +0200)]
ncurses: fix exit and clean-up

Make cancellation safe and simplify.

3 years agoncurses: identify playlist item by input item
Rémi Denis-Courmont [Sat, 19 Nov 2016 17:01:20 +0000 (19:01 +0200)]
ncurses: identify playlist item by input item

This fixes a whole lot of use-after-free bugs.

3 years agoncurses: missing lock
Rémi Denis-Courmont [Sat, 19 Nov 2016 16:30:31 +0000 (18:30 +0200)]
ncurses: missing lock

3 years agoosx_notification: Remove "item-change" callback
Marvin Scholz [Thu, 17 Nov 2016 21:53:21 +0000 (22:53 +0100)]
osx_notification: Remove "item-change" callback

Notify as soon as the current input item is changed,
as it is not a good idea to rely on an item-change
eventually happening later.
This will result in some items that are added and played
immediately lacking metadata in their notifications but
this seems like an acceptable edge case.

3 years agoplaylist: remove constant playlist_Add() parameter
Rémi Denis-Courmont [Sat, 19 Nov 2016 15:24:31 +0000 (17:24 +0200)]
playlist: remove constant playlist_Add() parameter

3 years agoplaylist: assert lock on (non-NULL) VIEWPLAY control
Rémi Denis-Courmont [Sat, 19 Nov 2016 15:15:43 +0000 (17:15 +0200)]
playlist: assert lock on (non-NULL) VIEWPLAY control

3 years agoplaylist: remove unnecessary shortcuts
Rémi Denis-Courmont [Sat, 19 Nov 2016 15:11:34 +0000 (17:11 +0200)]
playlist: remove unnecessary shortcuts

Controls can be processed if the playlist is empty. They just do
nothing. Conversely, stop need not be processed if the playlist was
already killed.

3 years agoncurses: only append to PL or ML
Rémi Denis-Courmont [Sat, 19 Nov 2016 15:09:15 +0000 (17:09 +0200)]
ncurses: only append to PL or ML

User interfaces should only add new items under the playlist or the
media library nodes. In particular, they have no business adding or
removing SD items.

3 years agoplaylist: remove useless intf-change object variable
Rémi Denis-Courmont [Sat, 19 Nov 2016 14:41:53 +0000 (16:41 +0200)]
playlist: remove useless intf-change object variable

3 years agoplaylist: remove position parameter to playlist_Add*
Rémi Denis-Courmont [Sat, 19 Nov 2016 11:26:21 +0000 (13:26 +0200)]
playlist: remove position parameter to playlist_Add*

This removes the insertion offset parameter to playlist_Add(),
playlist_AddExt() and playlist_AddInput().

intf_InsertItem() was the only occurrence where the offset was zero,
a side effect of how the command line is parsed backward. This is now
done explicitly with playlist_NodeAddInput().

A non-zero positive offset made no sense, since:
 - the number of children of a node can change asynchronously while the
   playlist lock,
 - the back-end asserted or triggered undefined behaviour if the offset
   was out of range.
DBus was the only occurrence of this bug (see also #17451 comment:4),
and has been fixed in an earlier change.

In all remaining call sites, the offset was PLAYLIST_END. This removes
the parameter which would otherwise constitute a useless constant.

3 years agointerface: clean up intf_InsertItem()
Rémi Denis-Courmont [Sat, 19 Nov 2016 11:19:06 +0000 (13:19 +0200)]
interface: clean up intf_InsertItem()

 - Handle and report errors.
 - Fix and improve documentation.

3 years agoplaylist: remove constant duration parameter to playlist_AddExt()
Rémi Denis-Courmont [Sat, 19 Nov 2016 10:59:11 +0000 (12:59 +0200)]
playlist: remove constant duration parameter to playlist_AddExt()

3 years agodbus: fix racy track addition (refs #17451)
Rémi Denis-Courmont [Sat, 19 Nov 2016 10:51:38 +0000 (12:51 +0200)]
dbus: fix racy track addition (refs #17451)

This fixes potential assertion failures. Insertion position must always
be valid.

3 years agoplaylist: remove useless parameter
Rémi Denis-Courmont [Sat, 19 Nov 2016 10:04:08 +0000 (12:04 +0200)]
playlist: remove useless parameter

3 years agoplaylist: remove unused NO_REBUILD flag
Rémi Denis-Courmont [Sat, 19 Nov 2016 10:01:47 +0000 (12:01 +0200)]
playlist: remove unused NO_REBUILD flag

The flag is sort-of-implemented, but never actually used.

3 years agoplaylist: remove APPEND and INSERT flags
Rémi Denis-Courmont [Sat, 19 Nov 2016 09:59:01 +0000 (11:59 +0200)]
playlist: remove APPEND and INSERT flags

They are set but never used. In practice, the (following) position
parameter determines whether the item is inserted or appended. A
positive position means insertion; PLAYLIST_END means appendment.

3 years agoplaylist: remove no-op PREPARSE and SPREPARSE flags
Rémi Denis-Courmont [Sat, 19 Nov 2016 09:52:58 +0000 (11:52 +0200)]
playlist: remove no-op PREPARSE and SPREPARSE flags

PLAYLIST_SPREPARSE is not used anywhere. PLAYLIST_PREPARSE is set but
has no effects whatsoever.

3 years agolua: remove cargo-cultivated cancellation
Rémi Denis-Courmont [Fri, 18 Nov 2016 22:26:02 +0000 (00:26 +0200)]
lua: remove cargo-cultivated cancellation

3 years agoskins2: remove playlist pointer copy and simplify
Rémi Denis-Courmont [Fri, 18 Nov 2016 22:06:17 +0000 (00:06 +0200)]
skins2: remove playlist pointer copy and simplify

3 years agoplaylist: remove playlist_item_t.p_playlist
Rémi Denis-Courmont [Fri, 18 Nov 2016 20:39:18 +0000 (22:39 +0200)]
playlist: remove playlist_item_t.p_playlist

3 years agoplaylist: remove use of playlist_item_t.p_playlist in callbacks
Rémi Denis-Courmont [Fri, 18 Nov 2016 20:33:00 +0000 (22:33 +0200)]
playlist: remove use of playlist_item_t.p_playlist in callbacks

3 years agoplaylist: delete array earlier during destruction
Rémi Denis-Courmont [Fri, 18 Nov 2016 20:29:20 +0000 (22:29 +0200)]
playlist: delete array earlier during destruction

Remove items from the arrays takes time, especially during playlist
destruction. In particular removing an item from the current array
requires a linear search and then a memmove(); that was the bottleneck
with quadratic complexity while deleting the playlist. (Now the
bottleneck is removing items from their parent.)

Destroying the arrays early on skips those steps entirely.

3 years agoplaylist: use search trees for ID and input to item mapping
Rémi Denis-Courmont [Fri, 18 Nov 2016 19:40:54 +0000 (21:40 +0200)]
playlist: use search trees for ID and input to item mapping

Regarding input item look-ups, this reduces asymptotic complexity from
linear to logarithmic time.

Regarding ID look-ups, this reduces insertion and deletion time to
logarithmic. Previously it degraded to linear time because of memcpy()
and memmove() in ARRAY_APPEND and ARRAY_REMOVE macros.

This removes the "all_items" array, and its missing error handlers.

Finally, this adds support for allocating more than INT_MAX items
during the entire lifetime of the VLC instance. (The maximum number of
_concurrent_ items is still INT_MAX, but memory would probably run out
before that is reached.)

Note: Item deletion still requires linear time. And playlist deletion
still consequently requires quadractic time because of the "current"
array.

3 years agoRevert "contrib: define CMAKE_SYSTEM_NAME when building for Android"
Jean-Baptiste Kempf [Fri, 18 Nov 2016 18:26:46 +0000 (19:26 +0100)]
Revert "contrib: define CMAKE_SYSTEM_NAME when building for Android"

This reverts commit 6290ebf68b5f59b7c1b272c242e62377fee4b557.

It behaves badly depending on the CMake Version, and on our buildbots...

3 years agowin32: don´t clobber process DLL directories (fixes #17663)
Rémi Denis-Courmont [Fri, 18 Nov 2016 18:13:34 +0000 (20:13 +0200)]
win32: don´t clobber process DLL directories (fixes #17663)

LibVLC has zero business messing with process state. This breaks
apps other than VLC and other libraries in the same process.

3 years agovlc: do not load kernel32.dll
Rémi Denis-Courmont [Fri, 18 Nov 2016 18:10:25 +0000 (20:10 +0200)]
vlc: do not load kernel32.dll

vlc.exe already depends on kernel32.dll, so there is no need to load it
explicitly. It can simply be looked up.

3 years agovlc: set default DLL directories
Rémi Denis-Courmont [Fri, 18 Nov 2016 18:07:58 +0000 (20:07 +0200)]
vlc: set default DLL directories

3 years agowin32: fix IDN support when building retardedly
Rémi Denis-Courmont [Fri, 18 Nov 2016 17:56:40 +0000 (19:56 +0200)]
win32: fix IDN support when building retardedly

3 years agowin32: don´t clobber process error mode
Rémi Denis-Courmont [Fri, 18 Nov 2016 17:46:27 +0000 (19:46 +0200)]
win32: don´t clobber process error mode

Regression from e44f6165c138a1e715b5dd738155f9eb6dc90143

3 years agoOpenGL: more const
Steve Lhomme [Thu, 17 Nov 2016 11:45:28 +0000 (12:45 +0100)]
OpenGL: more const

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agoOpenGL: fix typo in comment
Steve Lhomme [Thu, 17 Nov 2016 11:45:27 +0000 (12:45 +0100)]
OpenGL: fix typo in comment

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agoOpenGL: code cleaning
Steve Lhomme [Thu, 17 Nov 2016 11:45:26 +0000 (12:45 +0100)]
OpenGL: code cleaning

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agoOpenGL: the zoom translation depends on the sphere radius
Steve Lhomme [Thu, 17 Nov 2016 11:45:25 +0000 (12:45 +0100)]
OpenGL: the zoom translation depends on the sphere radius

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>
3 years agohotkeys: handle 360° zoom with mouse wheel
Thomas Guillem [Thu, 17 Nov 2016 11:06:08 +0000 (12:06 +0100)]
hotkeys: handle 360° zoom with mouse wheel

3 years agohotkeys: handle the mouse events for VR/360° navigation
Thomas Guillem [Thu, 17 Nov 2016 11:05:36 +0000 (12:05 +0100)]
hotkeys: handle the mouse events for VR/360° navigation

3 years agohotkeys: listen to new inputs and new vouts
Thomas Guillem [Wed, 16 Nov 2016 18:30:27 +0000 (19:30 +0100)]
hotkeys: listen to new inputs and new vouts

Listen to playlist "input-current" to update the current input_thead_t, and
listen to input "intf-event" to update the current vout. The vout/input don't
need to be fetched when processing an action, this removes one FIXME, but this
adds more lock complexity.

This will be mainly needed for the next commit, in order to listen to mouse
events from the vout.

3 years agocore: vout: notify when the viewpoint can change
Thomas Guillem [Wed, 16 Nov 2016 15:05:53 +0000 (16:05 +0100)]
core: vout: notify when the viewpoint can change

The boolean variable "viewpoint-changeable" will be used by interfaces in order to
know if 360° navigation should be handled.

3 years agolua/vlc: vlc_sd_probe_Open: fix memory-leak
Filip Roséen [Fri, 18 Nov 2016 08:02:30 +0000 (09:02 +0100)]
lua/vlc: vlc_sd_probe_Open: fix memory-leak

vlc_dictionary_all_keys requires the caller to not only release the
returned value, but also the entities this value refers to; these
changes fixes a memory-leak due to not doing the latter.

Signed-off-by: Pierre Ynard <linkfanel@yahoo.fr>