leeexin/Velix-MPV-Core
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
55,516 Commits 2 Branches 89 Tags
harmony-port
Commit Graph
100 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
sfan5 63ada87ec7 ci/mingw: add second download URL for iconv and freetype 
It appears this time GNU broke their ftpmirror, while the primary one is working again.

download.savannah.gnu.org also seems to be having issues.
Never thought I'd see the day where sourceforge is more reliable.
 2026-07-31 15:56:00 +02:00
sfan5 3f1b23abd0 stream_curl: fix libavformat version checks 
Unfortunately ffmpeg missed the backport of <https://code.ffmpeg.org/FFmpeg/FFmpeg/commit/1a7e802c8f970e76807f08c3d3fb3513ae5a4882>,
so custom IO is still segfaulting on the older versions.
 2026-06-29 14:58:15 +02:00
sfan5 2339eb7276 stream_curl: extend version checks that enable curl by default 
...to include older ffmpeg patch relases where the nested IO cleanup bug is fixed
 2026-06-23 11:03:09 +02:00
sfan5 1fe37856cb ci: fix comment workflow permissions 2026-05-19 18:54:30 +02:00
sfan5 8fc1a10a44 ci: fix necessary permissions for cleanup_caches and comment 
This undoes de5edca29c,
because I failed to read the documentation.
 2026-05-18 21:13:09 +02:00
sfan5 de5edca29c ci: clarify that cleanup_caches and comment don't use token permissions 2026-05-18 20:53:29 +02:00
sfan5 ea12f8efc8 ci/mingw: add libcurl 2026-05-16 18:20:42 +02:00
sfan5 7b70999b62 ci/mingw: cache tar downloads 2026-05-13 20:45:41 +02:00
sfan5 003c5433f6 ci/mingw: switch to zlib-ng 
gotta go fast
 2026-05-10 00:02:19 +02:00
sfan5 373cfc6924 ci/mingw: refactor tar downloading 2026-05-10 00:02:19 +02:00
sfan5 24fbc99701 ci/mingw: some small cleanups and updates 
Note: zlib 1.3.2 breaks with -Werror=undef, so we don't update it.
<https://github.com/madler/zlib/issues/1169>
 2026-05-09 12:07:53 +02:00
sfan5 07c3ff3725 command: reject negative offset for overlay-add 2026-03-09 15:43:32 +01:00
sfan5 60dd40d007 test/img_format: remove dead code 2026-02-08 11:51:26 +01:00
sfan5 e01165d3db test/repack: test draw_bmp at some uneven sizes 
intended to discover bugs like f3c1dcf7a1
 2026-02-08 11:51:26 +01:00
sfan5 272404533e test: add reference files for ffmpeg 8 2026-02-08 11:51:26 +01:00
sfan5 b6512e1f8d wayland_common: improve probing behavior and logging 
Any errors when opening the display are now logged.
The WAYLAND_DISPLAY check now also logs and applies only when probing.
This means explicitly setting a wayland `--gpu-context` will always
try to open the display instead of "helpfully" refusing (in some cases).

see also: f6f1721101
 2026-02-08 01:56:01 +01:00
sfan5 d0e24ab17a Update MPV_VERSION 2025-12-21 20:14:18 +01:00
sfan5 41f6a64506 Release 0.41.0 2025-12-21 20:08:20 +01:00
sfan5 3a4e0fc200 ci/mingw: update dependency versions 2025-11-30 17:21:28 +01:00
sfan5 393b3fc115 ci/mingw: use a mirror instead of ftp.gnu.org 
it's apparently constantly down these days...
 2025-11-30 17:21:28 +01:00
sfan5 584b368972 demux_playlist: clarify reason for HLS detection 
see 0b1c3e8de2
 2025-11-30 17:21:28 +01:00
sfan5 2ad27be8be demux_lavf: remove 'bin' format from blacklist 
A probing function was implemented in ffmpeg 4.0:
<https://github.com/FFmpeg/FFmpeg/commit/8f6a71f7f13d21614f5140e672976e96dca04af8>
 2025-11-30 17:21:28 +01:00
sfan5 609acc7cf2 meson: warn if building with SDL2 
I previously tried to enable sdl2-gamepad by default and the CI turned all red.
In detail:
- SDL_config.h has macros that conflict with ours
- memory leaks caused the tests to fail
- main=SDL_main redefinition broke MSYS builds

The options remain disabled by default, but lets warn our users a bit.
 2025-10-31 12:53:04 +01:00
sfan5 8649b64e57 meson: remove redundant sdl2 option 
This disables the individual options by default, since they were gated by the parent option before.
 2025-10-31 12:53:04 +01:00
sfan5and茶茶 e23e72bd49 vo/context_android: fix surface tearing on resize 
It's not clear why this is needed, but it works (on my device).

fixes https://github.com/mpv-android/mpv-android/issues/991

Co-authored-by: 茶茶 <chachakko@outlook.com>
 2025-10-27 23:30:04 +01:00
sfan5 80d2a6a7ed ytdl_hook.lua: don't crash if playlist entries don't have an usable URL 
see #16717
 2025-08-23 15:02:46 +02:00
sfan5 194ce88ac3 player/command: fix missing property update for some hwdec changes 
fixes #16710
 2025-08-22 08:38:02 +02:00
sfan5 653e241ac3 DOCS/lua: fix mistyped function name 2025-08-22 08:38:02 +02:00
sfan5 521e162420 DOCS/input: clarify value of hwdec-current 
This can never be false, or a boolean at all.
 2025-08-22 08:38:02 +02:00
sfan5 866da80979 vo_gpu_next: fix resetting --glsl-shader-opts 
fixes: #16638
 2025-08-07 18:35:42 +02:00
sfan5 a6f3236c3a ytdl_hook.lua: don't pass any format to youtube-dl by default 
yt-dlp changed its default format from `bv+ba/b` to `bv*+ba/b`, which we want to follow
since it fixes the behavior with some streaming sites.

The note that "sometimes youtube-dl returns a format that mpv cannot use" was added in
43c13e5ea2 but it's not clear what it refers to.

fixes #16388
 2025-07-28 11:22:13 +02:00
sfan5 2168e3d82c ytdl_hook.lua: get rid of outdated youtube-dl hint 
It has never worked for yt-dlp anyway and the relevant error messages
already tell people to update.
 2025-07-28 11:22:13 +02:00
sfan5 6b178e4f5a misc/path_utils: create folders with 0777 mode 
This better matches other Unix tools.
Note that in practice the umask downgrades this to 0755 or 0775, so folders don't actually end up world-readable.
 2025-06-02 15:59:05 +02:00
sfan5 9db49973f2 misc/path_utils: check existence first in mp_mkdirp() 2025-06-02 15:59:05 +02:00
sfan5 013e76e1c0 player: create missing folders for watch-history-path 
fixes: #16357
 2025-06-02 15:59:05 +02:00
sfan5 c4a00b2c85 player: handle unavailable options while writing watch-later file 
fixes: #16410
 2025-06-02 15:59:05 +02:00
sfan5 7182991bc4 sub/lavc_conv: also log decoder used 2025-06-02 15:59:05 +02:00
sfan5 6c6e81314a DOCS/libmpv: clarify that handle used by cplugin cannot be destroyed 2025-06-02 15:59:05 +02:00
sfan5 0f26d9f1ee sub/sd_lavc: fix condition for probing output type 
This fixes decoding with libaribb24.
closes #16384
 2025-06-02 15:59:05 +02:00
sfan5 063a9b8c89 sd_lavc: log some useful information 2025-05-31 09:59:54 +02:00
sfan5 c0efee0512 video/image_writer: improve logging of conversions and encoder 2025-05-31 09:59:54 +02:00
sfan5 cf00b07536 various: use MP_ARRAY_SIZE macro 2025-05-31 09:59:54 +02:00
sfan5 e663cbc037 vo: move vo_mediacodec_embed to not be the default on Android 2025-05-03 18:13:35 +03:00
sfan5 08939665fe DOCS: clarify some minor points in the release policy 2025-03-28 16:19:48 +01:00
sfan5 97cb16d683 RELEASE_NOTES: remove accidental BOM 
linting is good and all but does this check really need to fail on every single commit from now on?
 2025-03-25 18:39:00 +01:00
sfan5 2da0b49038 Update MPV_VERSION 2025-03-25 18:36:18 +01:00
sfan5 e48ac7ce08 Release 0.40.0 2025-03-25 18:33:27 +01:00
sfan5 425037a16a DOCS: refresh some parts of the release policy 2025-03-25 17:57:53 +01:00
sfan5 839aa31b7e player/client: adjust log message for client exit 
The old message is too generic and worse could be mistaken
for exit of the entire player.
 2025-03-07 09:44:19 +01:00
sfan5 53a6a6821c player/scripting: add logging around script un- and reloading 2025-03-07 09:44:19 +01:00
sfan5 bfa3e90867 video/out/android: don't assert-fail if WinID is not set 
This matches the behavior of every other GPU context and is the only
behavior that makes sense.
 2025-02-11 19:19:36 +01:00
sfan5 3b586ba6c3 ci/mingw: update library versions 
also moves shaderc to latest git again
 2025-01-25 15:39:36 +01:00
sfan5 87826224a2 ci/mingw: restrict shaderc version to fix build errors 
https://github.com/KhronosGroup/glslang/issues/3787 was fixed already
but shaderc uses pinned revisions, so let's just downgrade.
 2024-11-07 22:39:38 +01:00
sfan5 9737b99845 Revert "editorconfig: punish tab usage" 
This reverts commit 3840c98b36.
There are multiple reasons:
- the correct way to prevent tabs being mistakenly added is a CI linting step
- it can affect the editing experience of foreign source files
  under mpv's tree (e.g. subprojects)
- at least one editor (BBEdit) misinterprets the values and also turns
  soft tabs into 7 spaces
- it does not consistently apply to GH code view either
 2024-10-20 11:22:09 +02:00
sfan5 c11239be8c options: enable handling --no-hwdec as --hwdec=no 
Reusing M_OPT_ALLOW_NO has the side-effect of applying to --ab-loop-[ab],
which makes sense.
 2024-09-30 11:26:13 +02:00
sfan5 b64c53f730 ci/lint-commit-msg: make some minor adjustments to the rules 2024-09-23 19:49:43 +02:00
sfan5 0e6d050c3b Update MPV_VERSION 2024-09-23 19:37:35 +02:00
sfan5 a0fba7be57 Release 0.39.0 2024-09-23 19:36:41 +02:00
sfan5 5e9e756f53 TOOLS/gen-interface-changes: small usability improvement 2024-09-23 19:20:05 +02:00
sfan5 3ab989e554 ci/mingw: update dependency versions 2024-07-16 23:04:36 +02:00
sfan5 d2e3e6f3bd test/libmpv_lifetime: don't re-define ASAN macro 2024-07-16 23:04:36 +02:00
sfan5 03ca340835 stream: remove directory playlist hint 
The stream layer is the wrong place to handle this and the benefit
seems dubious.
 2024-05-15 22:45:18 +02:00
sfan5 553fb024d4 stream: check file descriptor passed to fd:// or fdclose:// 2024-05-15 22:45:18 +02:00
sfan5 54c755b08d various: add missing checks for directory streams 
All of this code opens a stream and expects to read stuff from a file.
But streams can also be directories and that has to be handled.
 2024-05-15 22:45:18 +02:00
sfan5 b69b58a12a {options,player}: fix stream leaks 2024-05-15 22:45:18 +02:00
sfan5 ae23556b6f stream: disallow reading or writing to directories 
Reading an entire file is a common operation, meanwhile
stream_file will happily open a directory. This doesn't match well.

Analogously for open_output_stream().
 2024-05-15 22:45:18 +02:00
sfan5 a33fd3b4cf Revert "stream: don't try to read from all-sparse/no-data files" 
The stream layer is just not the right place to make this change
since it's also used for completely unrelated purposes such as reading
configs.

This reverts commit bb7a485c09.
 2024-05-12 18:06:21 +02:00
sfan5 c4b6d0da38 f_lavfi: reject filters with too many in/outs 
The pad count serves as a minimum even if the filter has
dynamic inputs/outputs, so we should reject them here.

At least on FFmpeg 6.1 this results in no practical change
in the list accepted filters.

fixes: 8bf3fe7e2a
 2024-05-04 16:13:03 +02:00
sfan5 943a50b300 ci/mingw: update dependency versions 2024-05-03 23:44:51 +02:00
sfan5 426be8441a test: make libmpv_test abort if an error is logged 
Among others this provides a trivial test that built-in scripts aren't throwing
an error at load or init time.
 2024-04-24 12:58:50 +02:00
sfan5 1586ccc0c8 Update VERSION 2024-04-17 21:19:56 +02:00
sfan5 02254b92dd Release 0.38.0 2024-04-17 21:18:44 +02:00
sfan5 3b64bf908a test: add an encoding test with libmpv 2024-04-17 20:22:36 +02:00
sfan5 2af3a6e294 meson: improve option descriptions 2024-03-26 12:59:59 +01:00
sfan5 52638a4cac meson: refine dependencies of many feature checks 2024-03-26 12:59:59 +01:00
sfan5 af69b268c2 meson: remove some unneeded features from list 
There's no practical use in knowing that e.g. mpv was linked
to libm and these are not used for HAVE_* tests either.
 2024-03-26 12:59:59 +01:00
sfan5 7ccb1cbcd4 DOCS: update for new --pulse-latency-hacks defaults 
I missed this.
fixes: 8e3737ab63
 2024-03-25 22:45:58 +01:00
sfan5 8e3737ab63 ao_pulse: reenable latency hacks by default 
As far as I can tell PulseAudio introduced a bug in 16.0
where if a stream is (un)paused too often the reported latency
will momentarily spike by 3000% or more. Apparently in certain cases
just pausing once and waiting can also cause this.

Save the remaining users of PA the trouble of debugging the various
obscure issues that can arise from this (desync is a harmless example)
by enabling the latency hack code again.

ref: <https://github.com/mpv-player/mpv/issues/12057>
     <https://github.com/mpv-player/mpv/issues/10333>
 2024-03-24 09:58:41 +01:00
sfan5 1d8f28cea7 meson: conditionalize searching for ANGLE's EGL 2024-03-21 18:27:08 +01:00
sfan5 a392f91170 meson: get rid of 'egl-helpers' feature 
This was just redundant because it's always present together with 'egl'.
 2024-03-21 18:27:08 +01:00
sfan5 3679d18b54 meson: fix EGL version check 2024-03-21 18:27:08 +01:00
sfan5 6b8bd8072f video/egl_helpers: fix fallback logic for EGL_KHR_create_context 
Both possible paths had bugs:
For OpenGL it passed EGL_CONTEXT_CLIENT_VERSION, which - in older versions
of the standard - was not permitted.
For GLES it always assumed EGL_CONTEXT_FLAGS_KHR to work, which belongs to the
aforementioned extension.

Ironically this was never a problem (probably saved by implementations
not being overly strict) except in 2024 on an emulated Android 14 device
that trips over this edge case. It is a mystery.
 2024-03-21 18:27:08 +01:00
sfan5 b3ca600acb video/egl_helpers: log error for eglCreateContext 2024-03-21 18:27:08 +01:00
sfan5 ead9f892b3 various: use static assertions where appropriate 2024-03-17 20:04:04 +01:00
sfan5 830f6ccd6b vo_{drm,wlshm}: make query_format checks more correct 
We're using mp_sws here, so we should ask it for format support
and not the underlying library (usually swscale) directly.
 2024-03-16 13:27:34 +01:00
sfan5 aa75a0e9d2 vo_drm: add support for YUYV format 
As the first aligned format this required a fix to reconfig().

Adding the other component-swapped formats in this group would be trivial
but I checked the DRM database [1] and no driver exists that supports
one of those but not YUYV and this is quite fringe as-is, so I opted not to.

[1] <https://drmdb.emersion.fr/formats>
 2024-03-16 13:27:34 +01:00
sfan5 bc8038cffd vo_{drm,wlshm,x11}: add support for video-target-params 2024-03-16 13:27:34 +01:00
sfan5 2893b7d0f5 vo: move target_params into responsibility of VO 
The VO generic code tries to be helpful and resets this after
each reconfig. However for the simpler VOs the target params
are constant after a reconfig or even for the entire lifetime.
So it's clearly better to let the VO decide.

This also allows the VO to use a static buffer instead.
 2024-03-16 13:27:34 +01:00
sfan5 44f54357ca context_drm_egl: log fallback correctly 
Currently a theoretical concern because we handle all existing formats.
 2024-03-16 13:27:34 +01:00
sfan5 6b452788c4 vo_drm: add support for BGR formats 2024-03-16 13:27:34 +01:00
sfan5 93a5059e8c vo_drm: use native matching pixel format for XRGB2101010 2024-03-16 13:27:34 +01:00
sfan5 dafced8a8a wayland_common: fix type of dnd_action 
The default value of -1 and comparisons >= 0 only work properly
with an int type.
 2024-02-29 17:24:05 +01:00
sfan5 878b76f75e wayland_common: fix initialization order issue with protocols 
This broke DND under (apparently) GNOME and KWin, but not sway.

fixes: 2274311b25
 2024-02-29 17:24:05 +01:00
sfan5 78cedac844 wayland_common: fix DND read error handling 2024-02-29 17:24:05 +01:00
sfan5 dec29e82ac wayland_common: read DND data in larger chunks 2024-02-29 17:24:05 +01:00
sfan5 98f7f9e25e wayland_common: log if DND fails 
This can happen if the compositor or applications don't behave correctly,
so let the user know to aid debugging.
 2024-02-29 17:24:05 +01:00
sfan5 f36ab2c609 wayland_common: free DND resources on shutdown and error 
This leaked only in edge cases, if at all.
 2024-02-29 17:24:05 +01:00
sfan5 d955dfab29 misc/jni: reduce duplication in mapping struct 
'name' was in fact unused when reading fields or methods, so it can be merged with 'method'.
Also changed the type of 'mandatory' to bool.
 2024-02-28 16:11:54 +01:00
sfan5 5b1eaf3ff1 misc/jni: introduce macros for deleting references 2024-02-28 16:11:54 +01:00
sfan5 75ae44472b misc/jni: general code cleanup and refactor 
Make it align with mpv's coding conventions in general and avoid unecessary Lavu use.
 2024-02-28 16:11:54 +01:00