- Jan 31, 2025
-
-
-
If the p_extra is freed in the previous branch, it is also freed in this branch without being reset to NULL before. Regression from a95213e2.
-
Whenever an ES selection is forced, the algorithm taking care of default selecting the tracks at ES creation should take it into account. Otherwise, tracks with highest priority will always be selected when created after an user selected one. This happens a lot in DVDs since PS sets track priority. Fixes #24815
-
-
GL_EXT_YUV_target[^1] provides a way to access an external texture without applying a conversion to RGBA texture, which can be used to supply our own matrices for conversion. This can be used to workaround bogus RGBA conversion for some devices and can help integrating better with deinterlacing, or in general plane filtering, and some YUV pipeline like encoding. [^1]: https://registry.khronos.org/OpenGL/extensions/EXT/EXT_YUV_target.txt
-
-
When a filter could not load because of a sampler creation failure, there were no messages indicating that it came from the sampler, nor information on what failed in the sampler.
-
-
-
A new YUV list is created to avoid adding the new chroma to YUV fallbacks.
-
Steve Lhomme authored
It's different between C and C++, we should use the proper type no matter what. See https://learn.microsoft.com/en-us/windows/win32/api/dxgidebug/nf-dxgidebug-dxgigetdebuginterface
-
Steve Lhomme authored
This reverts commit fbbd68f8. There are in fact the "discard" and "mean" deinterlace filters that use half height.
-
Steve Lhomme authored
-
- Jan 30, 2025
-
-
This is used in other places in the script and if unset leads to the `-arch` argument being passed without any architecture, which is invalid and causes compiler invocations to fail. The reason this worked in CI is that the build script overwrites the ACTUAL_HOST_ARCH as it can be changed by an argument. However the env.build.sh script is used also outside the build.sh script in manual build scenarios, which is the whole reason it is a separate script to begin with.
-
Alexandre Janniaux authored
The tracer is designed to send the metrics towards a telegraf server using this kind of configuration: [[inputs.socket_listener]] service_address = "tcp://localhost:8094" The telegraf server can then forward the metrics towards an influxdb server for monitoring or directly to grafana live server[^1] for introspection. The influxdb database can also be used to query the metrics after they've been indexed. The application using the tracer can use the VLC_TELEGRAF_ENDPOINT environment variable (eg. VLC_TELEGRAF_ENDPOINT=tcp://127.0.0.1:8094) to set where the tracer will output the traces to. A bunch of notes and improvement left for later: - Unsafe code is still used for accessing the fields data since an union is used and union access is unsafe. It could probably be wrapped from the binding implementation. - There is no way to specify the address using the configuration since vlc_variable is not bound to the Rust bindings and no unsafe extern "C" code...
-
Alexandre Janniaux authored
This commit allows rust code to use the tracer API, through the trace!() macro, and also exposes a TracerCapability and TracerModuleLoader to create new tracer modules in Rust.
-
As configure.ac is patched
-
Steve Lhomme authored
-
Steve Lhomme authored
-
Steve Lhomme authored
-
- Jan 29, 2025
-
-
-
Emscripten is known to deal badly with function casts to incompatible types [^1]. This patch prevent us to accidentally merge harmful UB casts to wasm enabled code. The warning is really strict but is compliant to the standard. Code that wants to use function cast should be disabled from the wasm platform. [^1]: https://emscripten.org/docs/porting/guidelines/function_pointer_issues.html
-
The function signature does not match the libplacebo callback (gl_t* instead of void*). While this is supported by many platforms and compilers, it's undefined behavior strictly by the standard [1] and it's known to often fail with emscripten [^2]. [1]: Section 6.3.2.3, paragraph 8: > A pointer to a function of one type may be converted to a pointer to a > function of another type and back again; the result shall compare > equal to the original pointer. If a converted pointer is used to call > a function whose type is not compatible with the pointed-to type, the > behavior is undefined. Quoting section 6.7.5.1, paragraph 2 for pointer compatibility: > For two pointer types to be compatible, both shall be identically > qualified and both shall be pointers to compatible types. [^2]: https://emscripten.org/docs/porting/guidelines/function_pointer_issues.html
-
The function signature does not match the dvbpsi callback (demux_t* instead of void*). While this is supported by many platforms and compilers, it's undefined behavior strictly by the standard [1] and it's known to often fail with emscripten [^2]. [1]: Section 6.3.2.3, paragraph 8: > A pointer to a function of one type may be converted to a pointer to a > function of another type and back again; the result shall compare > equal to the original pointer. If a converted pointer is used to call > a function whose type is not compatible with the pointed-to type, the > behavior is undefined. Quoting section 6.7.5.1, paragraph 2 for pointer compatibility: > For two pointer types to be compatible, both shall be identically > qualified and both shall be pointers to compatible types. [^2]: https://emscripten.org/docs/porting/guidelines/function_pointer_issues.html
-
Steve Lhomme authored
Especially when mixing LPCWSTR and LPWSTR*.
-
Given contribs are bootstrapped with `--disable-optimizations` option, the CMAKE flag `-DCMAKE_BUILD_TYPE` is set to `Debug`, and when build configuration is set to `Debug` in libpng contrib, it CMAKE configuration append a `d` char to the static library name, resulting in a lib named `libpng16d.a` instead of `libpng16.a`. Then some vlc module depending on libpng aren't able to link properly as they can't find their lpng16 dependency.
-
- Jan 28, 2025
-
-
Fixes a regression in the SAP descriptor generation.
-
we can't rely on index to refer to an item across views as the ordering of the model is not stable across the views. This MR uses the artist MLId to set which subview should be loaded fix: #28556
-
this allows to request more data from the cache using the standart growth policy rather than requiring the caller to `refer` and item after what is currently loaded. This function ensure that the cache tries to load at least the current count + chunk size
-
-
This allows getting the position of an item in the model from its MLItemID
-
this allows to retrieve the content of and item given its id
-
-
-
-
Steve Lhomme authored
-
Steve Lhomme authored
It avoids the warning in the core that uses whatever default it expects. A SPU with absolute coordinates should always tell what dimensions it's referring to. For DVD SPU we know it has to correspond to the visible video area.
-
Steve Lhomme authored
-
Steve Lhomme authored
Even in 3.0 it's never set to true. Fixes #28952
-