Pierre Lamot authored 11 months ago and Steve Lhomme committed 11 months ago

IconToolButton relies on the font size since 3df6a79a

Pierre Lamot authored 11 months ago and Steve Lhomme committed 11 months ago

description field should be used since d6b6ac32

Pierre Lamot authored 11 months ago and Steve Lhomme committed 11 months ago

IconLabel already exists in Qt namespace

Opus 1.5.2 fixes several build issues that were discovered since the 1.5 release.
It also fixes a misalignment issue in the AVX2 code that could cause crashes under Windows.

Jiepeng Huang authored 11 months ago and Steve Lhomme committed 11 months ago

The `psz_lcname` is lowercase, so we need to use `strcasecmp`
instead.

Jiepeng Huang authored 11 months ago and Steve Lhomme committed 11 months ago

In macOS/iOS/tvOS, font collection file contains multiple fonts.
The 0 index should not be used by default, but the corresponding
index should be used. This will fix some font rendering bugs and
fallback font bugs.

Just like the other enums.

Alexandre Janniaux authored 11 months ago and Steve Lhomme committed 11 months ago

The upper class was used everywhere, leading to the pointer being
downcasted at the beginning of every function. By using the downcast
class everywhere, we don't introduce explicit downcast in most places.

This commit also reduces the usage of the es_out_t type as parameter,
which will be moved to another downclass in follow-up MR.

Alexandre Janniaux authored 11 months ago and Steve Lhomme committed 11 months ago

The upper class was used everywhere, leading to the pointer being
downcasted at the beginning of every function. By using the downcast
class everywhere, we don't introduce explicit downcast in most places.

This commit also reduces the usage of the es_out_t type as parameter,
which will be moved to another downclass in follow-up MR.

Alexandre Janniaux authored 11 months ago and Steve Lhomme committed 11 months ago

es_out_sys_t is the name used in es_out.c for the main es_out. By
providing a dedicated name, we gain more clarity and better integration
with tooling.

Alexandre Janniaux authored 11 months ago and Steve Lhomme committed 11 months ago

es_out_sys_t is the name used in es_out.c for the main es_out. By
providing a dedicated name, we gain more clarity and better integration
with tooling.

Alexandre Janniaux authored 11 months ago and Steve Lhomme committed 11 months ago

The includes are missing when including this file into another.

It's not supported [1]. NDK 26 will emit an error if you try to use it.

Upstream patch: x264!150

[1] https://android.googlesource.com/platform/bionic/+/main/docs/32-bit-abi.md

It's not supported [1]. NDK 26 will emit an error if you try to use it.

Upstream patch: https://github.com/pupnp/pupnp/pull/441

[1] https://android.googlesource.com/platform/bionic/+/main/docs/32-bit-abi.md

It's not supported [1]. NDK 26 will emit an error if you try to use it.

Upstream patch: https://github.com/xiph/flac/pull/691

[1] https://android.googlesource.com/platform/bionic/+/main/docs/32-bit-abi.md

It works for Android and should be the same on other platforms.

So the the proper system detection happens:

https://github.com/Kitware/CMake/blob/master/Modules/Platform/Android-Determine.cmake

We have a fake fxc that can be called under wine.
In Qt6.7 it may also call dxc (Shader Model 6) which can even be a Linux binary.

We need to add it to the PATH when calling qsb which assumes it's in the PATH.
qsb doesn't have an option to tell it where to look (in contribs). So if it's
not in the PATH it will never find it.

Then we can use the one from contribs.

Otherwise the package is not found on the system.

It doesn't a pkgconfig file so it was never found so far.

So any call to fxc will use our tool, as long as fxc is in the PATH.

We can also handle the option format in the script.

It will be pulled by Qt if needed.

It's only used by qtshadertools/qsb to optimize the Vulkan output.
We do not use that on any platform for now.

Even when using Vulkan it's optional and will work without.

François Cartegnie authored 11 months ago and Steve Lhomme committed 11 months ago

We could, but dont really want to patch audio rate from timescale,
as it could be meaningless for some codecs.

refs #28560

This is already the case for many of them.

This is already the case for many of them.