- Jul 23, 2024
-
-
Steve Lhomme authored
There should be no difference: - the vendor system is meant to pick the source needed to rebuild the package for all targets, it doesn't require a target, in fact there is no "--target" option. - someone even made a tool to do vendoring for a specific target: https://users.rust-lang.org/t/is-it-possible-to-vendor-dependencies-only-for-a-given-target/90261/2 - before and after the patch the rav1e-vendor-build folder is the same - using CARGO_NATIVE on Linux I still get a ton of windows packages in the rav1e-vendor folder It may start a little faster while the cross compiler is being downloaded.
-
Steve Lhomme authored
PKGS_TOOLS are native tools that are used by VLC so they are rebuilt when using prebuilt contribs. cargo capi is only used to build rav1e, it's not needed when using prebuilt contribs. This may be reverted if we ever want to use cargo capi in VLC.
-
Steve Lhomme authored
cargo-capi is a native tool.
-
Steve Lhomme authored
The C compiler and native tools should be the BUILDxxx ones. The RUSTFLAGS are not needed, especially the target platform specific ones.
-
Steve Lhomme authored
-
Steve Lhomme authored
It may be configured to default to another target. We need to check the RUST_HOST is supported because rustc returns a version even if the --target doesn't match a known target. Maybe in the future it will.
-
Steve Lhomme authored
If for some reason we don't have a proper value, we don't have a usable value, we cannot compile native rust code nor install the cross-compiler.
-
Steve Lhomme authored
-
Co-authored-by:Steve Lhomme <robux4@ycbcr.xyz>
-
Steve Lhomme authored
-
Steve Lhomme authored
-
- Jul 22, 2024
-
-
Only call UpdateListener() if the clock is valid. This fixes the following assert: vlc: ../../src/clock/clock.c:585: vlc_clock_monotonic_to_system: Assertion `ctx->wait_sync_ref.stream != VLC_TICK_INVALID' failed. This was occurring after a pause, a flush and a resume (that triggers a rate change). The assertion is new, but VLC_TICK_INVALID was used as a reference point before in that case.
-
No functional changes, this is needed by the next commit as UpdateListener() use cl->rate (and need to be up to date).
-
This already works fine in other places, I must have missed updating this in the last iteration. Brief explanation of how this works: - Tap handler should set the focus, even if the drag handler takes over. This means that the focus should be set in both `canceled()` and `singleTapped()` handlers. This merely corresponds to `pressed()` signal of mouse area. - Tap handler should not update the selection if the drag handler takes over. This means that the selection should not be updated upon `canceled()` but `singleTapped()` - Drag handler must update the selection when it becomes active, as long as the dragged item is not already selected.
-
The calculation is not necessary, when `when` is not set. So, there is no need to do the calculation. At the same time, the calculation must be done _after_ `when` is set. Currently, there is no guarantee for that and it causes issues with expanding items getting invisible unexpectedly.
-
-
-
- Jul 21, 2024
-
-
`tapped()` is a signal that is emitted on release, not press. Therefore, currently the state is not set properly.
-
It appears that setting the gesture policy is enough, and setting grab permission to `CanTakeOverFromAnything` seems to be conflicting with the drag handler.
-
Widget based user interface is no longer used, `QMainWindow` (which is a `QWidget`) is no longer useful.
-
-
-
Signed-off-by:Claudio Cambra <developer@claudiocambra.com>
-
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by: -
Signed-off-by:
-
