Rendering by line into a line buffer that's in IRAM can be faster than accessing framebuffers in slower PSRAM, so offer this by allowing users
to omit even the initial framebuffer.
They were causing infinite recursion because they were calling
themselves.
Also add the missing MapModel::reset
Yet another motivation for https://github.com/slint-ui/slint/issues/3888
as the code was mixing the `reset` function on the adapter meaning
"please re-apply the adapter filter/map/sort function" with the `reset`
function on Model which means "the subclass has changed and we should
notify listeners".
Fixes#4968
A MSRV update is required to update the image crate to 0.25, otherwise
we get link error that are ficed in newer version.
Also other dependency update such as env_logger needs a newer MSRV.
Update to 1.73 because it has functions like `with_borrow` on thread
storage which we already actually use in some platform.
This is also the last release before the the drop of macOs < 10.12
in Rust 1.74
The energy monitor declares a transition on an animated int property,
for which Property<int>::set_animated_binding_for_transition is called,
which in turn calls slint_property_set_animated_binding_helper. The
latter is overloaded for various property types, such as float, Color,
or Brush, and then calls specialized functions from ffi, such as
slint_property_set_animated_binding_(int|float|etc.).
slint_property_set_animated_binding_int uses i32 in Rust, which cbindgen
maps to int32_t, so the
slint_property_set_animated_binding_helper overload also uses int32_t.
Unfortunately, with esp-idf, int32_t is a distinct type from int, and
the overload resolution fails.
As remedy, this change uses c_int instead of i32 in the Rust ffi, which
maps to int.
This seems easier than changing Property<int> to Property<int32_t> :-)
- Bump esp-backtrace for esp32-p4 support, which removes the uart feature and delegates to esp-println
- Added esp-println dependency
- Add mapping for ESP32P4 and other risc-v esp-idf targets to the corresponding rust target triplet.
Unfortunately there's no generic pattern like for xtensa, as can be seen on
https://github.com/esp-rs/esp-idf-sys?tab=readme-ov-file#examples
It looks like the esp_lcd_rgb_panel.h header file is not unconditionally
available anymore. Guard it with the same #ifdef that's surrounding the
calls to esp_lcd_rgb_panel_* and include the header that provides it.
The trick is that the backend selector build by default with the
i-slint-backend-qt, but the "enable" feature is only enabled if the
qt-backend feature is enabled explicitly, or on linux from the slint
or slint-interpreter crate
They are not commonly used so no need to use semi-public api for that
and it's really easy to similate with ust two clicks anyway
The previous test function were added before there was public way to
send events
This was requested by a customer recently and it seems rather straight-forward to implement and offer. `clear()` mirrors `std::vector::clear()` and `set_vector` mirrors the `set_vec` we have in Rust.
Move the set_fullscreen function added to the WindowAdapter trait in 779aff0b39
to be a function in WindowProperties instead.
That way it'll be easier in the future to extend this with other window states without
having to modify or break the WindowAdapter trait API.
Move this back out of render() again and make it stateful in the renderer. Reduces the amount of book-keeping required and it's always the same callback anyway.
