In `0.22.0`, `eframe::start_web` has been replaced with `eframe::WebRunner`, which also installs a nice panic hook (no need for `console_error_panic_hook`).
A window may not always be available and may have already been closed by the time an eframe app is closing. An example of this is Android, where the main activity window may have been stopped or discarded because the app is no longer in the foreground, and then the user decides to close your app without resuming it using the multitasking view.
In this case, skip the window persistence step if it does not exist anymore by the time we are saving the persistence data. Currently eframe will panic with `winit window doesn't exist` instead.
* Triage for GL backend
* And cargo-fmt
* Changelog update with PR and issue
* Update crates/eframe/src/epi/mod.rs
Co-authored-by: Andreas Reich <r_andreas2@web.de>
* Update crates/egui-wgpu/src/winit.rs
Co-authored-by: Andreas Reich <r_andreas2@web.de>
* Add "supports_screenshot" to surface state
* Cranky fix
* fmt
---------
Co-authored-by: Andreas Reich <r_andreas2@web.de>
* Expose raw window and display handles in eframe
* Ensure that no one implements `Clone` in the future
* Cleanup
---------
Co-authored-by: Matti Virkkunen <mvirkkunen@gmail.com>
The approach of #1889 may remove observers in a view
twice, which produces the Obj-C Exception:
Cannot remove an observer <...> for the key path
"nextResponder" from <WinitView ...> because
it is not registered as an observer.
The above message can only be seen when attaching the
application to debugger. Users normally see:
[1] *** trace trap cargo run
This commit fixes it by only running `event_loop.run_return()`
twice on Windows. Besides:
* We have set `ControlFlow::Exit` on `Event::LoopDestroyed`,
`EventResult::Exit` and on error; therefore, it is safe
to not calling `set_exit()`.
* This commit also fix the persistence function in macOS.
It can't store the content in Memory due to this exception.
Fixed: #2768 (eframe: "App quit unexpectedly" on macOS)
Signed-off-by: pan93412 <pan93412@gmail.com>
* Add an optional app_id field to eframe's NativeOptions (#1600).
This is used in the window builder to set the application ID, which is e.g. important for a proper configuration in `.desktop` files under Wayland.
When no application ID is explicitly set, it defaults to the title of the window.
* Only enable NativeOptions::app_id under Linux.
The wayland feature is not sufficent as constraint and it won't compile e.g. under Windows.
While Wayland could also be used on other Unix-Systems like FreeBSD, this would probably need some specific testing. Winit uses the following definition as "wayland_platform" and on which the required packages are available:
> wayland_platform: { all(feature = "wayland", free_unix, not(wasm), not(redox)) },
* Do not use title as default application ID under Wayland.
The title might be used to also communicate state (opened file, ...) to the user and this might have unforeseen consequences for the application ID. It seems to be better to use the old behavior of not setting an application ID in this case. Also add an example on how to set the application ID in the documentation.
* Avoid as_deref(), which was a left-over of a previous version
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
* Detect panics during initialization and show them to the user
* PanicHandler now also logs the panics
* Add example of how to call into your app from JS
* Refactor: break out AppRunner and AppRunnerRef to own files
* Hide AppRunner
* Simplify user code
* AppRunnerRef -> WebRunner
* Better docs
* Don't paint until first animation frame
* Update multiple_apps.html
* Update web demo
* Cleanup and fixes
* left-align panic message in html
* Add Pointer::is_decidedly_dragging and could_any_button_be_click
This allows users to distinguish between click and drags
while they are not yet done.
* Fix warning in eframe
* fix typo
* Log warning instead of error when failing to decode RON in storage
* New web demo
* Clean up some warn/error logging
* Avoid deadlock that could happen on crash
* Log errors using console.warn, because console.error can cause crashes
* Use patched version of wasm-bindgen-cli, allowing >2GB memory
* New web demo
* Make wgpu webgl/gles opt-in (but still work out of the box via feature flag), workaround canvas creation issue
* missing allow unsafe code annotations
* add breaking change not to eframe release notes
* Add --webgpu flag to build_demo_web.sh
* Improve CHANGELOG docs
* Clean up, and prepare for having to wasm:s
* put canvas without workaround under `if false`
* fix spelling
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
* Refactor: remove extra store of events
* Remove unnecessary extra function
* Refactor: simplify event registering
* Store panic summary
* egui_demo_app: move web-part to own module
* index.html: await
* Properly unsubscribe from events on panic
* Better error handling
* Demo app html: hide the wasm canvas and show an error message on panic
* egui_demo_app: add panic button to test panic response on web
* fix typo
* Use a constructor to create WebHandle
* Refactor: less use of locks in the interfaces
* More consistent naming
* eframe: Set app icon on Mac and Windows
Also: correctly set window title on Mac when launching from
another process, e.g. python.
Co-authored-by: Wumpf <andreas@rerun.io>
* lint fixes
* Fix web build
* fix typo
* Try fix windows build
---------
Co-authored-by: Wumpf <andreas@rerun.io>
* Refactor repaint logic
* request_repaint_after also fires the request_repaint callback
* Bug fixes
* Add test to egui_demo_app
* build_demo_web: build debug unless --release is specified
* Fix the web backend too
* Run special clippy for wasm, forbidding some types/methods
* Remove wasm_bindgen_check.sh
* Fix typos
* Revert "Remove wasm_bindgen_check.sh"
This reverts commit 92dde253446a6930f34f2fcf67f76bc11669ec3b.
* Only run cranky/clippy once
* add method for requesting attention to the main window
* use another enum member for user attention type instead of nested `Option`s
(also, document the enum members now that they don't mirror `winit`)
* update the docstring
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
* add an example app for testing window attention requests
* Apply suggestions from code review
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
* remove `chrono` dependency and improve the attention example's readability
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
It is very easy for keys to become stuck when we alt-tab,
or a save-dialog opens by Ctrl+S, etc.
Therefore we new clear all the modifiers and down keys to avoid that.
* Remove dark-light dependency
Since https://github.com/emilk/egui/pull/2750 we now get what we need
straight from `winit`.
* fix warning
* Docstring formatting
* fix typo in check.sh
* Replace tracing crate with log
It's just so much simpler to use
* Add `bacon wasm` job
* eframe: add a WebLogger for piping log events to the web console
* Add raw mouse wheel event
The event is sent as it comes from the backend, so it will follow
different conventions depending on the target, and it is up to the user
code to deal with that. The goal is to allow advanced users to implement
alternative UI controls, e.g., using Ctrl to scroll the plot
horizontally instead of zooming, or use Shift to scroll faster instead
of changing direction.
* Change Pixel to Point for consistency
Apply suggestions from code review by emilk
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
* Inline mouse wheel raw event closure
It was meant only to be able to use the same variable names without
shadowing the rest of the code, but a simple block accomplishes the same
thing.
* Use wildcard on wheel event match
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
* Flip mouse wheel delta sign on web to match native
* Use wheel event data to generate scroll event
To avoid doing the same match and sign conversion twice.
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
Instead of depending on android-activity directly, this exposes the
android-native-activity and android-game-activity features from Winit.
This ensures that applications can choose what android-backend they use
while also relying on Winit to decide what version of android-activity to
use - without increasing the risk of a version conflict by having a direct
dependency.
_(NB: Egui doesn't currently use the android-activity API itself)_
Since android-activity provides the `android_main()` entry point for
Android applications it's not possible to link in multiple version of
the android-activity crate and so it's particularly important to
avoid unnecessary direct dependencies that could cause a version
conflict in the future.
To help avoid the need for applications to directly depend on
android-activity the Winit crate re-exports the android-activity API
and exposes features to configure the backend so that application crates
can instead rely on Winit to pull in a compatible version of
android-activity. (This way version bumps for android-activity only
need to be synchronized with the Winit crate).
CI now enables the `android-native-activity` feature for testing.
Fixes: #2829Fixes: #2720Closes: #2834
`EventResult::Exit` is hit before `Event::LoopDestroyed` is, and due to
possibly some order of operations or drops the window is never destroyed
on Linux. Adding a call to `winit_app.save_and_destroy();` where
`EventResult::Exit` is checked solves this.
Closes#2892
Signed-off-by: Luke D. Jones <luke@ljones.dev>
* Add ability to read the native window's focus state
* Add `eframe::Frame::focus()` for requesting the native window's focus
* rename the output field `active` → `focused` for consistency
Emil Ernerfeldt
jacekpoz
Ho Kim
Stephen M. Coakley
Matt Fellenz
icedrocket
amfaber
bilabila
τ
pan93412
Thomas Krause
Andreas Reich
Niklas Korz
TicClick
Dixeran
Jozef Číž
YgorSouza
Robert Bragg
Luke Jones