1
0
mirror of https://github.com/emilk/egui.git synced 2026-08-29 12:50:04 -04:00

Save state on suspend on Android and iOS (#5601)

<!--
Please read the "Making a PR" section of
[`CONTRIBUTING.md`](https://github.com/emilk/egui/blob/master/CONTRIBUTING.md)
before opening a Pull Request!

* Keep your PR:s small and focused.
* The PR title is what ends up in the changelog, so make it descriptive!
* If applicable, add a screenshot or gif.
* If it is a non-trivial addition, consider adding a demo for it to
`egui_demo_lib`, or a new example.
* Do NOT open PR:s from your `master` branch, as that makes it hard for
maintainers to test and add commits to your PR.
* Remember to run `cargo fmt` and `cargo clippy`.
* Open the PR as a draft until you have self-reviewed it and run
`./scripts/check.sh`.
* When you have addressed a PR comment, mark it as resolved.

Please be patient! I will review your PR, but my time is limited!
-->

This pull request fixes a subset of #5492 by saving the application
state when the `suspended` event is received on Android. This way, even
if the user exits the app and closes it manually right after changing
some state, it will be saved since `suspended` gets fired when the app
is exited. It does not fix the `on_exit` function not being fired - this
seems to be a winit bug (the `exiting` function in the winit application
handler trait is not called on exit). Once it gets fixed, it may be
possible to remove logic introduced by this PR (however, I am not sure
how it would handle the app being killed by the system when in the
background, that would have to be tested).

I've tested the logic by:
* Leaving from the app to the home screen, then killing it from the
"recent apps" menu
 * Leaving from the app to the "recent apps" menu and killing it
 * Restarting the device while the app was running

In all of these instances, the state was saved (the last one being a
pleasant surprise). It was tested on the repository mentioned in #5492
with my forked repository as the source for eframe (I unfortunately am
not able to test it in a larger project of mine due to dependence on
"3rd party" egui libraries (like egui_notify) which do not compile along
with the master branch of eframe (different versions of egui), but I
believe it should work in the same manner in all scenarios). Tests were
conducted on a Galaxy Tab S8 running Android 14, One UI 6.1.1.

CI passed on my fork.

* [x] I have followed the instructions in the PR template
This commit is contained in:
Pandicon
2025-01-27 08:14:49 +01:00
committed by GitHub
parent bc5f908b80
commit 93d2144294
4 changed files with 58 additions and 9 deletions

View File

@@ -366,6 +366,20 @@ impl WinitApp for GlowWinitApp<'_> {
.and_then(|r| r.glutin.borrow().window_from_viewport.get(&id).copied())
}
fn save(&mut self) {
log::debug!("WinitApp::save called");
if let Some(running) = self.running.as_mut() {
profiling::function_scope!();
// This is used because of the "save on suspend" logic on Android. Once the application is suspended, there is no window associated to it, which was causing panics when `.window().expect()` was used.
let window_opt = running.glutin.borrow().window_opt(ViewportId::ROOT);
running
.integration
.save(running.app.as_mut(), window_opt.as_deref());
}
}
fn save_and_destroy(&mut self) {
if let Some(mut running) = self.running.take() {
profiling::function_scope!();
@@ -413,7 +427,7 @@ impl WinitApp for GlowWinitApp<'_> {
if let Some(running) = &mut self.running {
running.glutin.borrow_mut().on_suspend()?;
}
Ok(EventResult::Wait)
Ok(EventResult::Save)
}
fn device_event(
@@ -1214,10 +1228,12 @@ impl GlutinWindowContext {
.expect("viewport doesn't exist")
}
fn window_opt(&self, viewport_id: ViewportId) -> Option<Arc<Window>> {
self.viewport(viewport_id).window.clone()
}
fn window(&self, viewport_id: ViewportId) -> Arc<Window> {
self.viewport(viewport_id)
.window
.clone()
self.window_opt(viewport_id)
.expect("winit window doesn't exist")
}