* Closes https://github.com/emilk/egui/issues/3501
The problem occurs when you want to render the same SVG at different
scales, either at the same time in different parts of your UI, or at two
different times (e.g. the DPI changes).
The solution is to use the `SizeHint` as part of the key.
However, when you have an SVG in a resizable container, that can lead to
hundreds of versions of the same SVG. So new eviction code is added to
handle this case.
Parley's bumped accesskit to 0.19, so we have to as well. It's a bit
concerning that we may end up locked to the version of accesskit that
Parley uses, but oh well.
[This kittest PR will have to be merged
first.](https://github.com/rerun-io/kittest/pull/11)
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
TextShape.visual_bounding_rect was not taking the text rotation into
account. I manually tested drawing the new bounding box on top of the
text for various rotations & anchor settings. For example:
<img width="191" alt="image"
src="https://github.com/user-attachments/assets/56528fc7-7e7d-45af-b92a-c1cd307ff205"
/>
The unit test I added will fail without this patch, but perhaps doesn't
add much value.
* [x] I have followed the instructions in the PR template
This PR corrects the deprecation note on ComboBox::from_id_source:
Changed:
`#[deprecated = "Renamed id_salt"]`
To:
`#[deprecated = "Renamed from_id_salt"]`
Fixes a regression introduced in https://github.com/emilk/egui/pull/5411
(possibly
d74bee536f)
that breaks `leading_space` handling.
I think this is what the condition should be but I haven't touched this
code in a while.
With kittest it was difficult to wait for images to be loaded before
taking a snapshot test.
This PR adds `Harness::with_wait_for_pending_images` (true by default)
which will cause `Harness::run` to sleep until all images are loaded (or
`HarnessBuilder::with_max_steps` is exceeded).
It also adds a new ImageLoader::has_pending and
BytesLoader::has_pending, which should be implemented if things are
loaded / decoded asynchronously.
It reverts https://github.com/emilk/egui/pull/6901 which was my previous
attempt to fix this (but this didn't work since only the tested crate is
compiled with cfg(test) and not it's dependencies)
<!--
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!
-->
Did not test on platforms other than my phone, but I can't imagine it
causing problems. AFAIK if highp isn't supported then `precision highp
float;` needs to still not cause the program to fail to link/compile or
anything; it should just silently use some other precision.
* Fixes https://github.com/emilk/egui/issues/4268 for me but I only
tested it on a native Android app and I don't know whether backends
other than glow are affected.
* [x] I have followed the instructions in the PR template (but the
change is trivial so I'm just doing it from the master branch)
Before:
After:
## What
Fail all PRs that are opened from the master/main branch of the fork.
## Why
PR:s opened from the `master` branch cannot be collaborated on. That is,
we maintainers cannot push our own commits to it (e.g. to fix smaller
problems with it before merging).
## How
Untested code straight from Claude 3.7 😅
https://github.com/emilk/egui/pull/5394 made it so images would load on
a background thread, which is great. But this makes snapshot tests that
have images via include_image!() flakey since they might or might not
load by the time the snapshot is rendered.
This is no perfect solution, since the underlying problem of "waiting
for something async to happen" still exists and we should add some more
general solution for that.
<!--
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!
-->
* Closes#2875
* Closes https://github.com/emilk/egui/pull/3340
* [x] I have followed the instructions in the PR template
Adds `create_native`. Similiar to `run_native` but it returns an
`EframeWinitApplication` which is a `winit::ApplicationHandler`. This
can be run on your own event loop. A helper fn `pump_eframe_app` is
provided to pump the event loop and get the control flow state back.
I have been using this approach for a few months.
---------
Co-authored-by: Will Brown <opensource@rebeagle.com>
This helped me benchmark the atomic layout (#5830) changes.
I also realized that the label benchmark wasn't testing the painting,
since the buttons at some point will be placed outside the screen_rect,
meaning it won't be painted.
This fixes it by benching the label in a child ui.
The `label &str` benchmark went from 483 ns to 535 ns with these
changes.
EDIT:
I fixed another benchmark problem, since the benchmark would show the
same widget millions of times for a single frame, the WidgetRects
hashmap would get huge, causing each iteration to slow down a bit more
and causing the benchmark to have unreliable results.
With this the `label &str` benchmark went from 535ns to 298ns. Also the
`label format!` benchmark now takes almost the same time (302 ns).
Before, it was a lot slower since it reused the same Context which
already had millions of widget ids.
* Closes <https://github.com/emilk/egui/issues/5406>
* [x] I have followed the instructions in the PR template
The changes follow what is described in the issue with a couple changes:
- Scroll bars are not hidden when dragging is disabled, for that
`ScrollArea::scroll_bar_visibility()` has to be used, this is as not to
limit the user configurability by imposing a specific function. The user
might want to retain the scrollbars visibility to show the current
position.
- The input for mouse wheel scrolling is unchanged. When I inspected the
code initially I made a mistake in recognizing the source of scrolling.
Current implementation is in fact using
`InputState::smooth_scroll_delta` and not `PassState::scroll_delta`,
therefore it is possible to prevent scrolling by setting the
`InputState::smooth_scroll_delta` to zero before painting the
`ScrollArea`.
A simple demo is available at
https://github.com/MStarha/egui_scroll_area_test
* [x] I have followed the instructions in the PR template
Previously, navigating text in `TextEdit` with Ctrl + left/right arrow
would jump inside words that contained combining characters (i.e.
diacritics). This PR introduces new dependency of `unicode-segmentation`
to handle grapheme encoding. The new implementation ignores whitespace
and other separators such as `-` (dash) between words, but respects `_`
(underscore).
---------
Co-authored-by: lucasmerlin <hi@lucasmerlin.me>
Fix: Popups do not appear in certain situations.
* Closes#5080
* Related #5107
The root cause is that `last_move_time` is not updated in certain
situations (slow situations?).
This adds an option for specifying the set of pointer buttons that can
be used to pan the scene via clicking and dragging.
The original behaviour where all buttons can pan the scene by default is
maintained.
Addresses part of #5891.
---
Edit: It looks like the failing test is unrelated and also appears on
master:
https://github.com/emilk/egui/actions/runs/14330259861/job/40164414607.
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
Starting with 77244cd4c5 the popup
open-state is cleaned up per memory pass.
This becomes problematic for implementations that share memory between
viewports (i.e. all of them, as far as i understand it), because each
viewport gets a context pass, and thus a memory pass, which cleans out
popup open state.
To illustrate my issue, i have modifed the multiple viewport example to
include a popup menu for the labels:
https://gist.github.com/mkalte666/4ecd6b658003df4c6d532ae2060c7595
(changes not included in this pr).
Then, when i try to use the popups, i get this:
https://github.com/user-attachments/assets/7d04b872-5396-4823-bf30-824122925528
Immediate viewports just break popup handling in general, while deferred
viewports kinda work, or dont. In this example ill be honest, it kind of
still did, sometimes. In my more complex app with multiple viewports
(where i encountered this first) it also just broke - even when just
showing root and one nother. Probably to do with the order wgpu vs glow
draws the viewports? Im not sure. In any case:
This commit adds `Memory::popup` (now `Memory::popups`) to the
per-viewport state of memory, including viewport aware cleanup, and
adding it to the list of things cleaned if a viewport goes away.
This results in the expected behavior:
https://github.com/user-attachments/assets/fd1f74b7-d3b2-4edc-8dc4-2ad3cfa2160e
I will note that with this, changing focus does not cause a popup to be
closed, which is consistent with current behavior on a single app.
Hope this helps
~Malte
<!--
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!
-->
* Closes <https://github.com/emilk/egui/issues/THE_RELEVANT_ISSUE>
* [x] I have followed the instructions in the PR template
* [x] ~~I have run check.sh locally~~ CI on the fork, including checks,
went through.
**Added**
* Create `svg_text` feature flag to support text rendering & loading of
system fonts.
**Changed**
* Updates `resvg` to `0.45`.
* Adds `usvg::Options` field to the `SvgLoader` structure.
* Change function signatures to support passing `usvg::Options` to
downstream `load_svg_bytes_with_size`.
**Additional Info**
* I used this PR as a reference:
https://github.com/emilk/egui/pull/4659. @xNWP can you see if this
adequately resolves your concern from your original PR?
* Closes https://github.com/emilk/egui/issues/5977 (we may want to open
another issue for my other thoughts in this issue)
* Also, I would like to thank @xNWP and their original PR for being a
good reference for this one.
* [x] I have followed the instructions in the PR template
For some reason the pipeline in https://github.com/emilk/egui/pull/5698
succeeded even though the snapshots should have been updated. This
updates the snapshots.
Current implementation of ColorTest infinitely expand horizontally at
each redraw if included in a Window.
The effect can be see replacing the Panel in the ColorTestApp::update
with a Window:
```
egui::CentralPanel::default().show(ctx, |ui| {
egui::Window::new("Colors").vscroll(true).show(ctx, |ui| {
if frame.is_web() {
ui.label(
"NOTE: Some old browsers stuck on WebGL1 without sRGB support will not pass the color test.",
);
ui.separator();
}
egui::ScrollArea::both().auto_shrink(false).show(ui, |ui| {
self.color_test.ui(ui);
});
self.color_test.ui(ui);
});
```
The cause is the is the _pixel_test_strokes_ function that, at each
redraw, tries to expand the target rectangle of 2.0 points in each
direction.
* Closes no issue, I just needed this for an app and figured it could be
useful.
* [x] I have followed the instructions in the PR template
This PR adds an `overline` option for `egui_extras::TableRow`, which is
useful for visually grouping rows. The overline consumes no layout
space.
A screenshot of the demo app, showing every 7th row getting an overline.
<img width="704" alt="Screenshot 2025-01-25 at 14 40 08"
src="https://github.com/user-attachments/assets/9ccbee3d-296d-4afd-9290-c669e4ede1c0"
/>
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
Emil Ernerfeldt
valadaptive
Patrick Marks
Piotr Podusowski
Abdellatif EL MIZEB
TÖRÖK Attila
Hubert Głuchowski
Lucas Meurer
Alexander Nadeau
Will Brown
mitchmindtree
Gaelan McMillan
MStarha
Bennet Bleßmann
rustbasic
Sven Niederberger
mkalte
Marek Bernat
Christopher Cerne
giuseppeferrari
Aarni Koskela