The default `Plot` formatter now picks precision intelligently based on
zoom level. The width of the Y axis are is now much smaller by default,
and expands as needed.
Also deprecates `Plot::y_axis_with`; replaced with `y_axis_min_width`.
Set `auto_shrink(false)` on `Window`s `ScrollArea`.
This ensures that the scroll bars are always the outermost.
* Closes https://github.com/emilk/egui/pull/4602
* Fixes https://github.com/rerun-io/rerun/issues/6638
* Related? https://github.com/emilk/egui/issues/4563
This improves how an eframe canvas works inside of a larger web page,
and how it works when there are multiple eframe apps in the same page.
`eframe` will set `tabindex="0"` on the canvas automatically, making it
focusable.
It will also set `outline: none` on the CSS, so the focused canvas won't
have an ugly outline.
## Breaking changes
You may wanna add this to your `index.html` to give the canvas focus on
startup:
```js
document.getElementById("the_canvas_id").focus();
```
## Test setup
```sh
./scripts/build_demo_web.sh
./scripts/start_server.sh
open http://localhost:8888/multiple_apps.html
```
Then open the "Input Event History" and "Text Edit" windows
## Tested
* Chromium
* [x] drag-and-drop of files
* Test both when a `TextEdit` is focused and when it is not:
* [x] `Event::Key`
* [x] `Event::Text`
* [x] copy-cut-paste
* [x] Wheel scroll
* [x] `Event::PointerGone`
* [x] Mouse drag
* [x] Mouse click
* [x] Mouse right-click
* [x] Defocus all eframe canvas, and then start typing text
* [x] Firefox (all of the above)
* [x] Desktop Safari (all of the above)
* [x] Mobile Safari
## Future work (pre-existing issues)
* https://github.com/emilk/egui/issues/4723
* https://github.com/emilk/egui/issues/4724
* https://github.com/emilk/egui/issues/4725
* https://github.com/emilk/egui/issues/4726
For a project I am currently attempting to use the acceskit output of
egui to preform end-to-end integration tests of my application. However,
the output is currently quite basic (buttons for example don't report
that they are disabled).
Before adding new features to the acceskit integration, I thought it
would be a good idea to add some testing to egui's accesskit output.
I have started with a simple test for button texts, to get feedback on
whether this is the correct direction to go in.
---------
Co-authored-by: Wybe Westra <w.westra@kwantcontrols.nl>
This PR adds `PopupCloseBehavior` to improve state of the
<https://github.com/emilk/egui/issues/4607>
`PopupCloseBehavior` determines when popup will be closed.
- `CloseOnClick` popup will be closed if the click happens anywhere even
in the popup's body
- `CloseOnClickAway` popup will be closed if the click happens somewhere
else but in the popup's body.
It also adds a test in the demo app which contains several popups
examples.
---
My ideas about <https://github.com/emilk/egui/issues/4607> is to make
every tooltip and popup a menu. So it will provide more control over
popups and tooltips (you will be able to close a popup by calling
something similar to the `ui.close_menu` if you need to). You won't need
to manually handle it's opening. And also will allow to have multiple
popups opened. That means you can have a popup inside a popup. And it
will also lead to the easier creation of the popups. (should we create a
tracking issue to track changes because to me it seems like a huge
amount of changes to be done?)
---
- Improvements on <https://github.com/emilk/egui/issues/4607>
Since Escape takes focus away from the text field, it makes more sense
for it to cancel the change rather than confirm it.
This only matters if `update_while_editing` is set to false, of course.
The menu buttons, combo box menus, and tooltips don't take layer
transforms into account when placing their popups, resulting in popups
being placed in the wrong location.
This PR makes the popups take layer transforms into account,
transforming the positions before displaying them on screen. I
implemented this fix for menu buttons, combo boxes, and tooltips; let me
know if there's anything I missed.
Scaling of the popups is purposefully ignored for now. Personally, I
think popup scaling isn't necessary but if it is required I can
implement it (also it would require doing more invasive things to the
code and I want to keep this as simple as possible.)
Before the fix: (with a modified version of the "Pan Zoom" web demo)
<video
src=https://github.com/emilk/egui/assets/104604363/a2d79757-c410-4815-8159-b64d6bd668ee>
After:
<video
src=https://github.com/emilk/egui/assets/104604363/48f8b1d1-d30f-44bf-961f-f10b85964a92>
The frame delay seems to be unavoidable unless we can move the popups
_after_ the layer transform is set (or apply the transforms to the popup
during painting). I think this would be better implemented once the
menu/popup/tooltip unification is done (#4669).
A useful debug-feature in egui is pressing down all modifiers keys and
hovering any widget to see its backtrace (only in `dev` builds).
Unfortunately this is incompatible with `panic="abort"`, something I
just now discovered.
So I removed `panic="abort"` from `Cargo.toml` for `dev` builds. If the
backtrace returns empty-handed, I also suggests this as a fix to the
user. Finally, I cleaned up the backtraces a bit, making them slightly
shorter and more readable.
When the layers are reordered at the end of the frame, the sublayers are
placed directly above their respective parents. This allows having Areas
inside Windows, e.g., for the pan-zoom container.
* Closes <https://github.com/emilk/egui/issues/4128>
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
# What
Adds `#[doc(alias = "top_left")]` as an alias for `left_top`, and so on
for `right_top`, `right_bottom`, `left_bottom`.
# Why
Extremely minor doc-only change, but I keep going to type "top left" to
look for the top left of a rectangle.
I'm unsure whether this is just a british-english thing or an
english-in-general thing, but `top left corner` is far more common than
`left top corner`.
These doc aliases don't conflict with anything, and mean that
rust-analyzer will suggest the correct function when I search for the
wrong thing.
This improves ergonomics and discoverability in my opinion, even if not
by much.
* Part of https://github.com/emilk/egui/issues/3653
<!--
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 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!
-->
It was surprising to me that this key wasn't sending `Event::Key` events
(only `Event::Text`), so I added a `Key` for it.
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
This example does not use the default features from eframe in order to
avoid accesskit, which panics when run from multiple threads, so it must
manually enable the other default features in order to compile correctly
on Linux.
* Closes <https://github.com/emilk/egui/issues/4682>
<!--
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 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!
-->
When layer transforms are present, the `Response::hover_pos` function in
egui returns `transform * pos` instead of `transform.inverse() * pos`.
This is incorrect and isn't consistent with [how egui calculates other
interaction
positions](https://github.com/emilk/egui/blob/master/crates/egui/src/context.rs#L1193).
See:
https://github.com/emilk/egui/blob/master/crates/egui/src/response.rs#L471
This PR fixes this bug, changing `transform` to `transform.inverse()`.
Nothing fancy here, just a one-line change. `scripts/check.sh` runs
successfully.
Below are videos of before and after with a modified version of the web
demo. I added another entry after
https://github.com/emilk/egui/blob/master/crates/egui_demo_lib/src/demo/pan_zoom.rs#L108
with a debug rectangle drawn at the return value of `hover_pos`:
```rust
(
egui::Pos2::new(120.0, 120.0),
Box::new(|ui, _state| {
let response =
ui.allocate_response(egui::Vec2::splat(128.0), egui::Sense::hover());
ui.painter().rect_filled(
egui::Rect::from_center_size(
response.hover_pos().unwrap_or_default(),
egui::Vec2::splat(8.0),
),
egui::Rounding::ZERO,
egui::Color32::DEBUG_COLOR,
);
response
}),
),
```
Without the fix:
https://github.com/emilk/egui/assets/104604363/241cfcab-88ab-459b-8f4d-3367da3aa392
With the fix:
https://github.com/emilk/egui/assets/104604363/e52a7263-44c7-42c1-be46-1ecadc025625
This fixes a bug which sometimes would make it possible to interact with
widgets that were outside the parent clip_rect.
Interaction with a widget is done with the `interact_rect`, which is the
intersection of the widget rect and the parent clip rect. If these
rectangles are disjoint (the widget is outside the parent clip rect),
this results in a _negative rectangle_ (a rectangle with a negative
width and/or height). The distance tests for negative rectangles were
broken, causing the bug.
* This is part of solving https://github.com/emilk/egui/issues/4475
* It is also likely this would have solved
https://github.com/emilk/egui/issues/4349 (which now has another fix for
it)
### Breaking changes
`Rect::distance_to_pos`, `distance_sq_to_pos`, `signed_distance_to_pos`
now all return `f32::INFINITY` if the rectangle is negative.
Emil Ernerfeldt