I believe that the underlying issue that caused the stuck modifier keys
was resolved in the 0.29 winit keyboard refactor.
This probably needs to tested on other desktop platforms however since I
am only able to test this on windows 11.
* Closes <https://github.com/emilk/egui/issues/2332>
* Closes https://github.com/emilk/egui/issues/3444
* Closes https://github.com/emilk/egui/issues/865
On a touch screen, if you press down on a widget and hold for 0.6
seconds (`MAX_CLICK_DURATION`), it will now trigger a secondary click,
i.e. `Response::secondary_clicked` will be `true`. This means you can
now open context menus on touch screens.
This is a refactor on the way to add support for opening context menus
on touch screens via press-and-hold.
This PR changes what `InputState::button_clicked` does (it was ver badly
named before), and also changes `Response::clicked_by` to no longer be
true if clicking with keyboard (i.e. a widget has keyboard focus and the
user presses Space or Enter).
This PR fixes two issues related to `ScrollArea`.
1) When a `ScrollArea` would have `drag_to_scroll` set to `false` (e.g.
because some custom logic is at play or some other reason), it would not
animate to the `target_offset`, effectively making
`Response::scroll_to_me()` ineffective.
2) Single-direction `ScrollArea`s would leak the `scroll_target`'s other
direction. In certain specific circumstances (e.g. an horizontal area
nested in a vertical one, or inversely), this _could_ work as intended,
but in many other cases it could cause unwanted effects. With this PR,
both `scroll_target` directions are consumed by nearest enclosing
`ScrollArea`, regardless of the actually enabled scroll axes.
Previously, any frames in flight (`requestAnimationFrame`) on web were
not being cancelled (`cancelAnimationFrame`) when `WebRunner::destroy`
was called. If a user called `destroy`, then immediately removed the
canvas from the DOM, eframe could panic with a "failed to find (canvas)
element by id" error message.
This PR changes two things:
- The canvas element is directly referenced everywhere it's needed
instead of being looked up by `canvas_id`[^1]
- The RAF handle is stored in `WebRunner` and `cancelAnimationFrame` is
called on it inside of `WebRunner::destroy`[^2]
[^1]: The WebGL/WGPU backends were already holding onto the canvas (and
associated GPU context), so the change is just converting all the
`get_element_by_id` lookups to retrieve the canvas from the web runner
handle.
[^2]: There is only ever one frame in flight, so we store it directly as
a scalar field.
<!--
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 cranky`.
* 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/4144>
* Closes#3846
* Closes#3878
Dear emilk.
Leaving aside other function,
I think this is all you need to fix to patch the panic that occurs when
Shift + TAB.
Thank you, emilk.
Hi!
I'm using egui for the UI of a VST3/Clap plugin, and this kind of
environment is rather picky on performance. I use
[assert_no_alloc](https://crates.io/crates/assert_no_alloc) to make sure
the audio thread is never allocating. The audio thread may request a
repaint of the GUI tho, and this is where a saw that it may allocate
when tracing the repaint reason.
Turns out it's not necessary, `Location::caller` is `'static`, so using
a `&'static str` instead of a `String` in `RepaintCause::file` will just
work, so this PR just does that.
Update modifier state from web mouse events. This allows modifiers to be
correctly updated when the window is not in focus but the mouse is still
moving over the window.
When scaling an `egui::Shape` of variant `Rect` using the new
`transform` function, corner rounding isn't taken into account.
The fix is to multiply the rounding by the scaling factor.
* Closes <https://github.com/emilk/egui/issues/4152>
# What's New
* eframe: Added `App::raw_input_hook` allows for the manipulation or
filtering of raw input events
A filter applied to raw input before [`Self::update`]
This allows for the manipulation or filtering of input events before
they are processed by egui.
This can be used to exclude specific keyboard shortcuts, mouse events,
etc.
Additionally, it can be used to add custom keyboard or mouse events
generated by a virtual keyboard.
* examples: Added an example to demonstrates how to implement a custom
virtual keyboard.
[eframe-custom-keypad.webm](https://github.com/emilk/egui/assets/1274171/a9dc8e34-2c35-4172-b7ef-41010b794fb8)
* Remove the leading slash from the path if the target OS is Windows.
This is because Windows paths are not supposed to start with a slash.
For example, `file:///C:/path/to/file` is a valid URI, but
`/C:/path/to/file` is not a valid path.
* Use the input URI consistently as the cache key.
Currently, the cache key is inconsistently set as either the path or the
URI, while the forget key is always the URI. This inconsistency should
be resolved.
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
The `Shape::translate` method has been replaced with `Shape::transform`,
which introduces breaking changes that could negatively impact existing
users.
This patch adds a `Shape::translate` wrapper to prevent these breaking
changes.
- Introduces `WidgetType::ProgressIndicator` and maps it to the
corresponding AccessKit role.
- A `Spinner` is now exposed as a widget indicating a progress for which
a completion state is not known.
- On the other hand, a `ProgressBar` reports a completion state and can
possibly be labeled. Note that a label is not used if not explicitly
asked by the user, as it would be redundant information. Assistive
technologies prefer the numerical value so they can, for instance, emit
beeps of which the frequency rise as the completion state increase. I
had to call `floor` on the progression as it seems all the ATs I tested
would round the value, hence reporting something different than what is
displayed on the label.
* Closes https://github.com/emilk/egui/issues/4113
Previously the `Id` of the menu `Area` was using the same id as the
thing that was clicked (i.e. the button opening menu), which lead to id
clashes
We plan to store input data for creating automated tests, hence the need
for more serde derives on input related structs.
---------
Co-authored-by: Georg Weisert <georg.weisert@freshx.de>
I often write constants at the top of my widget files, as a "config". I
kept writing stuff like that :
```rust
const DEFAULT_INNER_MARGIN: Margin = Margin { left: 17., right: 17., top: 7., bottom: 7. };
```
So I prefixed constructors for `Margin`, `Rounding` and `Shadow` const.
No code was changed.
I also added a `Shadow::new()` for similar reasons.
Raw mouse movement is unaccelerated and unclamped by screen boundaries,
and does not relate to any position on the screen.
It is useful in certain situations such as draggable values and 3D
cameras, where screen position does not matter.
https://github.com/emilk/egui/assets/1700581/1400e6a6-0573-41b9-99a1-a9cd305aa1a3
Added `Event::MouseMoved` for integrations to supply raw mouse movement.
Added `Response:drag_motion` to get the raw mouse movement, but will
fall back to delta in case the integration does not supply it.
Nothing should be breaking, but third-party integrations that can send
`Event::MouseMoved` should be updated to do so.
Based on #1614 but updated to the current version, and with better
fallback behaviour.
* Closes#1611
* Supersedes #1614
Add an alternative method for creating a [`ColorImage`] that accepts
`Iterator` as the argument. It can be useful when `&[u8]` is not
available but the iterator is.
<!--
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.
* 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 cranky`.
* 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 you PR, but my time is limited!
-->
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
Each cell in a table now has a `Ui` with a unique `Id` based on the row
and column.
This avoids Id-clashes, e.g. when putting a `CollapsingHeader` in a
table cell.
* Closes https://github.com/emilk/egui/issues/3936
* Closes https://github.com/emilk/egui/issues/3923
* Closes https://github.com/emilk/egui/pull/4058
The interaction code is now done at the start of the frame, using stored
`WidgetRect`s from the previous frame.
The intention is that the new interaction code should be more accurate,
making it easier to hit widgets, and better respecting the rules of
overlapping widgets.
There is a new `style::Interaction::interact_radius` controlling how far
away from a widget the cursor can be and still hit it. This helps big
fat fingers hit small widgets on touch screens.
This PR adds a new `Context::read_response` which lets you read the
`Response` of a `Widget` _before_ you create the widget. This can be
used for styling, or for reading the result of an interaction early (to
prevent frame-delay) for a widget you add late (so it is on top of other
widgets).
# ⚠️ BREAKING CHANGES
`Memory::dragged_id`, `Memory::set_dragged_id` etc have been moved to
`Context`.
The semantics for `Context::dragged_id` is slightly different: a widget
is not considered dragged until egui it is sure this is not a
click-in-progress. For a widget that is only sensitive to drags, that is
right away, but for widgets sensitive to both clicks and drags it is not
until the mouse has moved a certain distance.
# TODO
* [x] Fix panel resizing
* [x] Fix scroll hover weirdness
* [x] Fix Resize widget
* [x] Fix drag-and-drop
* [x] Test all of egui_demo_app
* [x] Change `is_dragging` API
* [x] Consistent naming of start/stop or begin/end drag
* [x] Test `egui_tiles`
* [x] Test Rerun
* [x] Document
* [x] Document breaking changes in PR description
* [x] Test one final time
# Saving for a later PR
* [ ] Fix https://github.com/emilk/egui/issues/4047
* [ ] Specify what the response order for e.g. `ui.horizontal` is
I think both these can be fixed if each `Ui` registers themselves as a
`WidgetRect`, with the possibility to interact with it later, as if the
interaction was under all widgets on top of it.
⚠️ Removes `Context::translate_layer`, replacing it with a sticky
`set_transform_layer`
Adds the capability to scale layers.
Allows interaction with scaled and transformed widgets inside
transformed layers.
I've also added a demo of how to have zooming and panning in a window
(see the video below).
This probably closes#1811. Having a panning and zooming container would
just be creating a new
`Area` with a new id, and applying zooming and panning with
`ctx.transform_layer`.
I've run the github workflow scripts in my repository, so hopefully the
formatting and `cargo cranky` is satisfied.
I'm not sure if all call sites where transforms would be relevant have
been handled. This might also be missing are transforming clipping
rects, but I'm not sure where / how to accomplish that. In the demo, the
clipping rect is transformed to match, which seems to work.
https://github.com/emilk/egui/assets/70821802/77e7e743-cdfe-402f-86e3-7744b3ee7b0f
---------
Co-authored-by: tweoss <fchua@puffer5.stanford.edu>
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
Eris
ming08108
rustbasic
Emil Ernerfeldt
Antoine Beyeler
Jan Procházka
lomekragow
Vincent Alsteen
CrazyCraftix
Varphone Wong
Arnold Loubriat
Cameron Taggart
Georg Weisert
0Qwel
Mike Tsao
Giantblargg
wangxiaochuTHU
vincent-sparks
Francis Chua
YgorSouza
Dunfan Lu