This was initially a PR to add kitdiff, but this now lives in it's own
crate: https://github.com/rerun-io/kitdiff
I needed to make the image loaders public, this way it's possible to
compose image loaders together (which allowed me to create a image diff
loader that uses two other image loaders). But you can't use the
`ctx.try_load_image` since that would deadlock, so you have to store a
reference to the other loader in the wrapping loader.
Currently, DatePickerButton will close without saving whenever a user
clicks a dropdown from year/month/date. The issue is caused because the
system mistakenly interprets the user as clicking off of the calendar.
This is unexpected and creates an unpleasant experience for the user.
This change now allows the user to use the dropdowns as expected; it
will close on save or cancel. The calendar still closes when user clicks
off of it, as before. The changes here are made in:
crates/egui_extras/src/datepicker/button.rs
I will admit that I am not an experienced Rust developer. The changes
were made with the help of ChatGPT 4.0.
I have tested the changes locally, as I am using the date picker in my
project.
* Closes <https://github.com/emilk/egui/issues/THE_RELEVANT_ISSUE>
* [x] I have followed the instructions in the PR template
---------
Co-authored-by: Lucas Meurer <hi@lucasmerlin.me>
* Recently CI runs started to hang randomly:
https://github.com/emilk/egui/actions/runs/17427449210/job/49477714447?pr=7359
This fixes the deadlock and adds the basic deadlock detection we also
added to Mutexes in #7468.
Also, interestingly, the more sophisticated deadlock detection (behind
the deadlock_detection feature) didn't catch this for some reason. I
wonder why it exists in the first place, when parking_lot also has built
in deadlock detection? It also seems to make tests slower, widget_tests
usually needs ~30s, with the deadlock detection removed its only ~12s.
Change the theme selection of the code editor
Before
- was showing the window theme selector
After
- is showing the code editor theme selector
<img width="327" height="262" alt="image"
src="https://github.com/user-attachments/assets/50776218-26cc-4f11-9b91-6f902c022394"
/>
* [x] I have followed the instructions in the PR template
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
This is the same issue that was fixed for the http bytes loader in
239ade9a59
* [x] I have followed the instructions in the PR template
----------------
Funnily, [this
comment](https://github.com/emilk/egui/issues/3747#issuecomment-1872192997)
describes exactly how I encountered this issue:
> That assert is wrong if something calls forget between the start of
the request and the end of it.
I'm displaying lots of images in a scrolling grid (20 or so visible at a
time). It seems like image textures are never freed up automatically so
it stacks up a lot meaning I have to unload the image textures manually
with `egui::Context::forget_image()` in each `eframe::App::update()`
call for the images that are no longer shown when scrolling.
---------
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
This pull request modifies the `BytesLoader` implementation for
`FileLoader` in `crates/egui_extras/src/loaders/file_loader.rs` to
improve thread safety and handle unexpected states more gracefully.
### Changes to thread safety and state handling:
* Updated the cache logic to check if the `uri` exists in the cache
before inserting the result. If the `uri` is not found, a log message is
added to indicate the loading was canceled. This change prevents
overwriting cache entries unexpectedly.
* Closes <https://github.com/emilk/egui/issues/6755>
* [x] I have followed the instructions in the PR template
<!--
Please read the "Making a PR" section of
[`CONTRIBUTING.md`](https://github.com/emilk/egui/blob/main/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/3964>
* [X] I have followed the instructions in the PR template
This PR adds support for syntax highlighting with custom
`syntect::parsing::SyntaxSet` and `syntect::highlighting::ThemeSet`. It
adds a new `egui_extras::highlight_with` function (enabled with `feature
= "syntect"`), which takes a new public`struct SyntectSettings`
containing the syntax and theme sets.
```rust
let mut builder = SyntaxSetBuilder::new();
builder.add_from_folder("syntax", true).unwrap();
let ps = builder.build();
let ts = syntect::highlighting::ThemeSet::load_defaults();
let syntax =
egui_extras::syntax_highlighting::SyntectSettings { ps, ts };
// ...elsewhere
egui_extras::syntax_highlighting::highlight_with(
ui.ctx(),
ui.style(),
&theme,
buf,
"rhai",
&syntax,
);
```
There's a little bit of architectural complexity, but it all emerges
naturally from the problem's constraints.
Previously, the `Highlighter` both contained the `syntect` settings
_and_ implemented `egui::cache::ComputerMut` to highlight a string; the
settings would never change. After this change, the `syntect` settings
have become part of the cache key, so we should redo highlighting if
they change. The `Highlighter` becomes an empty `struct` which just
serves to implement `ComputerMut`.
`SyntaxSet` and `ThemeSet` are not hasheable themselves, so can't be
used as cache keys direction. Instead, we can use the *address* of the
`&SyntectSettings` as the key. This requires an object with a custom
`Hash` implementation, so I added a new `HighlightSettings(&'a
SyntectSettings)`, implementing `Hash` using `std::ptr::hash` on the
reference. I think using the address is reasonable – it would be _weird_
for a user to be constantly moving around their `SyntectSettings`, and
there's a warning in the docstring to this effect.
To work _without_ custom settings, `SyntectSettings::default` preserves
the same behavior as before, using `SyntaxSet::load_defaults_newlines`
and `ThemeSet::load_defaults`. If the user doesn't provide custom
settings, then we instantiate a singleton `SyntectSettings` in `data`
and use it; this will only be constructed once.
Finally, in cases where the `syntect` feature is disabled,
`SyntectSettings` are replaced with a unit `()`. This adds a _tiny_
amount of overhead – one singleton `Arc<()>` allocation and a lookup in
`data` per `highlight` – but I think that's better than dramatically
different implementations. If this is an issue, I can refactor to make
it zero-cost when the feature is disabled.
Add the ability to set the `DatePickerButton`'s start and end years via
new `start_year` and `end_year` methods.
Continue to use the existing today - 100 years and today + 10 years
behavior if a year is not specified.
* This more fully closes <https://github.com/emilk/egui/issues/3597> and
expands on <https://github.com/emilk/egui/pull/3599>.
* [x] I have followed the instructions in the PR template
Today each widget does its own custom layout, which has some drawbacks:
- not very flexible
- you can add an `Image` to `Button` but it will always be shown on the
left side
- you can't add a `Image` to a e.g. a `SelectableLabel`
- a lot of duplicated code
This PR introduces `Atoms` and `AtomLayout` which abstracts over "widget
content" and layout within widgets, so it'd be possible to add images /
text / custom rendering (for e.g. the checkbox) to any widget.
A simple custom button implementation is now as easy as this:
```rs
pub struct ALButton<'a> {
al: AtomicLayout<'a>,
}
impl<'a> ALButton<'a> {
pub fn new(content: impl IntoAtomics) -> Self {
Self { al: content.into_atomics() }
}
}
impl<'a> Widget for ALButton<'a> {
fn ui(mut self, ui: &mut Ui) -> Response {
let response = ui.ctx().read_response(ui.next_auto_id());
let visuals = response.map_or(&ui.style().visuals.widgets.inactive, |response| {
ui.style().interact(&response)
});
self.al.frame = self
.al
.frame
.inner_margin(ui.style().spacing.button_padding)
.fill(visuals.bg_fill)
.stroke(visuals.bg_stroke)
.corner_radius(visuals.corner_radius);
self.al.show(ui)
}
}
```
The initial implementation only does very basic layout, just enough to
be able to implement most current egui widgets, so:
- only horizontal layout
- everything is centered
- a single item may grow/shrink based on the available space
- everything can be contained in a Frame
There is a trait `IntoAtoms` that conveniently allows you to construct
`Atoms` from a tuple
```
ui.button((Image::new("image.png"), "Click me!"))
```
to get a button with image and text.
This PR reimplements three egui widgets based on the new AtomLayout:
- Button
- matches the old button pixel-by-pixel
- Button with image is now [properly
aligned](https://github.com/emilk/egui/pull/5830/files#diff-962ce2c68ab50724b01c6b64c683c4067edd9b79fcdcb39a6071021e33ebe772)
in justified layouts
- selected button style now matches SelecatbleLabel look
- For some reason the DragValue text seems shifted by a pixel almost
everywhere, but I think it's more centered now, yay?
- Checkbox
- basically pixel-perfect but apparently the check mesh is very slightly
different so I had to update the snapshot
- somehow needs a bit more space in some snapshot tests?
- RadioButton
- pixel-perfect
- somehow needs a bit more space in some snapshot tests?
I plan on updating TextEdit based on AtomLayout in a separate PR (so
you could use it to add a icon within the textedit frame).
This fixes bugs related to how an `Image` follows the size of an SVG.
We track the "source size" of each image, i.e. the original width/height
of the SVG, which can be different from whatever it was rasterized as.
* 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.
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)
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.
* 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
**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
* 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>
This is the same change as in #4069 but as this is stale I wanted to
reopen a non stale PR
Modifies ImageLoader's load function to use background threads for the
image decoding work. This avoids blocking the main thread that is
especially noticeable when loading many images at once.
This was modelled after the other loader implementations that also use
threads.
* Closes <https://github.com/emilk/egui/issues/5375>
* [x] I have followed the instructions in the PR template
---------
Co-authored-by: lucasmerlin <lucasmeurer96@gmail.com>
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
Enabled the `missing_assert_message` lint
* [x] I have followed the instructions in the PR template
---------
Co-authored-by: Lucas Meurer <lucasmeurer96@gmail.com>
Hi !
I'm using egui and egui_extra to build a demo tool for a crate I'm
building. In this crate I load favicons from a lot of websites, and I
have some failures on some .ico files. After tracking this thing down, I
guess there is two issues (kinda), in the image crate,
'image/vnd.microsoft.icon' isn't recognized as the valid mime for an ico
image (I created a [PR](https://github.com/image-rs/image/pull/2434)),
and the code to detect if a given mime type is compatible with the image
crate decoders in this crate do not support multi-mime type formats.
[ImageFormat::to_mime_type](85f2412d55/src/image.rs (L216C12-L216C24))
is only returning one mime for a given format (which is fine), we
compare the result of this method to guess if the format is
valid/enabled. Retriveing the correct format using
[ImageFormat::from_mime_type](85f2412d55/src/image.rs (L166))
would allow more mime to be considered valid since multiple mime can
match the same format.
The same applies to the extension detection, which I also modified to
stay consistent
Thanks
Hi, after upgrading to 0.31.0 all of my beautiful static webp images
started failing to load. I use the image_loader to load those via the
`image` crate.
I noticed that with 0.31.0 there are additions to how animated image
types are handled with frames and such. And with those changes the frame
index is attached to the uri at the end. This was problematic for the
image_loader, because it wasn't updated to handle that frame tag at the
end of the uri, so when looking up the bytes, it would fail to match the
uri in the bytes cache (the bytes were being saved without the frame
index, but attempting to be fetched _with_ the frame index).
This fixes the image_loader for me with webp & gif. They don't load the
animations, but I think that is because I don't have the custom
image_loader set up so I'm not worried about that for myself. I'm not
sure if that part is problematic in general, or if its just the way I
have my features set up.
You can recreate the issue on master by swapping out the dependency
features in the `images` example like this:
```
# egui_extras = { workspace = true, features = ["default", "all_loaders"] }
# env_logger = { version = "0.10", default-features = false, features = [
# "auto-color",
# "humantime",
# ] }
# image = { workspace = true, features = ["jpeg", "png"] }
egui_extras = { workspace = true, features = ["image", "svg"] }
env_logger = { version = "0.10", default-features = false, features = [
"auto-color",
"humantime",
] }
image = { workspace = true, features = ["jpeg", "png", "webp", "gif"] }
```
* [x] I have followed the instructions in the PR template
---------
Co-authored-by: lucasmerlin <lucasmeurer96@gmail.com>
Breaking change!
* `Rounding` -> `CornerRadius`
* `rounding` -> `corner_radius`
This is to:
* Clarify
* Conform to other systems (e.g. Figma)
* Avoid confusion with `GuiRounding`
Current WebP loader assumes all WebP images to be RGBA, which is the
case if the image is animated (that's what `image` crate assumes at
least). Static images can instead choose to exclude its alpha channel,
though it seems to be more of a default choice to include it, even if
it's not being utilized. Currently, loading a static RGB WebP image will
cause a panic when `ColorImage::from_rgba_unmultiplied` gets called in
the loader
```
thread 'main' panicked at /home/aely/.cargo/git/checkouts/egui-226fc7cdd51201c1/f87219d/crates/epaint/src/image.rs:97:9:
assertion `left == right` failed
left: 29184
right: 21888
```
[ x ] I have ~~followed~~ _read_ the instructions in the PR template
Unfortunately i had several issues:
- Some snapshot-tests didn't run successfully on osx. diff shows errors
around fonts or missing menu items)
- cargo clippy doesn't run successfully (egui_kittest cannot find `wgpu`
and `image`)
- ./scripts/check.sh had other issues on my system (env: python: No such
file or directory), even if python3 can be called via python in my shell
Is there a system independent, standard way to run these tools (e.g. via
Docker?)
I submit the pr anyway, because there changes are very simple and
shouldn't cause issues.
Emil Ernerfeldt
Lucas Meurer
Nicolas
Aleksandr Strizhevskiy
n4n5
Sergey Ukolov
Vanadiae
Matt Keeter
Zach Bateman
Lucas Meurer
MStarha
Christopher Cerne
Aarni Koskela
Nicolas Gomez
Markus Krause
tyssyt
Onè
Braden Steffaniak
Aely
Markus Ineichen