When painting a scatter plot we sometimes want to paint hundreds of thousands of points (filled circles) on screen every frame.
In this PR the font texture atlas is pre-populated with some filled circled of various radii. These are then used when painting (small) filled circled, which means A LOT less triangles and vertices are generated for them.
In a new benchmark we can see a 10x speedup in circle tessellation, but the the real benefit comes in the painting of these circles: since we generate a lot less vertices, the backend painter has less to do.
In a real-life scenario with a lot of things being painted (including around 100k points) I saw tessellation go from 35ms -> 7ms and painting go from 45ms -> 1ms. This means the total frame time went from 80ms to 8ms, or a 10x speedup.
* Remove integration name (it is always eframe)
* Remove egui_web crate
* Move egui_web/CHANGELOG.md into eframe/CHANGELOG.md
* Remove all mentions of egui_web
* Remove epi crate and absorb into eframe
* egui_glow: only use puffin on native
* Remove WASM doc from CI (we don't generate it anyways!)
* Remove eframe::epi and improve eframe docs
* Returns openness in CollapsingResponse
* Make CollapsingState a building block for custom collapsing headers
* Add a demo of the custom collapsing header
* Revert to much simpler tree demo
* Add CollapsingState::is_open and CollapsingState::set_open
This looks like a Hyperlink, but doesn't do anything when clicked.
Or rather: it lets the user decide what happens on click.
Closes https://github.com/emilk/egui/issues/1152
* Move examples out of eframe/examples into examples/
Give each example a `Cargo.toml` and `src/main.rs`.
This makes it easier for people to use as templates.
* Update README.md with more deps needed on vanilla Ubuntu
* Install libgtk-3-dev on CI, hoping that will fix something
* Update to rust 1.60.0
* Rename the feature `convert_bytemuck` to `bytemuck`
* Rename the feature `serialize` to `serde`.
* Make use of the "weak dependency" cargo feature
* Set rust-version = "1.60" for all crates
* egui_glow: clipboard, links, persistence & winit are now opt-in features
Id clashes can cause subtle bugs.
egui already warns when the same Id is used to interact with different
parts of the screen.
This adds warnings about id clashes for some widgets that store state:
Grid, Plot, ScrollArea, Table.
The PR also adds `Context::check_for_id_clash` so users who create
their own widgets can add the same type of check.
* Rename AlphaImage to FontImage to discourage any other use for it
* Encode FontImage as f32 and postpone the alpha correction
* Interpret alpha coverage in a new, making dark text darker, improving contrast in bright mode
* Fix code that could lead to a possible deadlock.
Drop implementations are not called until the end of a statement. The statement changed in this commit therefore took 4 read locks on a RwLock which can lead to problems if a write lock is requested between any of these read locks. The code looks like it would only hold one lock at a time but it does not drop any of the locks until after the arithmatic operations complete, which leads to this situation. See https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=996079046184329f3a9df1cd19c87da8 to see this in action. The fix is to just take one lock and share it between the three calls to num_presses, letting it drop at the end of the scope.
* Fix code that may cause a deadlock in `MenuRoot::stationary_interaction`
The issue here is related to that in 9673b8f2a08302c10ffcfd063f2dbdec4301d3e2 in that the lock is not dropped when it is expected. Since the `RwLockReadGuard` produced by `ctx.input()` has a reference taken from it (one into `pointer`), the lock cannot be dropped until that reference is no longre valid, which is the end of the scope (in this case this function). The following `ctx.input()` then attempts to aquire a second lock on the `RwLock`, creating the same situation that was found in the referenced commit.
This has been resolved by holding one lock on the input for the whole function.
* Reference this PR from comments in the code for future maintainers
* Add the change to the changelog
* Use full link to PR
* Use full link to PR
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
This adds a callback (set by `Context::set_request_repaint_callback`)
which integration can use to wake up the UI thread.
eframe (egui_web and egui_glow) will use this, replacing
`epi::Frame::request_repaint`.
Existing code calling `epi::Frame::request_repaint` should be changed
to instead call `egui::Context::request_repaint`.
This is the first callback added to the egui API, which otherwise is
completely driven by data.
The purpose of this is to remove the confusion between the two
`request_repaint` methods (by removing one). Furthermore, it makes
`epi::Frame` a lot simpler, allowing future simplifications to it
(perhaps no longer having it be `Send+Sync+Clone`).
* Add Shape::Callback to do custom rendering inside of an egui UI
* Use Rc<glow::Context> everywhere
* Remove trait WebPainter
* Add glow::Context to epi::App::setup
* egui_web: use tracing crate
* egui_glow: use tracing crate
* Log at the debug level
* egui_demo_app: enable tracing to log to stdout
* Use tracing in egui-winit
* Add opt-in tracing support to egui
* Fix Orientation not exposed, although there are public fields with its type
* Implement formatters for X/Y axis labels
* Use array instead of separate X/Y formatters
* Swap axis formatters if charts are horizontal
* Review suggestions
* Refactor text layout: don't need &Fonts in all functions
* Replace indexing in Fonts with member function
* Wrap Fonts in a Mutex
* Remove mutex for Font::glyph_info_cache
* Remove RwLock around Font::characters
* Put FontsImpl and GalleyCache behind the same Mutex
* Round font sizes to whole pixels before deduplicating them
* Make TextStyle !Copy
* Implement user-named TextStyle:s
* round font size earlier
* Cache fonts based on family and size
* Move TextStyle into egui and Style
* Remove body_text_style
* Query graphics about max texture size and use that as font atlas size
* Recreate texture atlas when it is getting full
* Move texture allocation into epaint/egui proper
* Add TextureHandle
* egui_glow: cast using bytemuck instead of unsafe code
* Optimize glium painter
* Optimize WebGL
* Add example of loading an image from file
* Move all interior mutability from Context to CtxRef and make it a handle
* Rename `CtxRef` to `Context`
* The old `Context` is now `ContextImpl` and is non-pub
* Add benchmark Painter::rect
Co-authored-by: Daniel Keller <dklr433@gmail.com>
* Split `Event::Text` into `Text` and `Paste`
* Added explicit Event::Paste change
See #1043
* Link to PR in changelog (not the issue)
Co-authored-by: Emil Ernerfeldt <emil.ernerfeldt@gmail.com>
`begin_frame`, `end_frame` is more convenient when using egui in a game engine. In particular, 0.16.0 was incompatible with https://github.com/mvlabat/bevy_egui>.
Changes:
* New `BarChart` and `BoxPlot` diagrams
* New `FloatOrd` trait for total ordering of float types
* Refactoring of existing plot items
Co-authored-by: niladic <git@nil.choron.cc>
* Rename `CursorPair` to `CursorRange`
* Easymark editor: add keyboard shortcuts to toggle bold, italics etc
* Split up TextEdit into separate files
* Add TextEdit::show that returns a rich TextEditOutput object with response, galley and cursor
* Rename text_edit::State to TextEditState
This refactors the widget state storage introduced by @optozorax in https://github.com/emilk/egui/pull/257
* Unify the four buckets (`data`, `data_temp`, `id_data` and `id_data_temp`) into a single `data`.
* Less complexity, and also less chance of error (storing in one bucket, reading from another).
* Store data by `Id` and `TypeId`.
* Users can thus reuse the same `Id` to store many types.
* Uses a simple xor of id and typeid, which is fast and good since both id and typeid are already high-entropy hashes.
* Use different suffixes on the functions to pick if you want the data persisted or not (`get_temp`, `insert_persisted`, etc).
* Writing with one suffix and reading with the other works.
* To store state not bound to a specific `Id` (i.e. only based on type), use the new `Id::null` as the key.
* Make egui_glow and opt-in backend for eframe
* Add egui_glow to Cargo.toml and to CI
* Reference egui_glow where egui_glium is mentioned
* Remove path-patches from root Cargo.toml
* Add instructions on how to enable the glow backend of eframe
* Rename epaint feature "persistence" to "serialize"
* Add separate "serialize" feature to egui
* egui_demo_lib: separate serialize and persistence features
* Add App::persist_native_window and App::persist_egui_memory
Controls what gets persisted
Label text will now be centered, right-aligned and/or
justified based on the layout.
Galleys are no longer always pivoted in the left top corner,
so now have a Rect rather than just a size.
This PR introduces a completely rewritten text layout engine which is simpler and more powerful. It allows mixing different text styles (heading, body, etc) and formats (color, underlining, strikethrough, …) in the same layout pass, and baked into the same `Galley`.
This opens up the door to having a syntax-highlighed code editor, or a WYSIWYG markdown editor.
One major change is the color is now baked in at layout time. However, many widgets changes text color on hovered. But we need to do the text layout before we know if it is hovered. Therefor the painter has an option to override the text color of a galley.
## Performance
Text layout alone is about 20% slower, but a lot of that is because more tessellation is done upfront. Text tessellation is now a lot faster, but text layout + tessellation still lands at a net loss of 5-10% in performance. There are however a few tricks to speed it up (like using `smallvec`) which I am saving for later. Text layout is also cached, meaning that in most cases (when all text isn't changing each frame) text tessellation is actually more important (and that's more than 2x faster!).
Sadly, the actual text cache lookup is significantly slower (300ns -> 600ns). That's because the `TextLayoutJob` is a lot bigger (it has more options, like underlining, fonts etc), so it is slower to hash and compare. I have an idea how to speed this up, but I need to do some other work before I can implement that.
All in all, the performance impact on `demo_with_tesselate__realistic` is about 5-6% in the red. Not great; not terrible. The benefits are worth it, but I also think with some work I can get that down significantly, hopefully down to the old levels.
* First pass (missing rendering the bar)
* Render horizontal bars, and change Window scroll API
* emath: add impl Index + IndexMut for Align2
* Scrolling: fix subtle sizing bugs
* Add horizontal scrolling to color test
* try to wrap content before showing scrollbars, + add auto-shrink option
* Add hscroll to the misc demo window
* Fix for putting wrapping labels in an infinitely wide layout
* Add a egui_asserts to protect against nans in the layout engine
* Add line about horizontal scrolling to changelog
* Add example to docs of ScrollArea
* code cleanup
* Fix incorrect max_width/height of panels
* Fix set_width/set_min_width/set_height/set_min_height
Closes https://github.com/emilk/egui/issues/647
Broke in https://github.com/emilk/egui/pull/629
* Fix expand_to_include_x/expand_to_include_y
* Make minimum grid column width propagate properly
* Expand cursor when max_rect expands
* Add ui.expand_to_include_y
* Only expand cursor in advance
* demo: clean up font_book code
* Fix: Make sure `TextEdit` contents expand to fill width if applicable
* ProgressBar: minimum width and fix for having it in an infinite layout
* clippy fix
* Add methods for optionally displaying the background/axes of a `Plot`
These are particularly useful when using the `Plot` widget as an overlay
over an existing grid or some other content.
* Allow for showing each axis of a `Plot` individually
* Fix case where `Plot`'s `min_auto_bounds` can be ignored after first
I ran into an issue using `Plot` within my timeline widget where if I
zoom in and out of the timeline (not the plot), the `Plot` instances
would ignore the necessary changes to the `include_x` calls and in turn
would become skewed and misaligned with the timeline below.
This changes the `Plot` to check whether or not `min_auto_bounds` have
changed and, if so, reset the memory and recalculate the bounds.
See #562 for an image of my current use case.
* Carry hidden_items when updating plot for changed bounds
* added new line styles
* update changelog
* fix#524
Add missing functions to `HLine` and `VLine`
* add functions for creating points and dashes from a line
* apply suggestions
* clippy fix
* address comments
This allows for resizaeable grids, where the last column will be given the remainder of the width.
To demonstrate, the widget gallery window is now resizeable.
* Added plot items:
* Arrows, also called "Quiver plots" in matplotlib etc.
* Convex polygons
* Text
* Images
Other changes:
* Make HLine/VLine into PlotItems as well.
* Add a "fill" property to Line so that we can fill/shade the area between a line and a horizontal reference line.
* Add stems to Points, which are lines between the points and a horizontal reference line.
* Allow using .. when specifying ranges for values generated by explicit callback functions, as an alias for f64::NEG_INFINITY..f64::INFINITY
* Allow using ranges with exclusive end bounds for values generated by parametric callback functions to generate values where the first and last value are not the same.
* update changelog
* add legend background
Emil Ernerfeldt
Luke Newcomb
Alejandro Perea
trevyn
Jan Haller
Friz64
Lucas Kent
Collin Kemper
Zachary Kohnen
Juan Campa
Benedikt Terhechte
cat-state
Sven Niederberger
Alexander
Erlend Walstad
Manuel Innerhofer
nongiach
AlexxxRu
Simon Werner
kud1ing
Sven Niederberger
Cristian Dinu
mitchmindtree
Ezra Barrow
Jay Oster