fission-render-vello
Vello rendering backend for the Fission UI framework.
fission-render-vello is the primary GPU rendering backend for Fission. It implements theRenderer trait by translating DisplayList operations into a Vello Scene graph, which is
then rasterized on the GPU via wgpu. Text shaping and layout are handled byParley through the VelloTextMeasurer, which also
implements the TextMeasurer trait for the layout engine. The crate includes caches for text
layouts (simple and rich), decoded images, and parsed SVG paths to keep frame times low. Key types
| |
|---|
| Implements fission_render::Renderer. Writes display ops into a Vello Scene using the current affine transform stack. Handles rectangles, text, rich text (with per-run styling and background highlights), images, SVG paths, inline SVG, and box shadows. |
| Implements fission_layout::TextMeasurer. Wraps a Parley FontContext and LayoutContext with an LRU-style layout cache (4096 simple entries, 2048 rich entries). Provides measure, measure_rich_text, hit_test, get_line_metrics, and get_caret_position. |
| A simple [u8; 4] RGBA brush type used as the generic parameter for parley::layout::Layout. |
Supported display ops
DrawRect with fill, stroke, rounded corners, and box shadow
DrawText with font size, color, underline, and optional caret rendering
DrawRichText with per-run font size, color, underline, and background highlight
DrawImage with Fill, Contain, Cover, and None fit modes (image cache)
DrawPath from SVG path data
DrawSvg with viewBox-aware scaling and fill/stroke (SVG parse cache)
DrawSurface (video surface placeholder)
Save / Restore / ClipRect / ClipRoundedRect / Translate / Transform
Web font fallback
The Web renderer uses Parley for layout and shaping. Browser builds cannot enumerate fonts
installed on the user's device, so a missing glyph would otherwise become a tofu box. When a
Web text run contains a missing glyph, Fission rasterizes that run with Canvas 2D and the
browser's own font fallback stack, then paints the resulting image in the Vello scene. This
lets installed CJK and emoji fonts work without embedding large fallback files in every app.
The fallback image uses the width and vertical metrics already assigned by Parley and is cached
for reuse across frames.
Usage example
use fission_render_vello::{VelloRenderer, VelloTextMeasurer};
use parley::FontContext;
use std::sync::{Arc, Mutex};
use vello::Scene;
// Set up text measurement
let font_cx = Arc::new(Mutex::new(FontContext::new()));
let measurer = Arc::new(VelloTextMeasurer::new(font_cx));
// Render a frame
let mut scene = Scene::new();
let mut renderer = VelloRenderer::new(&mut scene, measurer, 2.0 /* scale factor */);
renderer.render(&display_list)?;
// `scene` is now ready to be submitted to a Vello `Renderer` backed by wgpu.
Status
Supported -- this is the default and recommended rendering backend for Fission on desktop.
Used in production by fission-shell-desktop.
License