iris-aiandClaude Fable 5.1 62a16b5608 Re-ask a dirty widget at its offer locally instead of deferring to its parent
A local redraw refused any widget whose given box was not as long as its
offer and marked its parent instead. Under the frame/extent protocol that
is nearly every widget beneath a self-sized container: a span hands its
children its own placement across itself, which is `FULL` while the span
is measured and its answer once it is placed, so the children's offer and
given frames differ on every such axis. A `many` frame at seed 13, depth
8 escalated 43 marks along chains up to seven levels and redrew 508 of
583 active widgets where e44dea3 redraws 159.

Retain the offer's frame beside the given one and ask the offer question
locally: the offer frame composed where the given one is, at the offer's
lengths and placement, then place at the given box where the two differ.
Seed 13 `many` goes from 4.73 ms to 1.28 ms against e44dea3's 0.90, and
294 distinct widgets a frame; size, scroll and repaint are unchanged.

Not sound yet: the suite, the debug oracle and the shrinker at 400 trees
of depth 5 pass, but the oracle at 1000 seeds of depth 6 diverges on seed
532 under reorder and seed 398 under every-size. Both reduce to a
self-sized container whose answer changes under a local redraw; the
reduced plans are in docs/HANDOFF.md of ai-app-2.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-17 20:51:36 -04:00
2026-01-05 17:01:09 -05:00
2025-11-17 16:08:29 -05:00
2026-01-29 16:39:19 -05:00
2026-09-13 00:24:29 -04:00

iris

My experimental attempt at a rust ui library (also my first ui library).

It's currently designed around using retained data structures (widgets), rather than diffing generated trees from data like xilem or iced. This is an experiment and I'm not sure if it's a good idea or not.

Examples are in examples, eg. cargo run --example tabs.

Goals, in general order:

  1. does what I want it to (text, images, video, animations)
  2. very easy to use ignoring ergonomic ref counting
  3. reasonably fast / efficient (a lot faster than electron, save battery life, try to beat iced and xilem)

dev details

not targeting web rn cause wanna use actual nice gpu features & entire point of this is to make desktop apps / not need a web browser

general ideas trynna use rn / experiment with:

  • retained mode
  • specifically designed around wgpu so there's no translation
  • postfix functions for most things to prevent unreadable indentation (going very well)
  • events can be done directly where you draw the widgets
  • almost no macros in user code & actual LSP typechecking (variadic generics if you can hear me please save us)
  • relative anchor + absolute offset coord system (+ "rest" / leftover during widget layout)
  • single threaded ui & pass context around to make non async usage straightforward (pretty unsure about this)
  • widgets store outside of the actual rendering so they can be moved around and swapped easily (unsure about this but seems to work good for now)

under heavy initial development so not gonna try to explain status, maybe check TODO for that; sizable chance it gets a rewrite once I know everything I need and what seems to work best

it's called iris because it's the structure around what you actually want to display and colorful

S
Description
No description provided
Readme
4.1 MiB
0 Stars 1 Watchers 0 Forks
Languages
Rust 96.3%
Shell 2.1%
WGSL 1.6%