iris-aiandClaude Opus 5
ef815dadfd
A resize had its own mechanism: `reads_output` recorded that a widget had looked at the output, `update` scanned every active widget for one whose `output_px` had moved, marked it and its whole reader chain, and `resize_marks` kept those marks from counting as content dirtiness -- while `on_resize` answered the same question for every other box. Two answers to "does this drawing survive its box changing length", and the one that applied to the window ignored what the widget had declared. With the output held as the root of the chain there is one question. A resize offers the root widget its box again, `try_reuse` answers per axis from `on_resize`, and `redraws_under` prices the subtree. Gone with it: `reads_output`, `resized`, `resize_marks`, the scan, the eager reader marking, and the shallowest-first branch in `redraw_updates`, which only existed because resize marking worked differently -- the settle loop now has one order. Two things this needed. An unslotted widget may be reused when only its parent's box changed length: it has nothing of its own to write, and what it drew is a fraction of that box, so the slot above it already carries the change. And `root_readers` holds the widgets whose size came from the output rather than their own box -- `MaxSize` -- since no box of theirs need have changed; they are marked per axis, from a set kept as they draw rather than by scanning. `a_resize_does_not_redraw_what_the_shader_can_move` now says `Scale`, which is what it was always describing, and `a_resize_redraws_what_does _not_scale` is its other half. `ReadsWidth` declares `Scale` across the axis it does not read, so per-axis precision comes from the widget rather than from which output axis it happened to touch. Resize phase, seed 1 depth 8: 6.45M instructions per frame to 5.84M. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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:
- does what I want it to (text, images, video, animations)
- very easy to use ignoring ergonomic ref counting
- 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