Shape a text once per width, not once per ask

A container measures a child by drawing it in a box it may not keep, so
one layout asks a text for a dozen widths and comes back to widths it
has already had -- the hottest text in the depth-8 tree draws 32 times.
Each ask re-ran the shaper, because the two caches in front of it held
one entry each and a trial width alternating with a final width evicts
the answer about to be wanted again. `perf record` put 63% of a resize
frame in text and 0.9% in `draw_inner`.

So keep more than one: a bounded store of shapings on `TextData`, keyed
by the text, the attrs and the width, holding the parley layout and the
glyphs placed from it. Bounding the store rather than each buffer is
what keeps it a fixed cost -- +4 MB on a tree of 4,000 texts, which is
19 MB less than the code before #16 holds after the same resizes.

`TextBuffer` now holds the glyphs of the shaping it is drawn as, which
is where `TextView::tex` was. That leaves one place to invalidate rather
than two, so the `MutDetect` flags on a view's text and attrs have no
reader and go, along with the `buf.changed = true` after every edit.

On a 40-row tree of distinct random paragraphs, 500 resize frames:
124.2M instructions per frame before, 17.7M after, and 45.9M when the
width never repeats. The five reference renders and the resize render
are byte-identical, and the 100-seed sweep passes.

`tests/revision_cost.rs` is that tree, written in the API subset
`43ce8c7` shares so the same source measures the code this replaced.
Report the worst frame and p99 beside the median, since a stutter is
what somebody sees. Count glyph placements, and count a text render per
ask rather than per shaping, so the store cannot hide how many times a
layout drew the same text.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

core/src/layout_diagnostics.rs

@@ -54,10 +54,11 @@ pub(crate) enum Counter {
     TextRenders,
     TextShapeHits,
     TextShapes,
+    GlyphPlacements,
 }
 
 impl Counter {
-    const COUNT: usize = Self::TextShapes as usize + 1;
+    const COUNT: usize = Self::GlyphPlacements as usize + 1;
 
     const NAMES: [&'static str; Self::COUNT] = [
         "updates",
@@ -89,6 +90,7 @@ impl Counter {
         "text renders",
         "text shape hits",
         "text shapes",
+        "glyph placements",
     ];
 }
 

core/src/primitive/text.rs

@@ -7,7 +7,10 @@ use parley::{
     Alignment, AlignmentOptions, FontContext, FontFamily, FontFamilyName, GenericFamily, Layout,
     LayoutContext, LineHeight, PositionedLayoutItem, StyleProperty,
 };
-use std::hash::{DefaultHasher, Hash, Hasher};
+use std::{
+    collections::VecDeque,
+    hash::{DefaultHasher, Hash, Hasher},
+};
 use swash::{
     FontRef,
     scale::{Render, ScaleContext, Source, StrikeWith},
@@ -19,8 +22,31 @@ pub struct TextData {
     pub layout_ctx: LayoutContext<UiColor>,
     scale_ctx: ScaleContext,
     pub atlas: GlyphAtlas,
+    spare: VecDeque<Shaping>,
 }
 
+/// One shaping of some text, and the glyphs placed from it. A buffer holds
+/// the one it is drawn as; these are the ones it had before, kept because a
+/// container measures a child by drawing it in a box it may not keep, and so
+/// comes back to widths it has already asked for.
+struct Shaping {
+    /// A shaping is a function of these three and nothing else, so no widget
+    /// or buffer identity is involved and two texts of the same words share
+    /// an answer.
+    text: String,
+    key: LayoutKey,
+    layout: Layout<UiColor>,
+    placed: Option<RenderedText>,
+}
+
+/// How many to keep. Bounding the whole cache rather than each buffer is
+/// what makes this a fixed cost instead of one a tree of ten thousand texts
+/// pays ten thousand times; the re-asks come from laying out one subtree, so
+/// they are close together and few are needed. On `tests/revision_cost.rs`
+/// under `SWEEP=1`, the case that cannot hit across frames, 32 is not enough
+/// (6.6 ms) and 64 is (4.4 ms).
+const SPARE_SHAPINGS: usize = 128;
+
 impl Default for TextData {
     fn default() -> Self {
         Self {
@@ -28,6 +54,7 @@ impl Default for TextData {
             layout_ctx: LayoutContext::new(),
             scale_ctx: ScaleContext::new(),
             atlas: GlyphAtlas::default(),
+            spare: VecDeque::new(),
         }
     }
 }
@@ -83,6 +110,9 @@ pub struct TextBuffer {
     text: String,
     layout: Layout<UiColor>,
     layout_key: Option<LayoutKey>,
+    /// The glyphs placed from `layout`, so drawing this text again at the
+    /// width it already has places them once.
+    placed: Option<RenderedText>,
 }
 
 #[derive(PartialEq)]
@@ -97,6 +127,7 @@ impl TextBuffer {
             text: text.into(),
             layout: Layout::new(),
             layout_key: None,
+            placed: None,
         }
     }
 
@@ -121,15 +152,28 @@ impl TextBuffer {
         if text != self.text {
             self.text = text;
             self.layout_key = None;
+            self.placed = None;
         }
     }
 
     /// Invalidates the layout and returns the underlying string for editing.
     pub fn edit(&mut self) -> &mut String {
         self.layout_key = None;
+        self.placed = None;
         &mut self.text
     }
 
+    /// The glyphs of the shaping it is drawn as, once they are placed.
+    pub fn rendered(&self) -> Option<&RenderedText> {
+        self.placed.as_ref()
+    }
+
+    /// The width its shaping wraps at, and `None` where it does not wrap or
+    /// has not been shaped.
+    pub fn wrap_width(&self) -> Option<f32> {
+        self.layout_key.as_ref()?.max_width
+    }
+
     pub fn size(&self) -> Vec2 {
         Vec2::new(self.layout.width(), self.layout.height())
     }
@@ -144,6 +188,23 @@ impl TextBuffer {
             diag::bump(Counter::TextShapeHits);
             return;
         }
+        let kept = data.take_shaping(&self.text, &layout_key);
+        if let Some(key) = self.layout_key.take() {
+            data.keep_shaping(Shaping {
+                text: self.text.clone(),
+                key,
+                layout: std::mem::replace(&mut self.layout, Layout::new()),
+                placed: self.placed.take(),
+            });
+        }
+        if let Some(shaping) = kept {
+            #[cfg(feature = "layout-diagnostics")]
+            diag::bump(Counter::TextShapeHits);
+            self.layout = shaping.layout;
+            self.placed = shaping.placed;
+            self.layout_key = Some(shaping.key);
+            return;
+        }
         #[cfg(feature = "layout-diagnostics")]
         diag::bump(Counter::TextShapes);
         #[cfg(feature = "layout-diagnostics")]
@@ -273,26 +334,48 @@ pub struct RenderedText {
 }
 
 impl TextData {
-    pub fn render(
+    /// The shaping for this text at this width, taken out of what is kept.
+    fn take_shaping(&mut self, text: &str, key: &LayoutKey) -> Option<Shaping> {
+        // From the newest, since a re-ask is usually of something recent.
+        let at = self
+            .spare
+            .iter()
+            .rposition(|spare| spare.key == *key && spare.text == text)?;
+        self.spare.remove(at)
+    }
+
+    fn keep_shaping(&mut self, shaping: Shaping) {
+        if self.spare.len() >= SPARE_SHAPINGS {
+            self.spare.pop_front();
+        }
+        self.spare.push_back(shaping);
+    }
+
+    pub fn render<'b>(
         &mut self,
-        buffer: &mut TextBuffer,
+        buffer: &'b mut TextBuffer,
         attrs: &TextAttrs,
         width: Option<f32>,
-    ) -> RenderedText {
+    ) -> &'b RenderedText {
         #[cfg(feature = "layout-diagnostics")]
         diag::bump(Counter::TextRenders);
         #[cfg(feature = "layout-diagnostics")]
         let _render = diag::timer(TimerKind::TextRender);
         buffer.shape(self, attrs, width);
-        let glyphs = {
-            #[cfg(feature = "layout-diagnostics")]
-            let _place = diag::timer(TimerKind::GlyphPlacement);
-            self.place(buffer)
+        let placed = match buffer.placed.take() {
+            Some(placed) => placed,
+            None => {
+                #[cfg(feature = "layout-diagnostics")]
+                diag::bump(Counter::GlyphPlacements);
+                #[cfg(feature = "layout-diagnostics")]
+                let _place = diag::timer(TimerKind::GlyphPlacement);
+                RenderedText {
+                    glyphs: self.place(buffer),
+                    size: buffer.size(),
+                    color: attrs.color,
+                }
+            }
         };
-        RenderedText {
-            glyphs,
-            size: buffer.size(),
-            color: attrs.color,
-        }
+        buffer.placed.insert(placed)
     }
 }

core/src/ui/painter.rs

@@ -239,12 +239,12 @@ impl<'a> Painter<'a> {
         }
     }
 
-    pub fn render_text(
+    pub fn render_text<'b>(
         &mut self,
-        buffer: &mut TextBuffer,
+        buffer: &'b mut TextBuffer,
         attrs: &TextAttrs,
         width: Option<f32>,
-    ) -> RenderedText {
+    ) -> &'b RenderedText {
         #[cfg(feature = "layout-diagnostics")]
         diag::render_text(self.id, self.rsc.widgets().label(self.id), width);
         let ui = self.rsc.ui_mut();

src/widget/text/edit.rs

@@ -130,7 +130,6 @@ impl<'a> TextEditCtx<'a> {
     pub fn set(&mut self, text: &str) {
         let text = self.string(text);
         self.text.view.buf.set_text(text);
-        self.text.view.buf.changed = true;
         self.text.selection = None;
     }
 
@@ -177,7 +176,6 @@ impl<'a> TextEditCtx<'a> {
         };
         let at = at.min(self.text.view.buf.text().len());
         self.text.view.buf.edit().insert_str(at, text);
-        self.text.view.buf.changed = true;
         self.set_caret(at + text.len());
     }
 
@@ -190,7 +188,6 @@ impl<'a> TextEditCtx<'a> {
         }
         let range = sel.text_range();
         self.text.view.buf.edit().replace_range(range.clone(), "");
-        self.text.view.buf.changed = true;
         self.set_caret(range.start);
         true
     }
@@ -268,7 +265,6 @@ impl<'a> TextEditCtx<'a> {
 
     fn delete_range(&mut self, start: usize, end: usize) {
         self.text.view.buf.edit().replace_range(start..end, "");
-        self.text.view.buf.changed = true;
         self.set_caret(start);
     }
 

src/widget/text/mod.rs

@@ -14,11 +14,8 @@ pub struct Text {
 }
 
 pub struct TextView {
-    pub attrs: MutDetect<TextAttrs>,
-    pub buf: MutDetect<TextBuffer>,
-    // cache
-    tex: Option<RenderedText>,
-    width: Option<f32>,
+    pub attrs: TextAttrs,
+    pub buf: TextBuffer,
     pub hint: Option<StrongWidget>,
 }
 
@@ -28,19 +25,13 @@ impl TextView {
     }
 
     pub fn wrap_width(&self) -> Option<f32> {
-        self.width
+        self.buf.wrap_width()
     }
 }
 
 impl TextView {
     pub fn new(buf: TextBuffer, attrs: TextAttrs, hint: Option<StrongWidget>) -> Self {
-        Self {
-            attrs: attrs.into(),
-            buf: buf.into(),
-            tex: None,
-            width: None,
-            hint,
-        }
+        Self { attrs, buf, hint }
     }
 
     /// region where the text should be draw
@@ -52,22 +43,20 @@ impl TextView {
             .align(self.align)
     }
 
+    /// The text shaped for the width it is drawn in. The buffer keeps its
+    /// answers under the attrs too, so changing those asks a new question
+    /// rather than invalidating anything.
     fn render(&mut self, painter: &mut Painter) -> &RenderedText {
         let width = if self.attrs.wrap {
             Some(painter.px_len(Axis::X))
         } else {
             None
         };
-        if width != self.width || self.tex.is_none() || self.attrs.changed || self.buf.changed {
-            self.width = width;
-            self.tex = Some(painter.render_text(&mut self.buf, &self.attrs, width));
-            self.attrs.changed = false;
-            self.buf.changed = false;
-        }
-        self.tex.as_ref().unwrap()
+        painter.render_text(&mut self.buf, &self.attrs, width)
     }
+
     pub fn tex(&self) -> Option<&RenderedText> {
-        self.tex.as_ref()
+        self.buf.rendered()
     }
     /// Draws the text, and says where the glyphs went and what they use.
     pub fn draw(&mut self, painter: &mut Painter) -> (UiRegion, Size) {

tests/layout_diagnostics.rs

@@ -109,10 +109,15 @@ fn warm(seed: u64, depth: usize) -> (Harness, Tree) {
 fn report(label: &str, mut elapsed: Vec<f64>, _harness: &Harness) {
     elapsed.sort_by(|a, b| a.partial_cmp(b).unwrap());
     let frames = elapsed.len();
+    // The worst frame is the stutter somebody sees, so it goes beside the
+    // median; p99 says whether it is the load or a single interruption.
     println!(
-        "{label}: {frames} frame(s), min {:.3} ms, median {:.3} ms, total {:.1} ms",
+        "{label}: {frames} frame(s), min {:.3} ms, median {:.3} ms, p99 {:.3} ms, \
+         max {:.3} ms, total {:.1} ms",
         elapsed[0],
         elapsed[frames / 2],
+        elapsed[frames * 99 / 100],
+        elapsed[frames - 1],
         elapsed.iter().sum::<f64>(),
     );
     #[cfg(feature = "layout-diagnostics")]

tests/revision_cost.rs

@@ -0,0 +1,203 @@
+//! What a resize frame costs and what it holds, on a tree the revision before
+//! #16 also builds.
+//!
+//! Deliberately written in the API subset `43ce8c7` and this branch share, so
+//! the same source can be dropped into an old worktree and measured there:
+//! that is the only like-for-like comparison with the code the retained
+//! layout replaced. The random tree cannot carry one, because the generator
+//! itself changed with the work.
+//!
+//!     ROWS=40 FRAMES=500 cargo test --release --test revision_cost \
+//!       -- --ignored --nocapture resize_cost
+//!     ROWS=2000 cargo test --release --test revision_cost \
+//!       -- --ignored --nocapture text_memory
+//!
+//! Wall time on this machine varies with CPU frequency; take the number from
+//! `perf stat -e instructions:u` on the test binary directly.
+
+use iris::harness::Harness;
+use iris::prelude::*;
+use std::time::Instant;
+
+/// xorshift64, so one seed is one set of paragraphs on any machine.
+struct Rng(u64);
+
+impl Rng {
+    fn bits(&mut self) -> u64 {
+        self.0 ^= self.0 << 13;
+        self.0 ^= self.0 >> 7;
+        self.0 ^= self.0 << 17;
+        self.0
+    }
+
+    fn below(&mut self, n: usize) -> usize {
+        (self.bits() % n as u64) as usize
+    }
+}
+
+const WORDS: [&str; 24] = [
+    "wrapping",
+    "shapes",
+    "one",
+    "source",
+    "into",
+    "as",
+    "many",
+    "lines",
+    "as",
+    "the",
+    "box",
+    "leaves",
+    "room",
+    "for",
+    "paragraph",
+    "height",
+    "answer",
+    "setting",
+    "container",
+    "width",
+    "before",
+    "knows",
+    "measured",
+    "again",
+];
+
+/// A run of its own words, so nothing here is fast for two texts being the
+/// same string.
+fn words(rng: &mut Rng, least: usize, most: usize) -> String {
+    let words = least + rng.below(most - least);
+    let mut out = String::new();
+    for _ in 0..words {
+        if !out.is_empty() {
+            out.push(' ');
+        }
+        out.push_str(WORDS[rng.below(WORDS.len())]);
+    }
+    out
+}
+
+const OUTPUT: (f32, f32) = (900.0, 1200.0);
+
+fn env<T: std::str::FromStr>(name: &str, fallback: T) -> T {
+    std::env::var(name)
+        .ok()
+        .and_then(|value| value.parse().ok())
+        .unwrap_or(fallback)
+}
+
+/// A row of a fixed-width rect beside a column of one wrapping and one
+/// overflowing text: the shape that makes a container measure a child in a
+/// box it will not keep.
+fn build(h: &mut Harness, rows: usize) -> Vec<WidgetId> {
+    let mut rng = Rng(1);
+    let mut paragraphs = Vec::new();
+    let mut col = Span::empty(Dir::DOWN);
+    for _ in 0..rows {
+        let mut row = Span::empty(Dir::RIGHT);
+        row.push(
+            rect(Color::RED)
+                .width(Len::abs(40.0))
+                .add_strong(&mut h.rsc),
+        );
+        let mut body = Span::empty(Dir::DOWN);
+        let para = wtext(words(&mut rng, 12, 52))
+            .size(16)
+            .wrap(true)
+            .add_strong(&mut h.rsc);
+        paragraphs.push(para.id());
+        body.push(para);
+        body.push(
+            // Short, or its unwrapped width decides the row and the
+            // paragraph beside it never wraps.
+            wtext(words(&mut rng, 2, 6))
+                .size(16)
+                .wrap(false)
+                .add_strong(&mut h.rsc),
+        );
+        row.push(body.add_strong(&mut h.rsc));
+        col.push(row.add_strong(&mut h.rsc));
+    }
+    let root = col.add(&mut h.rsc);
+    h.set_root(root);
+    paragraphs
+}
+
+#[test]
+#[ignore = "measurement, not a check"]
+fn resize_cost() {
+    let rows = env("ROWS", 40_usize);
+    let frames = env("FRAMES", 500_usize);
+    let mut h = Harness::new(OUTPUT);
+    let paragraphs = build(&mut h, rows);
+    // What it cost is only half the comparison: the old code is cheaper
+    // partly because it wraps at the container's whole width rather than the
+    // part left beside the rect, and draws past the edge of the output.
+    println!("output width {}", OUTPUT.0);
+    for (at, id) in paragraphs.iter().enumerate().take(3) {
+        println!("paragraph {at}: {:?}", h.region(id));
+    }
+
+    // Two widths in turn is the friendly case for anything that remembers an
+    // answer, so `SWEEP=1` never repeats one -- a drag rather than a toggle.
+    let sweep = env("SWEEP", 0_usize) != 0;
+    let mut elapsed = Vec::with_capacity(frames);
+    for frame in 0..frames {
+        let narrower = match sweep {
+            true => (frame % 256) as f32,
+            false => ((frame + 1) % 2) as f32 * 8.0,
+        };
+        h.resize((OUTPUT.0 - narrower, OUTPUT.1));
+        let start = Instant::now();
+        h.frame();
+        elapsed.push(start.elapsed().as_secs_f64() * 1000.0);
+    }
+    elapsed.sort_by(|a, b| a.partial_cmp(b).unwrap());
+    println!(
+        "resize: {frames} frames, min {:.3} ms, median {:.3} ms, p99 {:.3} ms, \
+         max {:.3} ms, total {:.1} ms",
+        elapsed[0],
+        elapsed[frames / 2],
+        elapsed[frames * 99 / 100],
+        elapsed[frames - 1],
+        elapsed.iter().sum::<f64>()
+    );
+}
+
+fn kb(field: &str) -> u64 {
+    std::fs::read_to_string("/proc/self/status")
+        .unwrap()
+        .lines()
+        .find(|line| line.starts_with(field))
+        .and_then(|line| line.split_whitespace().nth(1)?.parse().ok())
+        .unwrap()
+}
+
+fn report(label: &str) {
+    println!(
+        "{label:24} rss {:>7} kB   peak {:>7} kB",
+        kb("VmRSS:"),
+        kb("VmHWM:")
+    );
+}
+
+/// Run this one on its own: the figures are the whole process's.
+#[test]
+#[ignore = "measurement, not a check"]
+fn text_memory() {
+    let rows = env("ROWS", 2000_usize);
+    report("before");
+    let mut h = Harness::new(OUTPUT);
+    let paragraphs = build(&mut h, rows);
+    report("after cold frame");
+    for frame in 0..40 {
+        h.resize((OUTPUT.0 - ((frame + 1) % 2) as f32 * 8.0, OUTPUT.1));
+        h.frame();
+    }
+    report("after 40 resizes");
+    // Settled: the output holds still and one leaf repaints per frame.
+    for _ in 0..10 {
+        let _ = h.rsc.widgets_mut().get_dyn_mut(paragraphs[0]);
+        h.frame();
+    }
+    report("after settling");
+}