Measure a dirty widget where its parent asked, not in a box its answer decided

A local redraw drew a dirty widget in the box it was placed in. When a reader
decided that box from the widget's own answer -- an aligned span sized to its
children, a text at the tail of a row, a scroll's content -- the old answer is a
fixed point of measuring there whatever the content now says, so the layout had
two stable answers and which one it reached depended on the tree's history.
`tests/unsettled.rs` has the two shrunk cases: the four-widget aligned span,
and a scroll placing a pass-through `SetSize` in a box the content decided,
where the span under it was placed once and nothing at its own edge said so.

`ActiveData::offered_px` keeps the pixel size of the box the parent first asked
about the child in, whether through `known_len` or a first `place`, beside `px`,
the box it drew against. A dirty widget whose size reads an axis on which some
reader up its chain gave what it read a box other than the one it asked in is
not drawn locally: the chain is marked and the parent of the highest such
placement draws, since above it every box is a constraint rather than an
answer. The walk goes up the whole reader chain because a pass-through hands a
derived box down unchanged.

`Scroll` read its box's length for the clamp through `px_len`, which records
the reported size as depending on it, and it does not: its size is its
content's. That made every scroll tick a size question asked in a derived box,
at 34x the instructions. `Painter::px_len_for_draw` is the read that records
nothing. Instructions per frame on the depth-8 rig against the previous head:
`many` at 32 dirty 0.66M to 0.74M, at 130 dirty 27.7M to 26.5M, `resize` 15.8M
to 15.0M, `scroll`, `repaint` and `size` unchanged. The shrinking fuzzer passes
200 trees at depth 7 in all four cases, the hundred-seed sweep passes, and the
five reference renders and the resize render are byte-identical.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

tests/generated.rs

@@ -442,25 +442,11 @@ fn adding_and_removing_span_children_lands_where_growing_it_that_way_would() {
     }
 }
 
-/// Reproduces a divergence that predates the position chain: laying a tree out
-/// again does not always land where growing it cold does.
-///
-/// Every one seen so far is a wrapping text on a span's *own* axis, where the
-/// two draws do not agree. The span measures the child in the whole box, the
-/// child shapes to that width and reports the width it used, the span then
-/// places it in exactly that width -- which is a length change, so the child
-/// shapes again, and its longest line is shorter than the box it was just
-/// given. Each pass narrows it, so where the tree ends up depends on how many
-/// passes it has had, and a warm tree has had a different number from a cold
-/// one. Layout is supposed to be a function of the state alone.
-///
-/// A span whose axis is not the wrap axis is stable, which is every real
-/// column of text, and why nothing else has run into this.
-///
-/// 7 of these 90 diverge on `db1751f`, before the chain; 30 do with it, since
-/// a placed child reaches the second shaping more often. Both numbers are the
-/// same defect, and it wants fixing where the two draws meet -- LAYOUT.md §4 --
-/// rather than anywhere in the chain.
+/// The same property over a hundred seeds and every scenario. What it has
+/// found so far was never where the trees disagreed: a text measured in a box
+/// it was not going to get, and a widget re-measured in a box its own answer
+/// had decided. `tests/shrink.rs` is how a seed from here becomes a tree
+/// small enough to read.
 #[test]
 #[ignore = "a hundred seeds, rather than the seven the others check"]
 fn a_long_run_of_seeds_agrees() {

tests/unsettled.rs

@@ -1,10 +1,9 @@
-//! The smallest tree that lays out differently on a second frame, shrunk from
-//! a 402-widget one `tests/shrink.rs` grew. Both of these fail: a cold frame
-//! leaves a wrapping text shaped at a width it was measured in rather than the
-//! one it was given, and a repaint is what puts it right. So the warm-against-
-//! cold oracle in `generated.rs` has been comparing against a tree that had
-//! not settled, and some of what it called a warm defect is the cold side
-//! being wrong.
+//! The smallest trees that laid out differently warm than cold, each shrunk
+//! by `tests/shrink.rs` from hundreds of widgets. The first two are a cold
+//! frame that had not settled: a wrapping text shaped at a width it was
+//! measured in rather than the one it was given. The rest are a widget
+//! measured again in a box its own answer had decided, where the old answer
+//! is a fixed point whatever the content now says.
 
 use iris::harness::Harness;
 use iris::prelude::*;
@@ -213,3 +212,87 @@ fn swapping_two_children_lands_where_growing_them_that_way_does() {
     }
     assert!(wrong.is_empty(), "{}", wrong.join("\n"));
 }
+
+/// Eight widgets, shrunk from 80. The scroll decides how wide to make its
+/// content from what the content says, and hands that box down through a
+/// pass-through; the span under it was placed once, in that box, so nothing
+/// at its own edge says the box was its own answer.
+fn plant_scrolled(h: &mut Harness, swapped: bool) -> (Vec<WidgetId>, [WeakWidget<Span>; 2]) {
+    let words = "Wrapping shapes one source into as many lines as the box leaves room for,";
+    let text = wtext(words).size(16).wrap(true).add(&mut h.rsc);
+    let filler = rect(Color::RED).add(&mut h.rsc);
+    let mut inner_children: Vec<StrongWidget> =
+        vec![text.add_strong(&mut h.rsc), filler.add_strong(&mut h.rsc)];
+    if swapped {
+        inner_children.rotate_left(1);
+    }
+    let inner = Span {
+        children: inner_children,
+        dir: Dir::RIGHT,
+        gap: 0.0,
+    }
+    .add(&mut h.rsc);
+    let block = rect(Color::RED).add(&mut h.rsc);
+    let fixed = SetSize {
+        inner: block.add_strong(&mut h.rsc),
+        x: Some(Len::px(87.0)),
+        y: None,
+    }
+    .add(&mut h.rsc);
+    let mut outer_children: Vec<StrongWidget> =
+        vec![fixed.add_strong(&mut h.rsc), inner.add_strong(&mut h.rsc)];
+    if swapped {
+        outer_children.rotate_left(1);
+    }
+    let outer = Span {
+        children: outer_children,
+        dir: Dir::RIGHT,
+        gap: 0.0,
+    }
+    .add(&mut h.rsc);
+    let through = SetSize {
+        inner: outer.add_strong(&mut h.rsc),
+        x: None,
+        y: None,
+    }
+    .add(&mut h.rsc);
+    let scroll = Scroll::new(through.add_strong(&mut h.rsc), Axis::X).add(&mut h.rsc);
+    h.state.root = Some(scroll.add_strong(&mut h.rsc));
+    (
+        vec![
+            text.id(),
+            filler.id(),
+            inner.id(),
+            block.id(),
+            fixed.id(),
+            outer.id(),
+            through.id(),
+            scroll.id(),
+        ],
+        [inner, outer],
+    )
+}
+
+#[test]
+fn a_span_placed_once_in_a_box_its_answer_decided() {
+    let mut warm = Harness::new((640, 900));
+    let (ids, spans) = plant_scrolled(&mut warm, false);
+    warm.frame();
+    for span in spans {
+        warm.rsc[span].children.rotate_left(1);
+    }
+    warm.frame();
+
+    let mut cold = Harness::new((640, 900));
+    let (cold_ids, _) = plant_scrolled(&mut cold, true);
+    cold.frame();
+
+    let mut wrong = Vec::new();
+    for (i, (&w, &c)) in ids.iter().zip(&cold_ids).enumerate() {
+        let (got, want) = (warm.region(&w), cold.region(&c));
+        if got != want {
+            wrong.push(format!("widget {i}: warm {got:?} cold {want:?}"));
+        }
+    }
+    assert!(wrong.is_empty(), "{}", wrong.join("\n"));
+}