`frame` named a length, not a rectangle, which was the one word in the
layout vocabulary that lied about its own shape. It is `rel_base`: what a
fraction a widget declares or reports is a fraction of.
Three API changes with it, all for containers that do one simple thing:
- `widget_within(id, region)` returns, taking a box in the widget's own
coordinates and deriving the child's rel base from it. `Offset` and `Pad`
are one call each again. `Offset` also stops reading `region_len`, which
pinned its drawing to a box length it does not care about.
- `place_at` takes the rel base, returns the answer, and asks the child
where there is no answer to re-express. Which of the two happens is the
painter's to work out, so `Span`'s second pass is one call and its
`drawn_across` bookkeeping is gone.
- `Part::All` is a `Part::WHOLE` constant rather than a variant, since it
was exactly `Of(UiSpan::FULL)` and bought a separate arm in two matches.
Measured at 0.07% of instructions retired against 0.04% run-to-run noise.
Cold layout is byte-identical to `84dad21` over 400 depth-5 trees.
The split box was named `region` and `placement` on 2026-09-17; `frame`
came back as a length and survived, `extent` did not. It stayed as the
name for both halves, distinguished only by prose: `draw_at` bound the
caller's `part` to a parameter called `extent`, and `ActiveData` held two
`UiRegion`s that `draw_at` wrote `part: extent` from.
The box a parent asks a widget in is now the region, and where its
drawing ends up is its placement. `Painter`'s four holds accumulators
become the one `LayoutHolds` they were assembled into, which also drops
the name mapping between them.
The cold dump of 400 depth-5 trees is byte-identical across the change.
A widget draws in the box it is asked in and its answer is placed inside
that box by re-expressing the drawing; nothing is drawn again in a box an
answer chose. The offer machinery, whose job was to tell a measuring draw
from a placing one, goes with the placing draw. A span measures each child
from its cursor and moves fixed children to their slots with place_at; a
share child is asked once more in its decided slot with its frame narrowed
to it. A stack asks non-sizing children in the box its sizing child
decided. A scroll asks its content once and moves it to the scrolled
offset. A local redraw asks the retained question again and puts the
answer back where the parent placed it.
A symbolic length a child pinned composes through Part::Of exactly where
the part is the whole box less pixels, and pins the parent's own length
otherwise; dropping it let a pad reuse a drawing across a narrowed frame
of the same pixel length (shrinker seeds 60, 248 and 384 at depth 5).
Suite 114/114 including the two decided-box pins, fast oracle 11/11,
shrinker 400 seeds at depth 5 over all fifteen cases.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The review pass over the two commits before it:
- `LayoutHolds`'s comment said a pin does not compose into the parent. It
does, where the box it pinned is the parent's own box, which is the one
case where the parent's own length is what was pinned.
- `DrawInfo::offer` was stored beside the two fields it is computed from.
It is a method now, with the open question written where it is asked
rather than only in the handoff.
- `Painter::own` is `frame_own` beside `extent_own`, since a widget's own
box is the extent and the frame is what it is a part of.
- One expression for the length a rule gives a frame (`narrowed_by`) and
one for what a widget answered (`ActiveData::measured`), each of which
had two.
- The counter said "the placement it was pinned to" for what is now a
length; the deferral in `redraw` named the seeds that made it necessary
before the last commit rather than the ones that do now; three comments
claimed an inset that does not exist yet.
No behaviour change: 108 suite tests, 20 core, the 11 generated cases, the
same two shrinker seeds failing at 400/5, and `view` and `tabs` byte-identical
to the renders taken before it.
`Pad` and `Stack` read `Painter::placement` to put their children inside
their own drawing, and reading it is what says the drawing holds for that
placement alone. So a pad or a stack anywhere in a row was drawn again --
with its whole subtree -- the moment an earlier sibling changed length,
however little else had moved.
`widget_within` now takes a `DrawRegion`, and `DrawRegion::Extent(part)`
gives the child a part of the extent without reading it. What is retained
is the part rather than the box it resolved to, so moving the extent
re-places the child through the same rule instead of redrawing the parent:
`inherited_children` becomes `extent_children`, carrying `Inherit` for the
wrapper case `Painter::widget` already had and `Within(part)` for the new
one.
The dependency that goes up is a range on the container's extent rather
than on its frame, since only the part's *length* reaches the child and
where the part sits is re-placed. A declared length is unchanged: it is a
length of the frame wherever the box it sits in came from. What still pins
the placement is a report with a fraction in it -- the same fraction of a
different extent is a different length -- and that pin is on the answer,
which `extent_frames_keep_fractional_reports_and_numeric_dependencies_valid`
fails without.
Three tests from the first attempt at this come with it, and the
diagnostics rig now says which of the three contracts refused a reuse,
which is what found the above.
Measured, seed 1 at depth 8, median frame: `many` 0.667 -> 0.613 ms and
`resize` 48 -> 32 us; seed 13's `many` 6.35 -> 5.15 ms. Green: fmt, clippy,
109 suite and 20 core tests, the oracle at 100 seeds, the shrinker at 400
trees of depth 5, 1000 seeds at depth 6, and 2000 seeds at depth 4 over all
fifteen cases. The five reference renders are byte-identical to `0e107f0`
on Venus, as are `tabs` resized to 900x1200 and `random` to 1280x800
against cold renders there.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`UiScalar` was `Len` without the `leftover` weight, which is the separation
canonical `main` already had as `Len` beside `LayoutLen` and this branch
collapsed. It is needed back for the queued clamp: a cap may not contain a
share, because a cap has to read the report a rule otherwise makes moot, and
a share puts the container's division into the same equation -- two
self-consistent assignments, which is the multiple-fixed-point failure
generated seed 13 punished for orthogonal sizing. `min(report, cap)` is not
a `LayoutLen` either: it is a sum of parts, and the smaller of two of them
is not one.
So `UiScalar` is `Len`, what was `Len` is `LayoutLen`, and the two say in
their docs which is which: a `Len` is pixels plus a fraction of a box -- a
position being the length from the box's start, which is why a span is two
of them -- and a `LayoutLen` is a `Len` plus a claim only a container
dividing its room can answer. `From<Len> for LayoutLen` is the one-way step
between them.
Names only; the shader's `UiScalar` is renamed with them. Checked: fmt,
clippy, 105 tests, and `tabs`, `minimal`, `view`, `text` and `random`
byte-identical at 1920x1200.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Fixed point cost 3x in layout: `many` went from 0.179 ms a frame to 0.544,
and `scroll` from 0.011 to 0.030. The counters said why -- eight more "placed
by redrawing" a frame -- and the reason was mine rather than the grid's. A
retained drawing belongs to the layer it was made on, which `4e28f10` started
enforcing, and `Stack` measures the child that sizes it by drawing it on its
own layer and then draws it again on the child layer. So every stacked child
redrew twice a frame, forever.
`Painter::child_layer_at` addresses a child's layer rather than walking to
it, and `Stack` measures on the layer that child ends up on. The second ask
is then a reuse. Its glyphs are written once rather than once under the
background and once over it.
Measured on the same fixture: `scroll` 0.031 ms to 0.020, `many` 0.570 to
0.283, and the scroll phase's counters are back to what they were before
fixed point -- 4 widget draws and 12 draw requests a frame, exactly. What is
left above that baseline is not this.
`ReuseOutcome` could not say "another layer" or "the region-node choice
changed"; both returned without a counter, which is why the first look at
this said nothing. They have counters now.
Checked: fmt, clippy, 105 tests, five shrinker cases at 300 seeds, 100
generated seeds, and the examples byte-identical but for 36 pixels of
`random` at one level -- edges that were being drawn twice.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`Px` and `PxVec2` reach the last places a pixel was a float: the window, the
box a widget reads, the box it is compared against, and `PixelRegion`. A
pointer, a wheel notch and a shaped glyph advance still arrive as floats,
and each is put on the grid where it arrives.
`Holds` is an interval of `Px`. `HOLDS_EPSILON_PX` is gone with the
`exact`/tolerant split it existed for: `at` is the length a widget read, an
open end is the next step along, and `same_px` is equality. `Span`'s margin
from `5ed9e87` goes too -- the box a parent hands back and the sum of what
its children asked for are counts of the same step, so the boundary decides
the same way from either side.
Three things had to be true for that, and were not:
`Holds::through` inverts `px + rel * box`, which rounds -- so a part of a
given length came from a range of boxes, and inverting the length alone gave
a point that need not contain the box the part was drawn in. It now maps the
half step either side, and one more for a length composed down the chain
against the same length measured against the window.
`RegionRemap` translates when a box only moved, rather than dividing to find
each part's fraction and multiplying to place it again. Two roundings landed
a step from where growing the tree that way does; a move is exact on a grid,
which is the whole reason `tests/drift.rs` was written.
A pixel is `1/1024` rather than `1/64`. At `1/64` the residue of a length
reached two ways was one step, and one step was 0.016 px -- enough to move
a box. `PX_SHIFT` and `REL_SHIFT` are the only statement of the grid now,
and the shader's copy is prepended from them rather than written twice.
Checked: fmt, clippy, 102 tests, 100 generated seeds in 75 s, all five
shrinker cases at 300 seeds, and `tabs`, `view`, `minimal`, `text` and
`random` byte-identical at 1920x1200.
What the fuzzers ask for is now a step, not a twentieth of a pixel: the
shrinker's five cases agree within one (`resize` exactly), and the oracle's
two-operation cases within two. The residue is a single rounding either way
-- it scales with the grid rather than accumulating, which is why it is a
thousandth of a pixel now. Closing it means one way of asking how long a box
is, rather than a chain composed down and a length measured against the
window; that is a bigger change than this one.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`UiScalar` is `Rel` beside `Px` rather than two floats, so composing a
position down a chain of boxes adds exactly and rounds only at the two
multiplies `within` makes. `UiSpan`, `UiRegion` and `UiVec2` follow it, the
hand-written `Hash` goes away with the bits it hashed, and `impl_op!` grows a
`same` form for a type whose fields are not the same kind of number.
`Len` is still floats, so the seam converts: `Px::from_f32` where a span adds
a child's length to its cursor, and `to_f32` where something outside layout
wants pixels. Those go when `Len` follows.
The GPU reads what the CPU wrote: the instance attributes are `Sint32x2` and
the shader decodes by `1/64` and `1/2^24`, both exact in `f32`, then composes
the move chain in floats as before. It has to agree with itself frame to
frame rather than with the CPU to the last bit.
Two things fell out of making the numbers exact.
`floor` at the rasteriser was picking the pixel below wherever a fraction
divided a window exactly. A fifth of 1920 is 383.99998 through a rounded
`Rel` -- and was 384.0 through an `f32` that happened to round up -- so five
tabs each lost their last column. `snap_floor` takes a coordinate within half
a step of a boundary to be on it, which is the same rule as everywhere else
here: decide where values do not land.
A widget measured on one layer and drawn again on another kept the first
layer, because `try_reuse` compared everything about a retained drawing
except which list it sits in. `Stack` does exactly that for its background,
so every panel's text went under its own background. It only worked before
because the two asks differed by a rounding and forced a redraw;
`tests/retained.rs` pins it now, and `ReuseOutcome` can say `WrongLayer`.
Checked: fmt, clippy, 100 tests, 100 generated seeds in 70 s, all five
shrinker cases at 300 seeds. `tabs`, `view` and `minimal` render
byte-identical at 1920x1200; `random` differs in 36 pixels by one level;
`text` differs where glyph origins moved onto the grid -- same positions,
same spacing, different subpixel coverage, checked at 6x against the old
render.
Measured on the way: with the fuzzer comparing for *equality* rather than
within 0.05 px, `resize`, `repaint` and `size-change` already pass 100 seeds.
`reorder` fails one seed by exactly one step, which is the `Len` seam above.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Choosing which dirty widget to settle next asked every one of them how
deep it was, and answering meant walking its ancestry to the root. At
130 of 260 widgets dirty that was 25.8% of the frame -- more than laying
out or rendering.
A widget's depth is known where it is drawn: its parent's plus one. So
`Painter` carries it and `ActiveData` keeps it, and the choice reads a
field. Being reused counts as being visited, so the two reuse paths keep
it current too; only a subtree nothing looked at can hold an old one,
and nothing under an unvisited subtree is being ordered.
The order is unchanged, so nothing about the layout is: the five
reference renders and the resize render are byte-identical. What the
carried depth might get wrong is itself, so `depth` asserts it against
the ancestry in debug builds, and the hundred-seed sweep passes with
those assertions on -- including the reshuffles, which are what move a
widget to another parent.
Same load, 1000 frames, 130 of 260 dirty: 8.16M instructions per frame
to 7.14M, median 0.813 ms to 0.639, and the choosing from 25.8% of the
frame to 4.7%. What is left of it is iterating the dirty set itself,
which a `HashSet` walks by capacity rather than by length.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Only the line breaking depends on the width. The shaped runs under it --
the font selection, the unicode analysis, harfrust -- are a function of
the text and the attrs, and parley re-breaks them in place; its own
editor does exactly this on every resize. So a new width is a break and
a placement, not a shaping.
On the depth-8 tree that is 107 breaks at 0.119 ms where the shapings
they replace were 4.0 ms, and it holds however far the width moves,
which is what the store could not do: a width the layout has not seen
before is a miss, and a drag never sees one twice. Instructions per
frame over 500 resize frames of `tests/revision_cost.rs`, for widths
that alternate and widths that never repeat:
#18 head 124.2M 123.6M
a store of shapings 17.7M 45.9M
re-breaking alone 32.9M 32.8M
both 20.6M 24.2M
The store stays because re-breaking does not place the glyphs, so it now
holds those instead: fewer instructions than either alone in the case
that never repeats, and 3 MB rather than 4 MB on a tree of 4,000 texts,
against the 132 MB the code before #16 reaches after the same resizes.
The worst frame is 2.54 ms where that code's is 6.47 ms, and the two
gestures are within a millisecond of each other rather than a factor of
two apart.
Count the breaks and time them separately from shaping, since which of
the two a frame is doing is the whole question here.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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>
iris-ai
