[vjeux]

Depends on #9261 (on #8658, #8655, #8653 and #8676). Builds the swipe constructor that spec:AST-057 records (DEC-8). This pull request's own commits: vjeux/astryx@0aa861f...e870248

Status: draft. CI is running at this head. Already passed locally at this head: unit tests, lint, typecheck, check:repo, check:knowledge, the CLI and codemod suites (but for two release-cache cases that fail the same way at the base), Storybook's build, the docsite's preview build, the accessibility contracts (3 repeats), and the interaction proofs in Chromium (pinned to two cores at their default of one worker, every case; at 4 workers, every case but the animator's busy-main-thread proof, which starves there as at #9261's head) and in WebKit (four CPUs: the swipe proofs 10 times, and the animator, snap-point drag and gesture proofs).

User need

A list row with swipe actions is a touch accelerator: a finger, a pen or a trackpad's two-finger swipe drags the row aside to uncover its actions.

• An app should pay for the swipe only where it uses it, and a row should be swipeable as soon as it paints. Today the swipe reaches every import that renders an Item row through Item itself, and the gesture goes live only once a lazily loaded chunk arrives.
• On a trackpad, the row should move with the fingers and, when they lift, settle at their speed. Today it trails the fingers by 10–14 px and comes to a stop before it settles.
• On a finger, the row should keep UIKit's feel: a dead zone, a tracking spring, and UIKit's release rule.

Sheets and pagers already run one shared drag path. A row that runs its own copy of that machinery drifts from them, so the row should run the same path and declare only what is its own: its stops and its actions.

Problem and solution fit

• Swipe actions come through a constructor (spec:AST-057 DEC-8): swipe={createSwipe({leading, trailing, swipeBehavior})}. The constructor's module, @astryxdesign/core/Swipe, carries every module the swipe runs on, so an app that never imports createSwipe ships none of it, and a row given swipe is swipeable at its first paint (FR10).
• The row ran its own pan, its own frame loop and its own springs, so the trackpad behavior that spec:AST-094 FR43 gives every followed pan (follow from the first event past 8 px, settle by position and velocity, keep the fingers' speed) never reached it.
• The row now runs the shared path: the trackpad splitter, the pan, the snap-point drag and the animator (spec:AST-094 DEC-24). It declares its stops (rest, the open width, and out past its edge for a full swipe) and its actions, and the drag draws its travel into the one custom property every part of the panel reads (spec:AST-057 FR2).
• The finger's UIKit model is the one behavior the shared path does not have, so the row supplies it through the shared drag's model hooks instead of running its own drag.

Current specification

• spec:AST-057:
  • DEC-8 and FR1: one swipe prop, whose value createSwipe makes; Swipe is opaque.
  • FR2: the travel as one custom property. FR3: swipeBehavior, with a development warning for several entries on a commit side.
  • FR4 and DEC-7: a finger, a pen or a trackpad swipes; a mouse drag never does.
  • FR5: UIKit's release rule; a full swipe by position only; an action fires once the row's content has left its box. FR7: an open row's closes. AR2.
  • FR10: every module the swipe runs on is reachable only from createSwipe's module. FR11: an inline createSwipe call is fine.
  • DEP-0017: the released swipeActions and swipeBehavior are deprecated and keep working.
• spec:AST-094 FR43 and DEC-24: a followed trackpad pan follows the fingers one for one from its first event past 8 px and settles by position and velocity from their last speed; within one gesture an open row only closes; one shared path, and a component declares only its detents and actions.
• spec:AST-017 FR28, FR29 and DEC-17: a deprecation in a patch adds no runtime warning and changes no caller's behavior.
• spec:AST-107 FR7: Core uses the public layer except for the known public-API gaps that architecture:react-component-runtime INV13 lists. This pull request names the row's jobs there.

Public delta and caller burden

• API (packages/core/api-report.md shows each line):
  • adds createSwipe(options: CreateSwipeOptions): Swipe, from @astryxdesign/core/Swipe and the package root, with CreateSwipeOptions (leading?, trailing?, swipeBehavior?), the opaque Swipe, SwipeAction, SwipeActionVariant and SwipeBehavior;
  • adds swipe?: Swipe on Item and ListItem;
  • deprecates swipeActions, swipeBehavior and the types ItemSwipeAction, ItemSwipeActions, ItemSwipeActionVariant and ItemSwipeBehavior (DEP-0017). They keep working as released, with static @deprecated metadata and no runtime warning, until 0.8.0 removes them (CLN-0025). When swipe is set too, swipe wins, and a development-only note says the deprecated props are ignored.
  • The theme targets (item-swipe-panel, item-swipe-action) are unchanged.
• Behavior:
  • a row given swipe is swipeable at its first paint; a row given the deprecated props loads the swipe the first time it has them, as released;
  • a trackpad's two-finger swipe moves a row with the fingers and lets it go at their speed;
  • a mouse drag and a mouse wheel's notches never swipe a row;
  • a fast flick from an open row settles open, since speed alone never fires an entry (spec:AST-057 FR5), and FR43's long fast swipe fires only where spec:AST-057 allows a full swipe.
• Callers: nothing to change in this release. astryx upgrade rewrites static swipeActions and swipeBehavior on Item and ListItem to swipe={createSwipe(…)}, with a codemod staged in transforms/next/. It's an opt-in early migration from this release, and 0.8.0's upgrade runs it again with the removal. It adds the import, renames the four types, and drops a merged swipeBehavior="reveal" (the default). It reports computed values, every row that spreads props, and code that mutates the leading or trailing array of a value it retypes as CreateSwipeOptions, since those arrays are readonly.

Design and alternatives

Where the swipe's code lives (spec:AST-057 DEC-8, FR10). packages/core/src/Swipe/ holds createSwipe, the panels (SwipeActionsLayer.tsx), the row's gesture code (useSwipeAction.ts, swipeRelease.ts) and, through them, the shared drag path and the animator.

• A row given swipe. Item reads the value through swipeHost.ts, a module with no imports, and hands the runtime the value carries its root and the facts it already has: its role, whether it is disabled, whether it is the enlarged target of a control inside it, and whether its root is a link. The runtime decides whether it serves the row (DEC-3). If it does, it returns the root's style, the root's handlers and the layer, and Item renders them. Item holds no swipe code for this path:
  • the root's keys reach the row's swipe through a registry keyed on the root element, inside the runtime; the handlers are the same functions for every row and every value, so a new value from an inline call changes nothing on the root (FR11);
  • while the row is open, the runtime holds its root's press with a private attribute (privateAttributes/pressHeld.stylex.ts). The press controller treats a held surface like a disabled one: it takes the press and paints nothing, a pressable inside it keeps its own press, and the touch still counts for the brake. So a tap on an open row's content only closes it (FR7);
  • the gesture host swallows the click the browser synthesizes after a swipe.
• Swipe is opaque: a frozen brand with nothing to read. Only createSwipe records what a value carries, so a value built by hand swipes nothing.
• DEP-0017's deprecated path is the only swipe code in Item, for the deprecated props only, and leaves with them (CLN-0025):
  • the released loader, which brings the layer with import() the first time a row has the deprecated props; that layer brings the released commit warning;
  • the released row style, which such a row takes from its first render (spec:AST-017 DEC-17), and the released role gate;
  • the bridge the loaded layer fills, and the forwarder of the root's key events into it;
  • the coexistence diagnostic.
  • The record's "click guard" has no separate piece: the gesture host swallows the synthesized click on both paths.
• The bundle guard names the loader as the one import() of swipe code and counts it out of Item's import; without that exception, the 13 imports that render an Item row would carry the gesture core. Swipe joins the imports that carry it.

Internal options the shared primitives gain (no public API; they land with this, their first user):

• createPan, under the public createPanGesture: which pointers and scrolls start a drag, its claim distance, a begin gate and the selection lock; each report says what drove the drag.
• createSnapDrag, under the public createDragToSnapPoints, which keeps its behavior and tests:
  • a model of where a drag draws the element and where a release lands;
  • a custom property to draw into;
  • settleTo with a spring, a velocity, a touch release and the animator's start: 'next-frame' (chore(core): the animator takes start: 'next-frame' for a call made in an input event's own task #9261).
  • A drag's spring targets run on the drag's own clock: a target that waited for the browser's start would restart from rest on every move in WebKit.
• A settle-passes-offset callback: settleTo({onPass}) calls back on the first animation frame whose presented value has passed the offset, read on the settle's own curve. It uses no timer, and never fires for a settle that lands short, is replaced, or is destroyed.
• The animator draws a channel into one registered custom property, with an optional companion: a second custom property drawn from that value by a pure mapping, in the same keyframes, so it cannot drift from the travel.
• The press controller reads a held surface (privateAttributes/pressHeld.stylex.ts) as a disabled one.

What the row's own code holds, and why. useSwipeAction.ts is 675 code lines (comments and blank lines excluded):

• before the callback and the companion moved into the shared drag, it was 778;
• those two moves took it to 669, adding 21 lines to the shared drag and 29 to the animator;
• passing the start option on the row's input-driven settles added 6.

The row builds no pan recognizer, runs no frame loop and solves no spring. What remains:

• its stops and actions: measuring the entries, and the side geometry the panel's CSS reads;
• the finger's UIKit model: the dead zone, the tracking spring, the never-past-rest clamp, the confirm point and run-ahead, and UIKit's release rule (swipeRelease.ts);
• the full swipe's action timing (spec:AST-057 FR5);
• the open row's closes (spec:AST-057 FR7);
• the panel itself (SwipeActionsLayer.tsx).

Known limitation, inherited from main's released design

A long main-thread task during a drag or settle freezes the row, then it jumps; the released swipe behaves the same. spec:AST-057's current-state impact records it in this pull request.

• spec:AST-057 FR2 draws the travel into one registered custom property that every part reads, and Chromium animates a custom property on the main thread. The drag's tracking spring draws into the same property. Main's released swipe has the same design.
• Measured with a 700 ms block during a settle, in Chromium pinned to two cores at one worker, 10 runs:
  • the row moved on none of the 38–40 frames in which a compositor-only control moved, and held one position for the whole 660 ms judged;
  • the control's longest stall was 18–55 ms;
  • in a traced run, the row then jumped from 191 px to 97 px in one frame.
• spec:AST-057 keeps FR2 as approved, so each part's transform does not move on the compositor (fix(Item): swipe actions move on the compositor with UIKit springs #8097's approach, superseded by this pull request).

Evidence

• The constructor:

  • createSwipe makes a frozen value with nothing to read, carries the entries per side and swipeBehavior (reveal by default), keeps the entries it was given, and warns in development for several entries on a commit side (createSwipe.test.ts);
  • a value built by hand swipes nothing (Item.test.tsx);
  • importing and calling it in a fresh Node process, with no window or document, does no DOM work (createSwipe.server.test.ts);
  • the runtime serves a role-less row and a listitem, and no option or menuitem* row, disabled row, delegating row, link root or row without entries; it gives every row and every value the same root handlers and layer type.

• The press hold (red first, pressFeedback.test.ts): a touch on a held surface paints nothing and lights nothing around it; a pressable inside it keeps its own press; a mouse press on it paints nothing; it presses again once the hold is gone. On the row, the hold is written from the release that opens it and freed when it closes, or when the row loses its swipe while open.

• An inline createSwipe call (spec:AST-057 FR11): a new value keeps a closed row closed and an open or settling row open, fires the newest handler, repaints a changed entry list, creates no second drag and starts no second settle.

• The deprecated props: a row still rests open and fires a tapped entry, and still slides out under commit; it has the same root style as a row given swipe from its first render, before its chunk loads; the released role gate serves a listitem and not an option; swipe wins, with a development note; no runtime warning on their own; the released commit warning stays. Their tests wait for the deprecated layer's own load, never a clock: a test helper awaits the module the loader brings, inside act, and one test's mock loads it 1.5 s late.

• The import graph: Item statically imports only swipeHost.ts and the released row style from the Swipe module, and reaches the layer only through DEP-0017's loader; createSwipe brings the layer, and the layer the gesture, statically (Item.source-build.test.mjs, and the touch-support bundle guard).

• Module attribution (spec:AST-057 FR10). Each public import is built with the size report's esbuild options (minified ESM, code splitting, metafile). Every byte of the JavaScript it loads eagerly, its entry output and the chunks it reaches by static import, is attributed to its source module from the metafile's bytes in output. A module's gzipped share is its part of the eager bytes times the eager output's gzip size. The swipe modules are the gesture core, the animator, the snap-point drag, the panels' rendering and the row's gesture code. The touch-support bundle guard runs this attribution for Item alone and ListItem alone and fails on any byte of a swipe module; Item with createSwipe is its positive control, and a build where Item uses createSwipe fails it. Base (feat(core): createDragToSnapPoints and useDragToSnapPoints settle an element on snap points #8658's head) against head:

  Import	Base (gzip)	Head (gzip)	Swipe modules, head (min / gzip share)	swipe handoff and row style, head (min / gzip share)	Item's own modules, base → head (min)
  Item alone	13,531 B	13,527 B	0 B / 0 B	134 B / 47 B	7,612 → 7,140 B
  ListItem alone	51,980 B	51,986 B	0 B / 0 B	134 B / 45 B	7,599 → 7,127 B
  Item with createSwipe	—	54,606 B	113,005 B / 37,111 B	227 B / 74 B	— → 7,140 B
  ./List	52,757 B	52,743 B	0 B / 0 B	134 B / 45 B	7,599 → 7,127 B
  ./CheckboxList	54,875 B	54,848 B	0 B / 0 B	134 B / 45 B	7,597 → 7,120 B
  ./RadioList	50,440 B	50,449 B	0 B / 0 B	134 B / 44 B	7,596 → 7,127 B
  ./DropdownMenu	74,275 B	74,386 B	0 B / 0 B	134 B / 45 B	7,586 → 7,109 B
  ./ContextMenu	75,372 B	75,433 B	0 B / 0 B	134 B / 45 B	7,639 → 7,146 B
  ./MoreMenu	72,474 B	72,597 B	0 B / 0 B	134 B / 45 B	7,590 → 7,113 B
  ./Selector	76,657 B	76,652 B	0 B / 0 B	134 B / 43 B	7,647 → 7,163 B
  ./Table	222,107 B	222,676 B	0 B / 0 B	134 B / 43 B	7,613 → 7,128 B
  ./Breadcrumbs	68,379 B	68,491 B	0 B / 0 B	134 B / 44 B	7,639 → 7,146 B
  ./Pagination	85,883 B	85,719 B	0 B / 0 B	134 B / 43 B	7,631 → 7,148 B
  ./PowerSearch	185,843 B	186,065 B	0 B / 0 B	134 B / 42 B	7,608 → 7,123 B
  ./Markdown	147,227 B	147,193 B	0 B / 0 B	134 B / 45 B	7,635 → 7,139 B

  • No eager output without createSwipe holds a byte of a swipe module. From the Swipe module, each import that renders a row keeps only the swipe handoff (swipeHost.js, 57 B) and the deprecated path's row style (swipeHost.stylex.js, 77 B). Item's own modules shrink by about 470 B.
  • The press controller reads the press hold from privateAttributes/pressHeld.stylex.js, 73 B in every import that renders a row (FR7).
  • The deprecated loader's import() target is its own chunk. Bundling moves the parts of a module that this chunk shares with the eager code into the eager output, so the row importers carry 57 B more of utils/touchQueries.js and 36 B more of the press hold, and Table and PowerSearch also carry about 1,260 B more of motion/uikit/springs.js and 79 B more of motion/uikit/timing.js. Built without that import(), those modules match the base byte for byte. None of them is a swipe module, and the bytes leave with the loader (CLN-0025).
  • ListItem alone is 345 B smaller minified and 6 B larger gzipped.
  • Item.tsx is 944 lines (base 973), and ListItem.tsx 474 (base 475).

• The codemod: an inline object merged with swipeBehavior; a default swipeBehavior="reveal" dropped; a name guarded, and spread beside a swipeBehavior expression; aliases and namespace imports followed, other libraries left alone; an existing import reused, and a free alias taken; the deprecated props removed from a row that sets swipe; computed values and spreads reported once; mutations of a retyped value's leading or trailing array reported once; the four types renamed; idempotent.

• Trackpad, WebKit:

  • the row moves one for one on every wheel;
  • the release keeps the fingers' speed with no stop;
  • it never passes rest in one gesture;
  • a long fast swipe fires the outermost action;
  • a fast flick from open settles open;
  • Reduce Motion lands at once on a fine pointer;
  • a click closes a row a trackpad opened and reaches nothing.
  • Chromium's synthetic wheels carry a fixed wheelDelta, which FR43 reads as a mouse wheel's, so the trackpad cases skip there.

• A mouse wheel's notch never swipes a row (Chromium).

• Fingers, by real touch in both engines:

  • the tracking spring and its catch-up, read on the curve the row played;
  • UIKit's settle, its keyframes and every frame drawn;
  • a flick hands the row's speed to the settle;
  • a catch;
  • the release rule;
  • the full swipe;
  • the Reduce Motion cross-fades;
  • FR7's closes.

• A close held back by a later listener's block starts after the block, and every frame drawn lies on its curve: the row's input-driven settles pass start: 'next-frame'.

• On a loaded page:

  • the frame-by-frame settle proof is judged against the frames the page drew during the settle (its animation-frame times): at least 4 to judge, a new travel on each, each on UIKit's curve. A count at 60 Hz (more than 10) failed on a loaded runner that drew 5. Red first: with the page's main thread busy 150 ms at a time from the lift, the page drew 8 frames; the count fails 3 of 3, and the judgment per frame passes 3 of 3;
  • the accessibility proof's fling stamps its touches with the script's times, as a digitizer stamps a finger's. The gesture reads the touches' own times, but each dispatch returns only once the page has handled the touch, so unstamped, a busy page stretched the dispatches. Red first: with each touchmove costing the page 60 ms, the last two moves and the lift went out 108 ms and 100 ms apart, the lift fell past the velocity filter's 40 ms stillness, and the row closed (3 of 3); stamped, the same run opens it (3 of 3).

• Unit:

  • the row runs the shared drag and declares its stops there (Item.sharedDrag.test.tsx);
  • the shared drag's internal options, each red first (createSnapDrag.test.ts);
  • a disabled entry stays focusable and inert on an open row (spec:AST-085 FR1, FR2).

Supersedes #8069, #8096, #8097, #8132 and #8134

They were closed in its favor. Each one's tests, changeset and doc changes were checked against this branch:

• chore(core): swipe actions' UIKit values and release rule #8069 (UIKit values and release rule):

  • swipeActions.ts, swipeRelease.ts, its MEASUREMENTS.md rows and its motion/README.md row are here.
  • Of its 11 test titles, 4 are here unchanged and 1 is renamed (the lab batch numbers became "UIKit lab, iOS 26.2 simulator, Oct 23 2026"). fix(Item): a swipe row's full swipe confirms, runs ahead and fires like UIKit's #8134 replaced the other 6: the commit point became the confirm point, and one release case was reworded. Those replacements are here.
  • Its changeset is here, reworded for that model.

• fix(Item): swipe actions on the gesture system #8096 (the gesture system):

  • 14 of its 18 test titles are here unchanged.
  • Its test that other recognizers stay inactive under a fine pointer is updated for spec:AST-094 FR43: gestures attach for every pointer.
  • Its two swipe-layer render tests are here: a row given swipe renders its layer at its first render, and a row given the deprecated props renders it once the loader has brought it.
  • Its pins that the layer imports with the row are replaced by pins that createSwipe brings the layer and Item imports none of it.
  • Its changeset is here, with the trackpad added.

• fix(Item): an open swipe row closes on a tap on its content or a scroll #8132 (an open row's closes): 11 of its 12 test titles are here unchanged; one is renamed for the focus rule ("inside its open panel"). Its changeset is here, reworded the same way.

• fix(Item): a swipe row's full swipe confirms, runs ahead and fires like UIKit's #8134 (the full swipe):

  • 19 of its 30 test titles are here unchanged.
  • 4 release-rule tests lost their "(FR5: …)" departure notes, and the describe that grouped them is gone, since spec:AST-057 FR5 now records the measured rule.
  • Its swipeCurves unit tests (one describe, four cases) covered per-part curves. Under spec:AST-057 FR2 every part reads one custom property, so those per-part curves no longer exist; the browser proofs check every part against the travel on every frame.
  • Dropped on purpose: its proof that the slide-out advances through a main-thread block, for the reason under the known limitation above.
  • Its changeset is here unchanged.

• fix(Item): swipe actions move on the compositor with UIKit springs #8097 (swipe actions on the compositor with UIKit springs):

  • 20 of its 31 test titles are here unchanged; 3 are renamed for spec:AST-057 FR2's single travel property ("moves only by transforms that read one travel", "gives the root back its own styles when the swipe actions go", "fires on the next frame when the release is already past the row's edge").
  • Its slide-out fire cases (a slide-out the browser starts late; a pending fire dropped when the row leaves) are covered by the shared drag's settle-passes-offset tests in createSnapDrag.test.ts: it calls back once, on the first frame whose presented value has passed the offset, and never for a settle that lands short, is replaced, or is destroyed.
  • Its per-part tests (swipeParts, one describe and three cases, and the one-clock case) covered per-part transforms, which spec:AST-057 FR2 replaces with one custom property; the browser proofs check every part against the travel on every frame.
  • Dropped on purpose: its compositor change and its proof that the settle advances through a main-thread block (see the known limitation).
  • Its other behavior (UIKit's springs, a flick that opens or closes, Reduce Motion, a catch) is here, with its changeset line minus the compositor sentence.

Scope

• One capability is added.
• Unrelated fixes, cleanup, and policy changes were removed or split.
• Consumer docs, focused tests, and migration evidence ship with the feature.
• Public text and artifacts contain no internal Meta context.

Testing

• pnpm exec vitest run packages/core, plus the CLI, registry, codemod and touch-bundle guards.
• pnpm check:repo and pnpm check:knowledge. packages/core/api-report.md changes by exactly the lines pnpm api:report changes between the base's build and this head's.
• Interaction proofs: packages/core/src/Item/__tests__/ItemSwipeActions.interaction.browser.spec.ts, with the animator, snap-point drag and gesture specs, in Chromium and WebKit.
• Accessibility contracts: ItemSwipeActions.a11y.chromium.spec.ts and ItemSwipePinchZoom.a11y.chromium.spec.ts.
• Sizes: each public import built with the size report's esbuild options from a build of packages/core, base against head.
• The docsite's preview build (apps/docsite/vercel.json's build command). createSwipe's doc is a new hook page, so the ShadCN route lock (internal/shadcn-registry/routes.lock.json) gains one entry, hook-create-swipe at hooks/create-swipe; no existing name or path moves.

Writing rubric results

Records edited:

• architecture:react-component-runtime INV13: the known public-API gaps gain the row's jobs, for spec-owner review;

• spec:AST-057: the contract table's readiness cell and the current-state impact drop the implementation gap this pull request closes; the current-state impact says ListItem passes swipe to its Item, and records the long-task line above;

• component:ListItem (ListItem.spec.md): its swipe gap goes, and the readiness cell counts the two that remain.

• R1 narration: none. Each edit states what the system does, in the present tense; the removed lines described what was not yet built.

• R2 research: none. The long-task measurements stay in this description; the record keeps one sentence.

• R3 evidence: INV13's addition cites spec:AST-057 AR2, FR2 and FR5; no evidence cell names a pull request.

• R4 rejected alternatives: none added.

• R5 change references: none added; the records name no pull request or issue.

• R6 open questions: none.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
astryx	Ready	Preview	Nov 2, 2026 9:29am UTC

[astracat-bot]

AI review status for this pull request.

Review status	Updated
❓ Needs maintainer input (for maintainers only)	Nov 2, 2026, 9:44 AM UTC

[astracat-bot]

This pull request needs a spec decision before it can be approved: spec:AST-057 on main lists two-finger trackpad swipes as a non-goal and its AR2 keeps the swipe off pointers that can hover, while the DEC-7 cited here exists only in the open #8646, whose FR4 would also make a pen swipe, which this change forbids. An engineering owner records that decision by commenting /approve-spec 6df8dd02f0913da1c01e72453a6415dffd3b9cec on this pull request, and the author may do so as an engineering owner. Whatever that decision, two changes are needed:

• Land or retarget the dependency pull requests, so this diff carries only its own 381 lines instead of 53,279.
• When a trackpad has opened a row, a click on its content should close it without running the row's action, as a finger's tap already does.

[Automated review]

[astracat-bot]

Spec hold: spec:AST-057 on main lists two-finger trackpad swipes as a non-goal (line 138), and its AR2 keeps the gesture's paint, tab order and accessibility tree untouched on a pointer that can hover (lines 250–252). The DEC-7 cited here exists only in the open #8646, so this opt-in waits on an engineering owner's decision on that record.

[astracat-bot]

Now that a trackpad can open a row, this dismissal still takes only a finger. In Chromium, a click on the content of a trackpad-opened row reaches the row's onClick and the row stays open, while a finger's tap closes it and activates nothing (Item.test.tsx:1178). Letting any pointer's press dismiss while the row is open, and swallowing its click through the click guard, would match the finger path; a test for a wheel-opened row would pin it.

[vjeux]

Fixed in 08e6916: the dismissing tap now takes every pointer and the click guard swallows the click, so a mouse click on a trackpad-opened row closes it; tests added in both engines.

[github-actions]

PR Analysis Report

Current run: The current CI run concluded skipped. Current analysis is unavailable.

---

_{Generated by PR Enrichment workflow | View current CI run}

[astracat-bot]

This pull request needs a spec decision before it can be approved: spec:AST-057 on main lists two-finger trackpad swipes as a non-goal and keeps the swipe off pointers that can hover (AR2, DEC-4), while the DEC-7 cited here exists only in the open #8646, whose FR4 would also make a pen swipe, which this change forbids. An engineering owner records that decision by landing the spec:AST-057 amendment with the pen settled, or ruling otherwise, and commenting /approve-spec 08e6916d0e698009bca4d50efdd10c6824a0c381 on this pull request; the author may do so as an engineering owner. Whatever that decision, two changes are needed:

• Land or retarget the dependency pull requests, so this diff carries its own 458 lines instead of 53,360.
• Under Reduce Motion on a fine pointer, a trackpad or wheel swipe's settle should complete at once, as design:reduced-motion DR3 requires, instead of cross-fading like a touch release (useSwipeAction.ts:777, :834).

[Automated review]

[astracat-bot]

Under Reduce Motion this treats every release as a touch release, so when a trackpad or wheel gesture ends on a fine primary pointer the row's settle cross-fades for up to 200 ms. design:reduced-motion DR3 requires any other transition on a fine pointer to complete at once (DR2's exception covers only the release of a touch drag), and #8646's FR5 asks the same of a trackpad-started swipe. Passing touchRelease only when a finger's contact ended the drag, here and in the slide-out fade at line 834, with a Reduce Motion test for a wheel-opened row, would fix it.

[astracat-bot]

The trackpad, wheel and pen swipe now matches spec:AST-057 on main (FR4, FR5, DEC-7), but approval needs an engineering owner’s decision on two things this head adds without current spec: the new component:BottomSheet element job for the input hold’s shield (BottomSheet.spec.md, lines 232–236), and whether a mouse or pen press on an open row’s own content should close it and activate nothing (useSwipeAction.ts, lines 1308–1332). An engineering owner records that decision by commenting /approve-spec bfef65e146dc5a8089b53dd84dd6c6dfafb1d1bd on this pull request, or by ruling otherwise in a comment; the author may do so as an engineering owner. Whatever that decision, two changes are needed:

• Land or retarget the dependency pull requests, so this diff carries its own 637 lines instead of 54,294.
• In the swipeActions doc and JSDoc, say a swipe-only verb is unreachable “by keyboard and by a mouse drag” and that an open row closes on a press elsewhere (Item.doc.mjs:202, Item.tsx:280 and :283).

[Automated review]

[astracat-bot]

Spec hold: with no pointer gate, a mouse or pen press on an open row’s own content now closes it and its click is swallowed. spec:AST-057 on main states no content-press dismissal for any pointer (FR7 lists a pointer outside, Escape, firing an entry and unmount; DEC-7 covers reveal and fire), so this waits on an engineering owner’s decision.

[astracat-bot]

A mouse’s horizontal wheel now reaches a swipe-only verb (DEC-7), and a press from any pointer outside closes an open row (FR7). Saying “unreachable by keyboard and by a mouse drag”, as spec:AST-057 asks the docs to, and “on a press elsewhere” on line 283 would match; the same two phrases are in Item.doc.mjs:202.

[astracat-bot]

Spec hold: this adds a named element job that the record on main does not have, and its verification row fails an element in the open sheet’s frame with no named job. An engineering owner approves or rejects it, ideally with the pull request that adds the shield.

[astracat-bot]

The pen, trackpad and wheel swipe matches spec:AST-057 on main (FR4, FR5, DEC-7), but approval needs an engineering owner's decision on two things this head adds without current spec: whether a mouse or pen press on an open row's own content should close it and activate nothing (useSwipeAction.ts, lines 1308–1333), and the new component:BottomSheet element job for the input hold's shield (BottomSheet.spec.md, lines 232–236). An engineering owner records that decision by commenting /approve-spec 90bf79fdcdaa1bba8793bb0ba4d610bae4c3977b on this pull request, or by ruling otherwise in a comment; the author may do so as an engineering owner. Whatever that decision, three changes are needed:

• Land or retarget the dependency pull requests, so this diff carries its own 644 lines instead of 54,742.
• In the swipeActions doc and JSDoc, say a swipe-only verb is unreachable "by keyboard and by a mouse drag" and that an open row closes on a press elsewhere (Item.tsx lines 280 and 283, Item.doc.mjs lines 202 and 257).
• Add the physical-device check spec:AST-057's verification asks for: a real trackpad (a slow swipe and a quick flick) and a real pen on the swipe-actions story.

[Automated review]

[astracat-bot]

Spec hold: with no pointer gate, a mouse or pen press on an open row's own content now closes it and its click is swallowed. spec:AST-057 on main states no content-press dismissal for any pointer (FR7 lists a pointer outside, Escape, firing an entry and unmount; AR3 keeps the primary's activation), so this waits on an engineering owner's decision.

[astracat-bot]

A mouse's horizontal wheel now reaches a swipe-only verb (DEC-7). spec:AST-057 asks the docs to say "unreachable by keyboard and by a mouse drag"; the same phrase is in Item.doc.mjs:202 and :257.

[astracat-bot]

FR7 closes an open row on "a pointer landing anywhere outside it", which this head does for every pointer; "on a press elsewhere" would match here and in Item.doc.mjs:202.

[astracat-bot]

Spec hold: this names an element job the record on main does not have, and its verification row fails an element in the open sheet's frame with no named job. An engineering owner approves or rejects it, ideally with the pull request that adds the shield.

[astracat-bot]

The pen, trackpad and wheel swipe matches spec:AST-057 on main (FR4–FR6, DEC-7), but approval needs an engineering owner's decision on three spec deltas this head carries without current authority: a mouse or pen press on an open row's own content closing it without activating anything (useSwipeAction.ts, lines 1300–1324), the new component:BottomSheet element job for the entrance shield (BottomSheet.spec.md, lines 233–237), and the new spec:AST-094 FR41 rule for claimed mouse and pen drags (docs/specs/AST-094-touch-presentations/spec.md, lines 471–475). An engineering owner records that decision by commenting /approve-spec 6c08e005cc6ac6102d2a33aba151779fa96a843a on this pull request, or by ruling otherwise in a comment; the author may do so as an engineering owner. Whatever that decision, three changes are needed:

• Land or retarget the dependency pull requests, so this diff carries its own 596 lines instead of 54,299.
• In the swipeActions JSDoc and doc, say a swipe-only verb is unreachable “by keyboard and by a mouse drag” and that an open row closes on a press elsewhere (Item.tsx lines 280 and 283, Item.doc.mjs lines 202 and 257).
• Add the physical-device check spec:AST-057's verification asks for: a real trackpad (a slow swipe and a quick flick) and a real pen on the swipe-actions story.

[Automated review]

[astracat-bot]

Spec hold: with no pointer gate, a mouse or pen press on an open row's own content now closes it and its click is swallowed. spec:AST-057 on main states no content-press dismissal for any pointer (FR7 lists a pointer outside, Escape, firing an entry and unmount; AR3 keeps the primary's activation), so this waits on an engineering owner's decision.

[astracat-bot]

A mouse's horizontal wheel now reaches a swipe-only verb (spec:AST-057 DEC-7), and the record asks the docs to say “unreachable by keyboard and by a mouse drag”; the same phrase is in Item.doc.mjs (the swipeActions description and the guidance).

[astracat-bot]

FR7 closes an open row on “a pointer landing anywhere outside it”, which this head does for every pointer; “on a press elsewhere” would match here and in Item.doc.mjs.

[astracat-bot]

Spec hold: this names an element job the record on main does not have, and its verification row fails an element in the open sheet's frame with no named job. An engineering owner approves or rejects it, ideally with the pull request that adds the shield; the bullet also leaves the shield's accessibility exposure unstated.

[astracat-bot]

Spec hold: FR41 on main says a mouse “otherwise keeps its released behavior”; this adds a rule for drags a recognizer claims (no new selection, a later claim clearing the page's selection, a claim at pointerdown keeping it) and a verification row for it, which waits on an engineering owner's decision, ideally in the pull request that adds the every-pointer host.

[astracat-bot]

Approval waits on an engineering owner's spec decision, and some changes are needed either way. An engineering owner records that decision by commenting /approve-spec 2f7567fe6c17fa37ea31db67dd5e7a76e4d7fcf3 (the author may do so as an engineering owner) or rules otherwise. Three points need it: whether a mouse, pen or trackpad press on an open row's content should close it (spec:AST-057 FR7 and AR3 name a touch tap), the two new INV13 gap entries in architecture:react-component-runtime, and whether spec:AST-094 FR43 or spec:AST-057 FR4 and DEC-7 governs a mouse's horizontal wheel.

• The diff is 20,249 authored lines, including five open dependency pull requests; land or retarget them, then split the rest.
• The row's settle now animates a custom property on the main thread, which spec:AST-094 FR5 rules out, and the new spec:AST-057 line says main behaves the same although main settled with a compositor transform transition.
• The open row adds its own document pointerdown and scroll listeners beside subscribeInput; subscribe through it when the row opens instead.
• spec:AST-057 asks for a physical-device check of FR5 and FR6, item-swipe-gesture-system.md claims a finger fix that already worked on main, and the docs say an open row closes "on a touch elsewhere" where any pointer closes it.

[Automated review]

[astracat-bot]

Spec hold: any pointer's press on an open row's content now closes it and its click is swallowed. spec:AST-057 FR7 and AR3 name only a touch tap, so this waits on an engineering owner's decision.

[astracat-bot]

These document listeners sit beside subscribeInput (utils/inputStream.ts), the one entry for document-wide input in core (architecture:react-component-runtime INV16). The work is caused by the open transition, which setSwipeState already handles for the press hold. Subscribe there through subscribeInput and drop the Effect.

[astracat-bot]

A row's settle drawn into this custom property no longer advances through a main-thread block, which spec:AST-094 FR5 requires of every release. On main the root settled with a transform transition that the compositor runs.

[astracat-bot]

On main the row's root settles with a transform transition, which keeps moving through a long task, so "the released swipe behaves the same" does not hold for a settle.

[astracat-bot]

On main a finger already swiped the row whatever the primary pointer: the gate was pointerType !== 'touch' alone and the layer passed isEnabled: true. Suggested wording: "A pen now swipes the row as a finger does."

[astracat-bot]

Any pointer landing outside closes an open row (onOutside in Swipe/useSwipeAction.ts), as spec:AST-057 FR7 says. "on a press elsewhere" would match, here and in Item.doc.mjs.

[astracat-bot]

Approval waits on an engineering owner's spec decision, and some changes are needed either way. An engineering owner records that decision by commenting /approve-spec e870248ecc40ef36a28bd41753c59feb2457a527 (the author may do so as an engineering owner) or rules otherwise. Three points need it: whether a mouse, pen or trackpad press on an open row's content should close it (spec:AST-057 FR7 and AR3 name a touch tap), the two new INV13 gap entries in architecture:react-component-runtime, and whether spec:AST-094 FR43 or spec:AST-057 FR4 and DEC-7 governs a mouse's horizontal wheel.

• The diff is 20,253 authored lines, including five open dependency pull requests; land or retarget them, then split the rest.
• The row's settle now animates a custom property on the main thread, which spec:AST-094 FR5 rules out, and the new spec:AST-057 line says main behaves the same although main settled with a compositor transform transition.
• The open row adds its own document pointerdown and scroll listeners beside subscribeInput; subscribe through it when the row opens instead.
• spec:AST-057 asks for a physical-device check of FR5 and FR6, item-swipe-gesture-system.md claims a finger fix that already worked on main, and the docs say an open row closes "on a touch elsewhere" where any pointer closes it.

[Automated review]

[astracat-bot]

Spec hold: any pointer's press on an open row's content now closes it and its click is swallowed. spec:AST-057 FR7 and AR3 name only a touch tap, so this waits on an engineering owner's decision.

[astracat-bot]

These document listeners sit beside subscribeInput (utils/inputStream.ts), the one entry for document-wide input in core (architecture:react-component-runtime INV16). The work is caused by the open transition, which setSwipeState already handles for the press hold. Subscribe there through subscribeInput when the row opens, unsubscribe when it closes, and drop this Effect.

[astracat-bot]

A row's settle drawn into this custom property no longer advances through a main-thread block, which spec:AST-094 FR5 requires of every release. On main the root settled with a transform transition that the compositor runs.

[astracat-bot]

On main the row's root settles with a transform transition, which keeps moving through a long task, so "the released swipe behaves the same" does not hold for a settle.

[astracat-bot]

On main a finger already swiped the row whatever the primary pointer: the gate was pointerType !== 'touch' alone and the layer passed isEnabled: true. Suggested wording: "A pen now swipes the row as a finger does."

[astracat-bot]

Any pointer landing outside closes an open row (onOutside in Swipe/useSwipeAction.ts), as spec:AST-057 FR7 says. "on a press elsewhere" would match, here and in Item.doc.mjs.