feat(gestures): a recognizer created with claimsTouches: false is a watcher by cixzhang · Pull Request #8589 · meta/astryx

cixzhang · 2026-10-26T20:47:47Z

Depends on #8315 (and its predecessors); review this pull request's own commit (feat(gestures): a recognizer created with claimsTouches: false is a watcher), about 210 lines over #8315's head.

Split out of #8336, which keeps the gesture host's ownership changes.

User need

Some touch behavior only follows the person's scroll and never takes the touch: a tab bar that minimizes while the page scrolls down, a header that collapses. It has to see the finger without delaying the scroll, without stopping another recognizer, and without cancelling a press or click under it.

Problem and solution fit

claimsTouches: false makes a recognizer a watcher: it never claims, so its element gets only passive listeners and a scroll never waits for it.
It prevents no other recognizer and none prevents it, as spec:AST-094 FR15 now states for watchers (spec(AST-094): FR15 exempts watchers from recognizer exclusivity (DEC-7) #8517).
It cancels no press or click under it when it begins (FR11, FR19).
Its direction writes no touch-action: it restricts no scroll (FR20, DEC-7).
It keeps tracking a touch the browser takes for a scroll, reporting through the native scroll and ending at the lift (FR20, DEC-7).

Current specification

spec:AST-094 FR11, FR15, FR19, FR20, DEC-7.

Public delta and caller burden

None: an internal recognizer option, not exported from the package entry point; the changeset is [chore].

Design and alternatives

One option on the base recognizer, so any recognizer kind can watch; a separate watcher class would duplicate each recognizer's tracking.

Evidence

Unit tests (watcherRecognizer.test.ts): runs alongside a claiming pan on the same touch and claims nothing; cancels no press under it; follows a touch the browser took for a scroll until the lift; gets only passive touchmove listeners; writes no touch-action for its direction.

Writing rubric results

No record text of its own: AST-094 FR15's watcher exemption is on main (#8517); R1–R6: none.

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.

…ated consumer The gesture host has no delegated consumers and no scroll handoff, so its header drops both, and one changeset covers the press tracking, the one press writer and the hooks. Both texts are cixzhang's from c34bf73 on #8073, taken verbatim so the next #8076 merges without a conflict.

… the gesture hook fixes

…s pending waitForAnimation and watch.started() left pending animations (no start time yet) out of the running set, so an earlier animation that had finished answered for the one the action started. A pending animation in scope now keeps the wait going. Self-test: a short animation ends, then a longer one starts; the wait returns the longer one (it returned the 40 ms one before). The README and JSDoc say how a mid-run reverse or replay is misread. Co-authored-by: Christopher Chedeau <vjeuxx@gmail.com>

…ase is one handle, and REDUCED_TRANSITION keeps createAnimator's docs - attach rebuilt only channels away from identity, so an identity the animator held inline over a stylesheet that differs (set({opacity: 1}) over opacity 0) was never written on the node that replaced the element; the new node showed 0 while get() read 1. The held identities now carry through a detach to the next node. - decay under Reduce Motion with one axis past an edge and another coasting returned a handle that covered only the coasting axis; stop() left the cross-fade to land. The handle now covers both calls. - REDUCED_TRANSITION moves above createAnimator's doc comment.

Storybook's highlight module scans the page on every insertion in WebKit as in Chromium, so the self-test now counts the insertion's getComputedStyle calls in both engines; only Chromium's trace counts the style recalculations they force. It no longer serves Storybook for an engine it skips.

…handle for a mixed release); decay's Reduce Motion docs - attach(): #8055's held identities, plus a pseudo-element's: its identity hold (a static animation the attach cancels) is held again on the next node; a borrowed opacity is never held. Test: a ::backdrop's hold follows a node replacement. - decay(): #8055's single handle for a release with one axis past an edge, with `started` covering both calls. Under this layer's plan a stop finishes the cross-fade and lands that axis at once, while invisible, so #8055's test checks that nothing moves after the stop. - Docs: the header, decay()'s and MotionCurve's comments, the motion README and the changeset now say that a curve marked isTransition (a coast's spring back from past an edge) plays through the Reduce Motion policy: a cross-fade on a coarse pointer, at once elsewhere.

…lacement, one handle for a mixed-axis release) as a dependency

…8315

…e animations Inside a trace, waitForRest waits on timers for each animation's scheduled end. A scroll- or view-timeline animation (DateInput's wheel rows) has none: its time is a percentage, the wait came out NaN, and the timer re-armed with no delay until the deadline, where the untraced wait rests. It now skips animations that are not on the document timeline, as holdAnimations does. Self-test: a scroll-timeline row inside a trace; it rejected at the deadline before (Chromium, 3x repeat pass).

…on is pending) into the frame evidence

…t moved frame, and the harness's waits run on the real timers With a compositor record, the moved criterion counted every compositor frame in the judged part, but only a gap between two moved frames can be credited to the compositor, so the frames before the first and after the last moved frame read as misses: at one screencast frame in six, a composited animation failed in 5 of 7 runs, where the same frames passed without the record. The criterion now counts the compositor's frames from the first to the last moved frame; the gaps before and after are the stall check's, which already measures the window's edges. blockMainThread(..., {during}) and sampleRects waited on the page's own requestAnimationFrame and setTimeout, which freezeTimers holds, so neither settled under a frozen timer gate, though the harness's own waits are documented as not held. They now run on the gate's real frame and timeout (the gate exposes its real setTimeout too). A unit case at one frame in six and a self-test under a frozen gate fail on the previous code.

…up a wait waitForAnimation and watch.started() counted a scroll- or view-timeline animation (DateInput's wheel rows) as running; it never progresses on its own, so the wait never answered with the animation that finished and rejected at its timeout. Only document-timeline animations count now, as for holdAnimations. Self-test: a scroll-timeline row beside a short animation that ends; it rejected before (Chromium, 3x repeat pass). The installCurveEnd doc comment sits above installCurveEnd again. Co-authored-by: Christopher Chedeau <vjeuxx@gmail.com>

…s this layer has, and coast says a transition lands at once on a fine pointer

…ator's browser proofs

… pointer); the changeset says the same animator.test.ts keeps this layer's header (no compositor browser spec in this layer). The changeset now says decay sends a spring back from past an edge through the Reduce Motion policy: a cross-fade on a coarse pointer, at once elsewhere.

…side a scroll view The press profiles' README said a cell's highlight always waits the content-touch delay. PressTracker waits it inside a scroll view and lights at once outside one (spec:AST-094 FR11), as its own header says; the README now says so too.

…wording) as a dependency

What one call becomes under Reduce Motion (an opacity-only cross-fade whose travel lands while the element is transparent, or everything at once; a dip on a borrowed stylesheet opacity) moves from createAnimator's closure to reducedMotionPlan.ts, a pure planner that reads the animator through a view and returns the plan; the animator turns the plan into its segments and runs the landings. No behavior change: the animator's tests pass unchanged, and the planner has its own. Co-authored-by: Christopher Chedeau <vjeuxx@gmail.com>

…ed pull requests This pull request keeps the animator's Reduce Motion plan (now its own planner module), the browser-set start of a call from rest, promotion, atRest, rebase and animatorFor. Two features it carried move up the stack, each with its own tests and changeset: - play(curves), baking any closed-form curve like a spring; - driving a pseudo-element (a dialog's ::backdrop). The changeset is split the same way. Co-authored-by: Christopher Chedeau <vjeuxx@gmail.com>

play(curves) bakes any curve given as value and velocity functions of time (a spring chasing a moving target from solveSpringTowardRamp, a fitted UIKit curve) exactly like a spring: presentation values, velocities, stop, finished and the shared clock work the same, a later animate starts from the curve's value and velocity, and under Reduce Motion a curve that moves is travel the policy replaces. Tests and a changeset of its own. Co-authored-by: Christopher Chedeau <vjeuxx@gmail.com>

createAnimator(element, {pseudoElement: '::backdrop'}) animates a pseudo-element: with no inline style, a static filling animation holds its value after a curve ends and after set; a custom property on its host promotes it; a Reduce Motion dip borrows and hands back the pseudo-element's own opacity, and its holds follow a node replacement. animatorFor keeps one animator per element and pseudo-element. Tests and a changeset of its own. Co-authored-by: Christopher Chedeau <vjeuxx@gmail.com>

…ot the page's frames around it "a harness read inside a trace is not the page's long task…" also asserted that no long task started anywhere in the read's window. On CI's runner each frame of the fixture's 1,200 running rows took 51-54 ms, the page's own long tasks, and failed it; at 4x CPU locally they take 126-220 ms. The read itself was set apart correctly. The test now asserts the read's own classification, and removes the rows before the timer, click and direct-call windows, so each holds only its own work. It passes at 4x CPU, where the previous assertion failed.

…old up a wait) into the frame evidence

…anddown

…ter the browser takes the touch A host that stopped feeding the finger at the takeover and still delivered the lift would have passed. Each transition now records the list's scrollTop, and the test expects a 'changed' while the list is scrolled (spec:AST-094 FR20: it follows the scroll to the lift).

…her browser test) into the arena fixes The host-owner changeset keeps this PR's own listeners for the scroll lock and finger-scroll watch, with meta#8336's lone-finger scroll-edge hold.

vjeux · 2026-10-27T05:17:12Z

The lint job's @typescript-eslint/array-type rule forbids Array<T>, and the watcher test now writes its record type as Array<{to: string; scrollTop: number}> three times (gestures.browser.spec.ts lines 1170, 1172 and 1195 on this head, b87b40b). That's the job's only error, and it fails lint here and on every head that merges this one (#8336, #8320). A fix that writes {to: string; scrollTop: number}[] instead, on this head: vjeux@42ada84 (branch uikit/f2-8589-array-type).

CI's lint rejected the three Array<T> types the previous commit added (@typescript-eslint/array-type); the test now names its record once (WatchTransition) and reuses it.

astracat-bot

Requesting changes. The watcher itself (claimsTouches: false) follows spec:AST-094 FR11, FR15, FR18, FR19, FR20 and DEC-7, and its real-touch browser case now checks that it keeps reporting moves after the browser takes the touch. Against main, though, this pull request carries 20,634 authored lines, all but the watcher's 525 from the open #8315 line, and it conflicts with main in packages/core/src/motion/animator.ts, so no CI test has run on this head. Please land #8315 and its predecessors and merge main, or base this pull request on a branch in this repository that holds #8315's head, so the diff is the eight-file watcher.

For the stack's own pull requests: the screen-edge pan decides at 10 px inside its band on a wide view (ScreenEdgePanGestureRecognizer.ts:117-125) against FR18's 15 px, and the fine-pointer action group cites a family:action-groups FR5 that has no record on main.

[Automated review]

…finger, meta#8589's watcher test, its lint fix; meta#8076 3bfd7c3; the coarse-pointer query in utils/) into meta#8293

vjeux · 2026-10-27T06:07:13Z

The arena resets a recognizer to possible after it ends, as UIKit does, and the transition listener records that reset too. So the watcher test's last transition is possible, and expect(states.at(-1)?.to).toBe('ended') fails in Chromium on this head (b57ac07, 2/2) and on every head that carries it. A fix that asserts the ended-then-possible tail, on this head: vjeux@1b6e19f (branch uikit/f2-8589-watcher-tail). With it, the test passes 3/3 and this head's Chromium gesture suite passes 32/32.

…t reads its end before the arena's reset) into facebook#8336

…8589's watcher test reads its end) into meta#8320

…8589's watcher test reads its end) into meta#8293

…8589's watcher test reads its end) into meta#8356

…8589's watcher test reads its end) into meta#8358

…8589's watcher test reads its end; the gestures-table gate retired) into meta#8360

…s watcher test fixed)

fZ retires the long-press check that kept DropdownMenu off the gestures table (meta#8358) and carries meta#8589's watcher-test fix and meta#8076 e6ab0b9.

's base fZ carries meta#8076 e6ab0b9 (DEC-22's 10 px rule and meta#8641's arena fix), the meta#8589 watcher-test fix and the meta#8358 gate retirement; taken at this stack's real push (LANES 01:13).

and the meta#8589 watcher-test fix) into c24-8147-base

vjeux and others added 30 commits October 26, 2026 03:27

Merge #8309 into #8315

2a44226

Merge #8309 (dd8e946: #8076 ee34b25 with the harness line, main) into…

b9c536f

… the gesture hook fixes

Merge f60e87d (#8056 on #8055's 7ad0455: held identities across a rep…

6eb2193

…lacement, one handle for a mixed-axis release) as a dependency

Merge 79a7a90 (main) as a dependency

86ac77d

Merge #8270 (main 5b691cd) into #8309

a2ea848

Merge cixzhang's #8309 head (dd8e946: #8076 ee34b25 and main 5b691cd)

50f39fb

Merge main (5b691cd) into #8270, as cixzhang's #8309 does

d0200c3

Merge #8309 (her third pass folded: #8076 ee34b25, main 5b691cd) into #…

e84b0ad

…8315

Merge cixzhang's #8315 head (b9c536f: her #8309 merge)

f8faec3

Merge #8505 (9320644: a wait keeps waiting while the action's animati…

b4234d7

…on is pending) into the frame evidence

docs(motion): the animator's SYNC list and test header name only file…

8fc7f2e

…s this layer has, and coast says a transition lands at once on a fine pointer

Merge #8446 (9d5c20a: the harness line on #8505's waits) into useAnim…

0543b78

…ator's browser proofs

Merge 79dc151 (#8056 on #8055's 8fc7f2e: the Reduce Motion changeset …

a4b10f9

…wording) as a dependency

Merge #8505 (462d82c: a scroll- or view-timeline animation does not h…

fa8c8e2

…old up a wait) into the frame evidence

Merge branch 'interaction-harness-3-frames' into uikit/f3-recorder-st…

1ffb20e

…anddown

vjeux mentioned this pull request Oct 27, 2026

fix(gestures): pointer-read fingers get the blur refusal and the pinch rule of the Touch Events path #8309

Draft

4 tasks done

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8589 (b87b40b) for meta#8336's review link

709e619

vercel Bot deployed to Preview October 27, 2026 05:04

View deployment

vjeux mentioned this pull request Oct 27, 2026

fix(core): gesture hooks follow their element; recognizer plumbing is protected #8315

Closed

test(gestures): the watcher's browser test names its record type (lint)

b57ac07

CI's lint rejected the three Array<T> types the previous commit added (@typescript-eslint/array-type); the test now names its record once (WatchTransition) and reuses it.

cixzhang added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8336 (meta#8589's lint fix) into the arena fixes

ea8531c

astracat-bot Bot requested changes Oct 27, 2026

View reviewed changes

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8320 (cixzhang's ea8531c: the scroll-edge hold for a lone …

45b25e4

…finger, meta#8589's watcher test, its lint fix; meta#8076 3bfd7c3; the coarse-pointer query in utils/) into meta#8293

vercel Bot had a problem deploying to Preview October 27, 2026 06:15 Failure

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge F2's meta#8320 (ea8531c: meta#8589's lint fix) into meta#8068

0f732ea

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge the facebook#8589 watcher-test fix (1b6e19f: the real-touch tes…

03155c3

…t reads its end before the arena's reset) into facebook#8336

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8336 (meta#8076 e6ab0b9: FR18 per DEC-22, meta#8641; meta#…

3d7fa2e

…8589's watcher test reads its end) into meta#8320

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8320 (meta#8076 e6ab0b9: FR18 per DEC-22, meta#8641; meta#…

d835d1c

…8589's watcher test reads its end) into meta#8293

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge facebook#8293 (facebook#8076 e6ab0b9: FR18 per DEC-22, facebook…

ee5387a

…8589's watcher test reads its end) into meta#8356

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8356 (meta#8076 e6ab0b9: FR18 per DEC-22, meta#8641; meta#…

82fe8b7

…8589's watcher test reads its end) into meta#8358

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge meta#8358 (meta#8076 e6ab0b9: FR18 per DEC-22, meta#8641; meta#…

23ade34

…8589's watcher test reads its end; the gestures-table gate retired) into meta#8360

vjeux mentioned this pull request Oct 27, 2026

feat(Slider): under a finger the thumb tracks like UISlider #8190

Draft

4 tasks done

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Review base for meta#8657: and F2's fZ (meta#8360 23ade34: meta#8589'…

29f3fb7

…s watcher test fixed)

vjeux mentioned this pull request Oct 27, 2026

refactor(core): BottomSheet's drag runs on the gesture recognizers #8386

Draft

3 tasks done

vjeux mentioned this pull request Oct 27, 2026

fix(Popover): under a finger, a tap during the presentation is ignored, as on iOS #8223

Draft

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 27, 2026

Merge commit '3d7fa2ebcb' (meta#8320, F2's fZ: carries meta#8076 e6ab0b9

31d7643

and the meta#8589 watcher-test fix) into c24-8147-base

Navigation Menu

Repository navigation

feat(gestures): a recognizer created with claimsTouches: false is a watcher - #8589

feat(gestures): a recognizer created with claimsTouches: false is a watcher#8589
cixzhang wants to merge 1042 commits into
mainfrom
uikit/gesture-watcher

cixzhang commented Oct 26, 2026

Uh oh!

vjeux commented Oct 27, 2026

Uh oh!

astracat-bot Bot left a comment

Uh oh!

vjeux commented Oct 27, 2026

Uh oh!

This branch had an error being deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

cixzhang commented Oct 26, 2026

User need

Problem and solution fit

Current specification

Public delta and caller burden

Design and alternatives

Evidence

Writing rubric results

Scope

Uh oh!

vjeux commented Oct 27, 2026

Uh oh!

astracat-bot Bot left a comment

Choose a reason for hiding this comment

Uh oh!

vjeux commented Oct 27, 2026

Uh oh!

This branch had an error being deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants