Skip to content

feat(liftFromTrigger): one opt-in lift-from-trigger presentation - #8485

Draft
vjeux wants to merge 77 commits into
meta:mainfrom
vjeux:uikit/c20-lift
Draft

vjeux wants to merge 77 commits into
meta:mainfrom
vjeux:uikit/c20-lift

Conversation

@vjeux

@vjeux vjeux commented Oct 26, 2026 •

Copy link
Copy Markdown
Contributor

Stack: #9764 → #9769 → #8485 → #8201.

Behavior module: the trigger lift (liftFromTrigger).

Maintainer impact

The UIKit-parity program brings iOS 26's touch presentations to Astryx, and two of them lift their trigger: ContextMenu's long-press preview (#8201) and Lightbox's zoom (#8231). Each was building its own copy. ContextMenu's clone mounted the trigger's content twice, as a second React tree. Lightbox kept its own zoom copy. The two motions would drift apart, and the page paid for both.

Intended invariant

  • architecture:layer-runtime INV15: one opt-in lift module, apart from the runtime's hold and exit copy.
  • architecture:component-style-authoring INV7 and spec:AST-017 FR60: nothing at rest, at most five elements lifted.
  • spec:AST-107 FR7: a listed public-API gap, so it stays internal.

Change

There were three options. Each surface could keep its own copy, which doubles ContextMenu's tree. A clone of the trigger's DOM loses a React trigger's state and listeners. Or the trigger lifts itself. This takes the third, with an empty placeholder holding its spot: packages/core/src/liftFromTrigger/liftFromTrigger.ts exposes createLiftFromTrigger() and surfacePlacement. Lightbox's copy returns with its first user. Everything moves on the animator: the trigger on #9769's rect morph, and the tint on the layer's ::backdrop channel. A drop while either rises turns it with its velocity. The lifted trigger casts no shadow. The lift keeps the page's pinch-zoom (touch-action: pinch-zoom on its layer) and fits between the viewport's gutters on both axes. surfacePlacement reports when a surface fits on neither side (spec:AST-094 DEC-19); the layer runtime's fallback is then the caller's path. The caller names the lift's marks with a private attribute it owns, so the module writes none of its own. A caller can show its own surface once the layer shows, before the trigger lifts (onLayerShown). It can pin styles on the lifted trigger, which the lift puts back with its own (pin). Nothing imports the module yet.

Evidence

  • Before, ContextMenu's preview mounted the trigger twice; on this module, one trigger lifts.
  • At an earlier head, ContextMenu's touch story (feat(ContextMenu): a held finger lifts a block trigger above a tint #8201) in Chromium at 4× CPU slowdown. Opening: 51–71 ms of main thread before, 38–54 ms after. Closing: 38–57 ms before, 22–41 ms after. Twenty opens and closes leave nodes and listeners at baseline.

Scope

  • No intended public API, product behavior, visual direction, or policy change.
  • Product changes discovered during the work were removed or split.
  • Public text and artifacts contain no internal Meta context.

Testing

This round, at aa2f8e5, over #9769's head 18fcf70. The owners asked for the cleanups, and review asked for the rest. Nothing imports the module yet, so no product behavior changes:

  • fc3aaf9 drops liftCopy, hideSource and pictureOf, with their types and tests. They return with their first user.
  • aed8400 drops the uncalled dismissSurface and its exit-copy path, and the fixture's close.
  • 8cf7cdd folds the geometry into liftFromTrigger.ts, and 0cd53f6 drops useLiftFromTrigger.
  • 6f44618 drops presentSurface's unused press parameter and its surfaceStartWidth option.
  • 3016130, records only: architecture:layer-runtime INV15 names the entry points and says a surface beside the lift keeps its own exit.
  • 86d8f07: the pointer watch subscribes to the window's input stream, which Core keeps for document-wide input, in place of its own document listeners.
  • 20be748 merges feat(core): one rect morph for touch presentations that grow out of a source rect #9769's head 6c17632 with no conflict.
  • 215cff7, comments only, from review: the attribute option's doc names the marks the lift sets, its layer, anchor box and placeholder.
  • 2095d4f merges feat(core): one rect morph for touch presentations that grow out of a source rect #9769's head 18fcf70, which merges main and resolves its conflict there. It merges with no conflict, and this pull request's own diff is unchanged (same patch-id and files).
  • 0a995e2, red first, from review: a second drop while the source returns changes nothing, and the return goes on. At 2095d4f, the lift ended 16 ms after the second drop.
  • 3a5a35c fixes it: the first drop marks the lift as dropping, and a later drop changes nothing until the lift ends.
  • c16093a, records and comments only, from review: the layer runtime's gaps paragraph states INV15's gap as what a reader can see. ContextMenu and Lightbox do not use the module. onLayerShown's doc says it runs before the source's writes, and what the page holds then.
  • 4380723, comments and docs only, from review: the README says the lift's listeners end with it, since it adds no document listener. onLayerShown's doc says presentSurface, drop and dispose do nothing inside it.
  • 5eb0610, red first, from review: the fixture rounds its last row through li:last-child, and a new case lifts that row. At 4380723 the row rose with square bottom corners, since the placeholder after it ended the last-child match.
  • 8db9bf4 fixes it: the lift keeps the four corner radii as laid out, with the borders and padding it already keeps.
  • aa2f8e5, records only: architecture:layer-runtime INV15 lists the module's four functions and why one module serves every surface that lifts.

Local checks at 8db9bf4, one worker (aa2f8e5 changes one record; check:knowledge passes at it):

  • the tests this change reaches over feat(core): one rect morph for touch presentations that grow out of a source rect #9769's head (test:ui --changed): 1 file, 26 passed, the module's own unit file (nothing else imports it);
  • the lift's browser spec: Chromium 14 passed; WebKit under the CI profile 12 passed, 2 skipped. With the corner fix reverted, the new case fails in both engines;
  • the core typecheck, strict eslint and prettier on the changed files, check:changesets and check:knowledge.

Restacked onto #9769's head 6b6310c, which holds main's input hold, exit copy, presentation hold and one eased progress. This pull request carries only its own commits:

  • the module, its tests and README;
  • the layer-runtime gaps paragraph, re-based on main's text, so only the lift clause changes;
  • four adaptations to rules that landed since. Each spring reaches the rect morph as a {spring} motion. The caller names the lift's marks: attribute is required for a lift, and a zoom's copy was marked only when given one, until fc3aaf9 removed the copy. Comments state reasons, not records, in sentences of at most 25 words. The hold proof taps with a finger.

The pull requests it carried are merged (#8342, #8186, #8076) or superseded (#8329, by #9764 and #9769).

Main's input hold judges each event by its own pointer, so a mouse keeps its clicks during a presentation. The lift is a touch presentation, so its hold proof taps with a finger. The proofs import Playwright through the interaction harness.

The review rounds add fourteen commits, each test ahead of its fix and red before it:

  • A tall trigger fits between the gutters, and placement can answer none (3 units red).
  • A pinch on the lifted backdrop zooms the page (red in Chromium: the scale stayed 1).
  • A drop while the tint rises turns it with its velocity (red in both engines: the clear restarted from rest).
  • The shadow moves onto its own rect morph; the frame-for-frame shadow proof covers it unchanged. A later commit drops the shadow at the spec owner's request: the page gains three elements while lifted, not four.
  • A dropped row's own margin returns with no transition running. This proof guards the restore's style flush: it is red when the flush is removed. The flush now reads transitionProperty, not width, so a lift that replaces another lays the page out once. No proof can count layouts from page script.
  • The growth proof read the menu two frames in, and a late WebKit frame let it run past its bound once. It then read the menu's own animations 30 ms in. The menu has since stopped growing, and its proof checks a fade.
  • A source laid out a fraction of a pixel wider than its rounded width lifted narrower, and its text wrapped past its box. The lift now pins the layout size (red in both engines).
  • A caller's surface shown after the lift's writes recalculated the lifted page first, 4–11 ms in WebKit. onLayerShown runs once the layer shows, before the source's writes (a unit case pins the order).
  • A caller that styled the lifted source put those styles back itself. pin takes them with the lift's own (a unit case pins the restore).

Local checks, one worker:

  • At an earlier head, a0eec83: the liftFromTrigger and motion units (1,075 passed), the core typecheck, eslint and check:changesets. The lift spec: Chromium 16 passed; WebKit under the CI profile 14 passed, 2 skipped.
  • At 4aa36aa: the fraction case in Chromium and WebKit, red without the fix.
  • At the same code before a comment split and the case's wait removal: the lift spec. Chromium 16 passed; WebKit under the CI profile 14 passed, 2 skipped. Also the liftFromTrigger units (29 passed), eslint and the core typecheck.
  • At eb7e753: eslint on the spec, and the growth proof in Chromium (2 passed) and in WebKit under the CI profile (3 passed).
  • At 0374ace, which lacks only that proof's presence assert: the lift spec in WebKit under the CI profile (13 passed, 2 skipped).
  • At ff13ea4, before the growth proof's fix: core typecheck, the liftFromTrigger units (29 passed), eslint, and the lift spec in Chromium (15 passed).
  • At 106ec92: check:changesets.
  • At eac90f8, the same own diff on feat(core): one rect morph for touch presentations that grow out of a source rect #9769's previous head 99d352d: eslint on the module (0 errors). Also the lift spec three times in Chromium (48 passed) and in WebKit under the CI profile (42 passed, 6 skipped).
  • At 34a81be, on feat(core): one rect morph for touch presentations that grow out of a source rect #9769's head 3bfa524, before the comment rewrap and the pinch proof's settled wait. CI=true pnpm lint (0 errors), check:knowledge, check:changesets and test:node (17,544 passed).

In the lift fixture (pnpm test:interaction packages/core/src/liftFromTrigger, Chromium and WebKit):

  1. Lift a list item: it rises into the top layer at its size, with no shadow, and its list keeps only li children.
  2. Drop it while it grows: it lands, and ten lifts leave no node or listener behind. The turn-around is the rect morph's own proof: "on screen, the box keeps moving through the switch without a jump".
  3. Tap while it grows: nothing receives the tap.
  4. Pinch on the backdrop while it is lifted: the page zooms, and nothing receives a tap. That runs in Chromium; WebKit pinch-zooms from injected touches only on the iOS simulator.
  5. Drop while the tint rises: the tint rises a little more, then clears.
  6. Drop a row with its own margin and transitions: the margin returns at once, and no transition runs.
  7. Lift the last row, which li:last-child rounds: it keeps its corners while lifted.

The unit tests (pnpm vitest run packages/core/src/liftFromTrigger) cover the lift's elements, its holds, the restore, a repeated drop and the geometry.

Writing rubric results

INV15's closing sentences list the module's four functions and why one module serves every surface that lifts.

R1–R7: no narration, research, evidence-cell changes, rejected alternatives, change references, open questions or project status; the layer-runtime gaps paragraph only states which gaps remain.

Public and internal API delta

Public: none.

Internal, new: liftFromTrigger/liftFromTrigger.ts, named in architecture:layer-runtime INV15. Nothing imports it yet; #8201 is its first user.

  • createLiftFromTrigger() lifts a source, presents a surface beside it, drops and ends. Usage: const lift = createLiftFromTrigger();.
  • surfacePlacement(lifted, surfaceHeight, viewport, gap) picks the surface's side, or answers none. Usage: const {side} = surfacePlacement(lifted, menuHeight, viewport, gap);.
  • liftFromTriggerScale and liftFromTriggerRect, the geometry, are exported for the module's unit tests only.

Why: ContextMenu's preview mounted its trigger twice, and Lightbox kept its own copy. One module lifts the trigger itself. Existing ones considered: the layer runtime's hold and exit copy (INV14), which play a layer's own presentation, not a trigger lifted out of the page. #9769's rect morph moves the lift. A clone of the trigger's DOM loses a React trigger's state and listeners (Change). An engineering owner's API decision on the module is pending.

Lines, over #9769's head: product +1,076/−0 (liftFromTrigger.ts), tests +1,455/−0, README +51/−0, records +14/−8, changesets +7/−0.

App-level size, eager JS gzip of the example-vite page in a Vite production build, over #9769's head: no change as shipped, with one ContextMenu or with one Switch (+0 B each). Nothing imports the module yet.

@vercel

vercel Bot commented Oct 26, 2026 •

Copy link
Copy Markdown

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

Project Deployment Actions Updated
astryx Ready Ready Preview Nov 12, 2026 7:53am UTC

Request Review

@astracat-bot

astracat-bot Bot commented Oct 26, 2026 •

Copy link
Copy Markdown
Contributor

AI review status for this pull request.

Review status Updated
🟢 Reviewed (for maintainers only) Nov 12, 2026, 8:10 AM UTC

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approval is on hold for an engineering owner's spec decision: the branch adds architecture:layer-runtime INV13 and its Exit copy section, rewrites the gesture-tuning ruling in spec:AST-057, and ships a press-feedback exception to spec:AST-094 FR19 that its own README calls pending. An engineering owner can record approval of the head's records by commenting /approve-spec <full head SHA> on this pull request (the author may do this as an engineering owner), while the FR19 exception needs its amendment on main or should come out. Separately, please land the stacked dependencies and rebase so this pull request carries only the lift commit, since the branch now has 39,463 authored changed lines.

  • .changeset/touch-selection-lock.md announces a [fix] for sliders, switches and segmented controls that no component is wired to here.
  • liftScale in packages/core/src/lift/liftGeometry.ts bounds only the width, so a trigger taller than the viewport lifts past the bottom gutter and surfacePlacement puts its menu above the top.

[Automated review]

Comment thread .changeset/touch-selection-lock.md Resolved
Comment thread docs/architecture/layer-runtime.md Resolved
Comment thread packages/core/src/gestures/README.md Resolved
vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026
# Conflicts:
#	packages/core/src/motion/README.md
vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026
vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026
vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026
# Conflicts:
#	packages/cli/assets/docs/motion.doc.mjs
#	packages/cli/test/__golden__/no-icon-set/docs-topics.json
#	packages/core/src/motion/roleKinds.generated.ts
vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026
vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approval is on hold for an engineering owner's spec decision: the branch adds architecture:layer-runtime INV13 and its Exit copy section, rewrites the gesture-tuning ruling in spec:AST-057, and ships a press-feedback exception to spec:AST-094 FR19 that its own README calls pending. An engineering owner can approve the head's records by commenting /approve-spec <full head SHA> on this pull request (the author may do so as an engineering owner), while the FR19 exception needs its amendment on main or should come out. Separately, please land the stacked dependencies and rebase so this pull request carries only the lift (31,587 authored changed lines today), and fix these:

  • createLayer in packages/core/src/lift/lift.ts gives the viewport-covering lift layer touch-action: none, so a pinch cannot zoom the page while a trigger is lifted (spec:AST-094 FR10, FR35); touch-action: pinch-zoom still keeps one-finger pans off.
  • liftScale in packages/core/src/lift/liftGeometry.ts bounds only the width, so a trigger taller than the viewport less its gutters lifts past the bottom gutter, and surfacePlacement then answers above with no room there.
  • The spec:AST-057 edit calls the row's motion roles theme motion roles and drops durations from its Not public list, which spec:AST-094 FR4 and DEC-11 do not allow yet.

[Automated review]

Comment thread docs/specs/AST-057-row-actions/spec.md Resolved
@cixzhang

Copy link
Copy Markdown
Contributor

Posted by Cindy's agent on her behalf.

A naming change to match the records: please rename the module and its exports from lift to liftFromTrigger, for example liftFromTrigger/ and createLiftFromTrigger (architecture:layer-runtime INV15 calls it the lift-from-trigger module). In spec:AST-094, "lift" means the finger leaving the screen, so a bare lift name would read as that. If this stays idle, we'll make the rename after about 30 minutes, keeping your authorship.

@vjeux

vjeux commented Oct 26, 2026

Copy link
Copy Markdown
Contributor Author

On it. This pull request's next push renames the module and its exports to match architecture:layer-runtime INV15:

  • packages/core/src/liftFromTrigger/
  • createLiftFromTrigger, useLiftFromTrigger and the exported types

ContextMenu's imports follow in #8201. That push also folds the new foundation heads.

vjeux added a commit to vjeux/astryx that referenced this pull request Oct 26, 2026
@vjeux

vjeux commented Oct 26, 2026

Copy link
Copy Markdown
Contributor Author

Renamed in 6004a1e: liftFromTrigger/, createLiftFromTrigger and the related types.

cixzhang and others added 8 commits November 10, 2026 21:43
…functions

A float64 hash of easedProgress's values is not the same on every
platform: macOS arm64 and CI's Linux give different hashes for main's
code. The test now keeps copies of easedProgress and
solveAnimatedPageScroll as they are on main (ebb9dde), and checks that
the shipped functions give the same number, bit for bit (Object.is), for
each position and velocity on the grid. From rest, the pager's velocity
at t <= 0 stays the one value left out.

Based on vjeux's test (vjeux/astryx coverage/rect-morph-first-instant)
liftCopy had no caller. It goes with its hide count per source, its
picture read, its corner read and their tests, fixture and README part.
It returns with the viewer that uses it.
No caller closed a surface through the lift: a menu closes with its own
exit. dismissSurface goes with its exit copy and the fields only it
read. The drop now waits for the source alone, and the fixture, the
browser proofs, the README and the changeset no longer describe the
close.
liftFromTriggerGeometry.ts served only the lift and one placement read.
Its three functions and their types move into liftFromTrigger.ts, and
their unit tests into liftFromTrigger.test.ts, unchanged.
useLiftFromTrigger wrapped createLiftFromTrigger for its one caller.
That caller creates the lift and ends it at unmount itself.
No caller grew a surface out of a press: ContextMenu fades its menu in
where it rests. presentSurface loses its press, the lift loses
surfaceStartWidth, and the proof of the growth from a press goes.
INV15 names createLiftFromTrigger and surfacePlacement, the module's
two entry points. With the surface close gone from the module, a
surface beside the lift keeps its own exit.

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs an engineering owner's decision, plus one fix. This pull request was reviewed on its own, because its Stack line names two tops (#8485 and #8080) and #8508 does not carry it.

Approval waits on an engineering owner deciding this branch's edit to architecture:layer-runtime INV15, which removes "the exit, played through the runtime's exit copy (INV14)" from the lift module, says a surface beside the lift keeps its own exit, and names createLiftFromTrigger and surfacePlacement as the entry points. An engineering owner records that decision by commenting /approve-spec <full head SHA> on this pull request (the author may, as an engineering owner); otherwise restore the exit-copy path or drop the INV15 edit.

Blocking:

  • Regression: since #9724 landed on main, packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts:44-45 conflicts with main, so GitHub cannot merge this pull request and CI cannot run; merge main and keep both imports.

Automated review (maintainer link)

Comment thread docs/architecture/layer-runtime.md Resolved
vjeux and others added 3 commits November 11, 2026 08:47
The lift marks its layer, its anchor box and its placeholder. The exit
copy's mark went with the exit copy. Comment only.
…ect-morph

# Conflicts:
#	packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs an engineering owner's decision, plus one fix. This pull request was reviewed on its own, because #9764, #9769, #8485 and #8201 carry different Stack lines.

Approval waits on an engineering owner deciding this branch's edit to architecture:layer-runtime INV15, which removes "the exit, played through the runtime's exit copy (INV14)" from the lift module, says a surface beside the lift keeps its own exit, and names createLiftFromTrigger and surfacePlacement as its entry points. An engineering owner records that decision by commenting /approve-spec <full head SHA> on this pull request (the author may, as an engineering owner); otherwise restore the exit-copy path or drop the INV15 edit.

Blocking:

  • Miss: a second drop() while the trigger is still returning settles the first dismissal and ends the lift at once (packages/core/src/liftFromTrigger/liftFromTrigger.ts:1020), so the trigger snaps to its place and the tint vanishes instead of the return continuing as spec:AST-094 FR6 requires; make a repeated drop() a no-op while one is in flight and add a unit case for it.

Automated review (maintainer link)

A second drop while the source returns must change nothing: the
source keeps returning on its spring, and the tint keeps clearing.
A drop while the source returned retargeted the return and settled
the first one at once: the source snapped to its place and the tint
vanished. The first drop now marks the lift as dropping, and a later
drop changes nothing until the lift ends.

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs an engineering owner's decision and one fix, and approval waits for CI. This pull request was reviewed on its own, since #9764, #9769, #8485 and #8201 carry different Stack lines; the repeated drop() snap from the last review is fixed.

Approval waits on an engineering owner deciding this branch's architecture:layer-runtime INV15 edit, which removes "the exit, played through the runtime's exit copy (INV14)" from the lift module, says a surface beside the lift keeps its own exit, and names createLiftFromTrigger and surfacePlacement as its entry points: an engineering owner records it by commenting /approve-spec <full head SHA> on this pull request (the author may, as an engineering owner), or else the exit-copy path comes back or the INV15 edit leaves the branch. CI has not run on this head because it is a draft from a fork: once it is ready for review, a maintainer adds the ci:run label so CI runs.

Blocking:

  • Newly live check: the description's Testing section still lists a liftCopy zoom proof ("In the lift fixture", step 3, and "The unit tests ... cover the same") that this head removed with liftCopy, so drop that step, point step 2's turn-around at the rect morph's own proof, and label the "At this head" checks that name an older commit as earlier.
Claims and files checked
  • Changeset lift-module.md: verified (three elements while lifted, translate, scale and opacity only, no shadow, not exported, no component uses it).
  • Changeset motion-morph-path.md: verified (lerpPath, edgeGrowPath, morphCurves; not exported).
  • Changeset motion-rect-morph.md: verified (createRectMorph on a copy or itself; easedProgress keeps its values; timedCurve for the pager and the morph).
  • Description, Change and API sections (createLiftFromTrigger, surfacePlacement and none, rect morph and ::backdrop tint, velocity on a drop, no shadow, touch-action: pinch-zoom, both gutters, caller-named marks, onLayerShown, pin, nothing imports the module, line counts): verified.
  • Description, Testing commit list and local-check counts (26 unit cases; 13 browser tests, 2 skipped in WebKit): match the code; not rerun here.
  • Description, "In the lift fixture" step 3 and "cover the same": false (no liftCopy or zoom proof at this head); step 2's turn-around: not proven in the lift fixture; checks labelled "At this head" that name an older commit: stale.
  • Description, performance numbers and app-level size: from an earlier head, not checkable here.
  • Files with a finding: docs/architecture/layer-runtime.md (INV15 edit on hold; gaps paragraph status, non-blocking), packages/core/src/liftFromTrigger/liftFromTrigger.ts (onLayerShown doc, non-blocking).
  • Files clear: .changeset/lift-module.md, .changeset/motion-morph-path.md, .changeset/motion-rect-morph.md, docs/architecture/react-component-runtime.md, docs/specs/AST-094-touch-presentations/spec.md, packages/core/src/Popover/popoverSurfaceMotion.ts, packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts, packages/core/src/liftFromTrigger/README.md, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.fixture.ts, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.interaction.browser.spec.ts, packages/core/src/liftFromTrigger/liftFromTrigger.test.ts, packages/core/src/motion/README.md, packages/core/src/motion/__tests__/callerMotionValues.test.ts, packages/core/src/motion/__tests__/motionBundle.ts, packages/core/src/motion/__tests__/rectMorph.interaction.browser.spec.ts, packages/core/src/motion/morphPath.test.ts, packages/core/src/motion/morphPath.ts, packages/core/src/motion/paging.test.ts, packages/core/src/motion/paging.ts, packages/core/src/motion/progress.test.ts, packages/core/src/motion/progress.ts, packages/core/src/motion/rectMorph.test.ts, packages/core/src/motion/rectMorph.ts, packages/core/src/motion/themeTiming.server.test.ts, packages/core/src/motion/themeTiming.test.ts, packages/core/src/motion/themeTiming.ts, scripts/snap-paging-bundle.test.mjs.

Automated review (maintainer link)

The layer runtime's gaps paragraph states INV15's gap as what a reader
can see: ContextMenu and Lightbox do not use the module. The
onLayerShown doc says it runs before the source's writes, and what the
page holds then. Records and comments only.

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs an engineering owner's decision on one record edit; no blocking defect remains. Approval waits on an engineering owner deciding this branch's architecture:layer-runtime INV15 edit, which removes “the exit, played through the runtime's exit copy (INV14)” from the lift module, says a surface beside the lift keeps its own exit, and names createLiftFromTrigger and surfacePlacement as its entry points: an engineering owner records it by commenting /approve-spec <full head SHA> on this pull request (the author may, as an engineering owner), or else the exit-copy path comes back or the INV15 edit leaves the branch. CI has not run on this head because it is a draft from a fork; once it is ready for review, a maintainer adds the ci:run label so CI runs. This pull request was reviewed on its own, since #9764, #9769, #8485 and #8201 carry different Stack lines.

Non-blocking:

  • Miss: the lift's README says its “document listeners” end with it, but the module listens on its own layer and the window's input stream, so “listeners” keeps it true.
  • Still open from the last review: onLayerShown's doc could say that presentSurface(), drop() and dispose() do nothing inside it, because the lift records its state after the callback.
  • Miss: two history items in the description (the zoom copy's marks, the menu growth proof) describe code this head no longer has, and the title's scope lift could read liftFromTrigger to match the module.
Claims and files checked
  • Changeset for the lift module: verified (the trigger lifts above a tint, a surface fades in on the rect morph, translate, scale and opacity only, three elements while lifted, no shadow, not exported, no component uses it).
  • Changeset for the morph paths: verified (straight and edge-growth paths, an interrupted motion continuing in each path's law, compositor curves; not exported; no component uses them).
  • Changeset for the rect morph: verified (present and dismiss on a copy or the element, a duration and easing or a spring, Reduced Motion left to the animator, the eased progress moved with identical values, the timed curve for the pager and the morph).
  • Description, invariants: one module apart from the hold and the exit copy, nothing at rest and at most five elements, a listed public-API gap: verified.
  • Description, change: the two entry points, the trigger and the tint on the animator, a drop keeping velocity, no shadow, pinch-zoom kept, both gutters, a none placement, caller-named marks, onLayerShown before the source's writes, pin restored, nothing importing the module: verified.
  • Description, this round's commit list and notes: verified against history and the head.
  • Description, local check counts (26 unit cases; 13 browser tests, 2 skipped in WebKit): match the files; not rerun here.
  • Description, fixture steps 1 to 6 and unit coverage: verified.
  • Description, line counts over the rect-morph pull request: verified exactly.
  • Description, scope checkboxes: no public API, verified; no product behavior change, verified for the lift (the included rect-morph commits change the pager only for a retarget in a move's first instant and a 0 ms move).
  • Description, history sections, performance numbers and app size: from earlier heads, not checkable here; two history items describe code since removed.
  • Files with a note: docs/architecture/layer-runtime.md (INV15 edit awaiting a decision), packages/core/src/liftFromTrigger/liftFromTrigger.ts (onLayerShown doc), packages/core/src/liftFromTrigger/README.md (listener wording).
  • Files clear: .changeset/lift-module.md, .changeset/motion-morph-path.md, .changeset/motion-rect-morph.md, docs/architecture/react-component-runtime.md, docs/specs/AST-094-touch-presentations/spec.md, packages/core/src/Popover/popoverSurfaceMotion.ts, packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.fixture.ts, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.interaction.browser.spec.ts, packages/core/src/liftFromTrigger/liftFromTrigger.test.ts, packages/core/src/motion/README.md, packages/core/src/motion/__tests__/callerMotionValues.test.ts, packages/core/src/motion/__tests__/motionBundle.ts, packages/core/src/motion/__tests__/rectMorph.interaction.browser.spec.ts, packages/core/src/motion/morphPath.test.ts, packages/core/src/motion/morphPath.ts, packages/core/src/motion/paging.test.ts, packages/core/src/motion/paging.ts, packages/core/src/motion/progress.test.ts, packages/core/src/motion/progress.ts, packages/core/src/motion/rectMorph.test.ts, packages/core/src/motion/rectMorph.ts, packages/core/src/motion/themeTiming.server.test.ts, packages/core/src/motion/themeTiming.test.ts, packages/core/src/motion/themeTiming.ts, scripts/snap-paging-bundle.test.mjs.

Automated review (maintainer link)

The lift adds no document listener: it listens on its own layer and
on the window's input stream, so the README says its listeners end
with it. The onLayerShown doc says presentSurface, drop and dispose
do nothing inside it, since the lift records its state once it
returns. Comments and docs only.

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs an engineering owner's decision on one record edit, plus one fix. Approval waits on an engineering owner deciding this branch's architecture:layer-runtime INV15 edit, which removes “the exit, played through the runtime's exit copy (INV14)” from the lift module, says a surface beside the lift keeps its own exit, and names createLiftFromTrigger and surfacePlacement as its entry points: an engineering owner records that decision by commenting /approve-spec <full head SHA> on this pull request (the author may, as an engineering owner), or else the exit-copy path comes back or the INV15 edit leaves the branch. CI has not run on this head because it is a draft from a fork; once it is ready for review, a maintainer adds the ci:run label so CI runs. This pull request was reviewed on its own, since #9764, #9769, #8485 and #8201 carry different Stack lines.

Blocking:

  • Miss: lifting the last row of a list rounded by li:last-child squares its bottom corners for as long as it is lifted, because the placeholder becomes the row's next sibling and KEPT_PROPERTIES in liftFromTrigger.ts pins no border radius; pinning the four border-*-radius longhands as laid out, with a look proof that lifts the fixture's last row, fixes it.
Claims and files checked
  • Changeset for the lift module: verified (the trigger lifts above a tint, a surface fades in on the rect morph, translate, scale and opacity only, nothing at rest, three elements while lifted, no shadow, not exported, no component uses it).
  • Changeset for the morph paths: verified (straight and edge-growth paths, an interrupted motion continuing in each path's law, compositor curves; not exported; no component uses them).
  • Changeset for the rect morph: verified (present and dismiss on a copy or the element, a duration and easing or a spring, Reduced Motion left to the animator, the eased progress moved with identical values, the timed curve for the pager and the morph).
  • Description, invariants (one module apart from the hold and the exit copy, nothing at rest and at most five elements, a listed public-API gap): verified.
  • Description, change (the two entry points, the trigger and the tint on the animator, a drop keeping velocity, no shadow, pinch-zoom kept, both gutters, a none placement, caller-named marks, onLayerShown before the source's writes, pin restored, nothing importing the module): verified; the lifted trigger keeps its look except where that look depends on being the last child (blocking finding).
  • Description, this round's commit list and notes: verified against the history and the head.
  • Description, local check counts (26 unit cases; 13 browser tests, 2 skipped in WebKit): match the files; not rerun here.
  • Description, fixture steps 1 to 6 and unit coverage: verified.
  • Description, line counts over the rect-morph pull request: verified exactly.
  • Description, scope checkboxes: no public API, verified; no product behavior change, verified for this pull request's own commits (the included rect-morph commits change the pager only for a retarget in a move's first instant).
  • Description, history sections, performance numbers and app size: from earlier heads, not checkable here.
  • Files with a finding: docs/architecture/layer-runtime.md (INV15 edit awaiting a decision), packages/core/src/liftFromTrigger/liftFromTrigger.ts (last-row radius), packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.interaction.browser.spec.ts (the look proof lifts only the middle row).
  • Files clear: .changeset/lift-module.md, .changeset/motion-morph-path.md, .changeset/motion-rect-morph.md, docs/architecture/react-component-runtime.md, docs/specs/AST-094-touch-presentations/spec.md, packages/core/src/Popover/popoverSurfaceMotion.ts, packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts, packages/core/src/liftFromTrigger/README.md, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.fixture.ts, packages/core/src/liftFromTrigger/liftFromTrigger.test.ts, packages/core/src/motion/README.md, packages/core/src/motion/__tests__/callerMotionValues.test.ts, packages/core/src/motion/__tests__/motionBundle.ts, packages/core/src/motion/__tests__/rectMorph.interaction.browser.spec.ts, packages/core/src/motion/morphPath.test.ts, packages/core/src/motion/morphPath.ts, packages/core/src/motion/paging.test.ts, packages/core/src/motion/paging.ts, packages/core/src/motion/progress.test.ts, packages/core/src/motion/progress.ts, packages/core/src/motion/rectMorph.test.ts, packages/core/src/motion/rectMorph.ts, packages/core/src/motion/themeTiming.server.test.ts, packages/core/src/motion/themeTiming.test.ts, packages/core/src/motion/themeTiming.ts, scripts/snap-paging-bundle.test.mjs.

Automated review (maintainer link)

The fixture rounds its last row through li:last-child. Lifted, that
row must keep its corners, though the placeholder after it ends the
last-child match.
The placeholder after the lifted source ends a :last-child match, so a
grouped list's last row rose with square bottom corners. The lift now
keeps the four corner radii as laid out, with the borders and padding
it already keeps.

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs an engineering owner's decision on one record edit; the last-row radius fix holds and no blocking defect remains.

Approval waits on an engineering owner deciding this branch's architecture:layer-runtime INV15 edit, which removes “the exit, played through the runtime's exit copy (INV14)” from the lift module, says a surface beside the lift keeps its own exit, and names createLiftFromTrigger and surfacePlacement as its entry points: an engineering owner records that decision by commenting /approve-spec <full head SHA> on this pull request (the author may, as an engineering owner), or else the exit-copy path comes back or the INV15 edit leaves the branch. CI has not run on this head because it is a draft from a fork; once it is ready for review, a maintainer adds the ci:run label so CI runs. This pull request was reviewed on its own, since #9764, #9769, #8485 and #8201 carry different Stack lines.

Claims and files checked
  • Changeset for the lift module: verified (the trigger lifts above a tint, a surface fades in on the rect morph, translate, scale and opacity only, nothing at rest, three elements while lifted, no shadow, not exported, no component uses it).
  • Changeset for the morph paths: verified (straight and edge-growth paths, an interrupted motion continuing in each path's law, compositor curves; not exported; no component uses them).
  • Changeset for the rect morph: verified (present and dismiss on a copy or the element, a duration and easing or a spring, Reduced Motion left to the animator, the eased progress moved with identical values, the timed curve for the pager and the morph).
  • Description, invariants (one module apart from the hold and the exit copy, nothing at rest and at most five elements, a listed public-API gap): verified.
  • Description, change (the two entry points, the trigger and the tint on the animator, a drop keeping velocity, no shadow, pinch-zoom kept, both gutters, a none placement, caller-named marks, onLayerShown before the source's writes, pin restored, nothing importing the module): verified.
  • Description, this round's commit list, including the red-first last-row case and the corner fix: verified against the history and the head.
  • Description, local check counts (26 unit cases; 14 browser tests, 2 skipped in WebKit): match the files; not rerun here.
  • Description, fixture steps 1 to 7 and unit coverage: verified.
  • Description, line counts over the rect-morph pull request: verified exactly.
  • Description, scope checkboxes: no public API, verified; no product behavior change, verified for this pull request's own commits (the included rect-morph commits change the pager only for a retarget in a move's first instant).
  • Description, history sections, performance numbers and app size: from earlier heads, not checkable here.
  • File with a finding: docs/architecture/layer-runtime.md (INV15 edit awaiting a decision).
  • Files clear: .changeset/lift-module.md, .changeset/motion-morph-path.md, .changeset/motion-rect-morph.md, docs/architecture/react-component-runtime.md, docs/specs/AST-094-touch-presentations/spec.md, packages/core/src/Popover/popoverSurfaceMotion.ts, packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts, packages/core/src/liftFromTrigger/README.md, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.fixture.ts, packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.interaction.browser.spec.ts, packages/core/src/liftFromTrigger/liftFromTrigger.test.ts, packages/core/src/liftFromTrigger/liftFromTrigger.ts, packages/core/src/motion/README.md, packages/core/src/motion/__tests__/callerMotionValues.test.ts, packages/core/src/motion/__tests__/motionBundle.ts, packages/core/src/motion/__tests__/rectMorph.interaction.browser.spec.ts, packages/core/src/motion/morphPath.test.ts, packages/core/src/motion/morphPath.ts, packages/core/src/motion/paging.test.ts, packages/core/src/motion/paging.ts, packages/core/src/motion/progress.test.ts, packages/core/src/motion/progress.ts, packages/core/src/motion/rectMorph.test.ts, packages/core/src/motion/rectMorph.ts, packages/core/src/motion/themeTiming.server.test.ts, packages/core/src/motion/themeTiming.test.ts, packages/core/src/motion/themeTiming.ts, scripts/snap-paging-bundle.test.mjs.

Automated review (maintainer link)

INV15 names the lift module's four functions and why one module
serves every surface that lifts. Records only.
@cixzhang

Copy link
Copy Markdown
Contributor

/approve-spec aa2f8e5

@cixzhang

Copy link
Copy Markdown
Contributor

Decision: D-20261112-210

@astracat-bot astracat-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved pending CI: CI has not run on this head yet. This is still a draft from a fork, so once it is marked ready for review, a maintainer adds the ci:run label so CI runs.

The lift-from-trigger module follows the layer runtime's one-module rule, and the rewritten INV15 (a surface beside the lift keeps its own exit; the module's four functions) carries an engineering owner's spec approval for this head. The earlier findings on the repeated drop, the last row's corners and the merge conflicts with main are fixed, and no blocking finding remains.

Non-blocking:

  • This pull request was reviewed on its own: its Stack line differs from the one #9764 and #9769 carry, and this branch does not contain their current heads, so merging #9769's head and one identical Stack line in all four bodies would keep the branches from drifting.
Claims and files checked
  • Changeset (lift): one opt-in lift for touch surfaces; the trigger rises above a tint, a surface fades in beside it on the shared rect morph, and the trigger returns. Verified: liftFromTrigger.ts lift, presentSurface, drop.
  • Changeset (lift): translate, scale and opacity only; nothing at rest, three elements while lifted, no shadow. Verified: the motion-properties and element-count browser cases and the unit count; the tint's colour is a zero-length hold, not a motion.
  • Changeset (lift, paths, morph): not exported from the package entry point, no component uses it yet. Verified: Core's exports and index are unchanged and nothing imports the modules outside their tests.
  • Changeset (paths): the straight line and the popover's edge growth give the rect at each progress and continue an interrupted motion in their own law; morphCurves makes compositor curves. Verified: morphPath.ts and the popover turn-around unit case.
  • Changeset (morph): present along a path, dismiss on a copy or on itself, each continuing the other; timing is a duration and easing or a spring from a given velocity; Reduced Motion is the animator's. Verified: rectMorph.ts and its unit cases.
  • Changeset (morph): easedProgress moves with the same values; timedCurve keeps a retargeted velocity for the pager and the morph. Verified: the bit-identical comparison test, paging.ts, rectMorph.ts.
  • Description: the invariants it serves (one lift module; minimal DOM with at most five more elements while open; internal use of a listed gap). Verified against the current records.
  • Description: the trigger rides the rect morph and the tint the layer's backdrop channel; a drop while either rises keeps its velocity. Verified: createRectMorph(source), the backdrop animator, the tint turn-around browser case.
  • Description: no shadow, page pinch-zoom kept, the lift fits between the gutters on both axes, placement can answer none. Verified: the layer's touch-action, liftFromTriggerScale, surfacePlacement and their tests.
  • Description: the caller names the marks, can show its surface once the layer shows, and can pin styles. Verified: the required attribute option, the onLayerShown call before the source's writes, the pin option and their unit cases.
  • Description: nothing imports the module; no public API or product behavior changes. Verified (the pager's change is nil on its default curve).
  • Description: this round's commit list. Verified: every listed commit is present with a matching subject; the last one changes only the INV15 record.
  • Description: unit and browser case counts (26; 14 in Chromium, 12 plus 2 skipped in WebKit) and the line counts over the rect morph branch. Verified by reading the suites and the numstat; the runs themselves were not repeated.
  • Description: fixture steps 1 to 7 and the unit coverage list. Verified: each step has its browser case; the unit file covers each listed area.
  • Description: performance numbers, app-size numbers and history from earlier heads and other pull requests. Not checkable at this head; no claim of this head depends on them.
  • .changeset/lift-module.md: clear; claims verified
  • .changeset/motion-morph-path.md: clear; claims verified
  • .changeset/motion-rect-morph.md: clear; claims verified
  • docs/architecture/layer-runtime.md: clear; INV15 rewrite and gaps sentence approved for this head, consistent with the module
  • docs/architecture/react-component-runtime.md: clear; the gap note states the glide still plays its own translate, which it does
  • docs/specs/AST-094-touch-presentations/spec.md: clear; the new verification path exists
  • packages/core/src/Popover/popoverSurfaceMotion.ts: clear; import moved, two internal exports for the morph proofs
  • packages/core/src/SegmentedControl/useSegmentedSelectionMotion.ts: clear; import moved, values unchanged
  • packages/core/src/liftFromTrigger/README.md: clear; matches the module (three elements, four with the shield, no shadow, listeners end with it)
  • packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.fixture.ts: clear
  • packages/core/src/liftFromTrigger/__tests__/liftFromTrigger.interaction.browser.spec.ts: clear; 14 cases, two skipped in WebKit
  • packages/core/src/liftFromTrigger/liftFromTrigger.test.ts: clear; 26 cases
  • packages/core/src/liftFromTrigger/liftFromTrigger.ts: clear
  • packages/core/src/motion/README.md: clear
  • packages/core/src/motion/__tests__/callerMotionValues.test.ts: clear; compares with copies of the earlier functions
  • packages/core/src/motion/__tests__/motionBundle.ts: clear
  • packages/core/src/motion/__tests__/rectMorph.interaction.browser.spec.ts: clear
  • packages/core/src/motion/morphPath.test.ts: clear
  • packages/core/src/motion/morphPath.ts: clear
  • packages/core/src/motion/paging.test.ts: clear
  • packages/core/src/motion/paging.ts: clear; plays timedCurve
  • packages/core/src/motion/progress.test.ts: clear
  • packages/core/src/motion/progress.ts: clear
  • packages/core/src/motion/rectMorph.test.ts: clear
  • packages/core/src/motion/rectMorph.ts: clear
  • packages/core/src/motion/themeTiming.server.test.ts: clear
  • packages/core/src/motion/themeTiming.test.ts: clear; its cases moved to progress.test.ts
  • packages/core/src/motion/themeTiming.ts: clear; easedProgress moved out, still imported for its own use
  • scripts/snap-paging-bundle.test.mjs: clear; the pager bundle now includes the progress module

Automated review (maintainer link)

This branch was successfully deployed

1 active deployment
Preview — aa2f8e51 Deployed Nov 12, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Meta Open Source bot. needs:design-review Affects visuals — Design should review

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants