[imdreamrunner]

Prototype for evaluation. A native-first alternative to #7706. Both move the native <dialog> from BottomSheetSwitcher to each child BottomSheet; this one then asks the browser to do everything a modal dialog can do and keeps custom code only where it measurably cannot. Not a proposal to merge as-is; see "Recommendation".

Primary intent

Find out how far <dialog> and CSS carry a bottom sheet — standalone and as a step of a BottomSheetSwitcher — when every sheet owns its own dialog, and measure exactly where custom JavaScript must stay. The public API is unchanged; the mechanisms behind page scroll lock, focus containment, Escape and the platform close, the scrim tap, focus return, and the scrim change, and three observable details move with the structure.

The other templates do not fit: nothing is broken (bug-fix), no capability is added (new-feature), and pixels are unchanged at rest (visual-update).

User or maintainer impact

Consumers see the same flows: one steady scrim, the step handoff, focus in the new step and back on the opener, page scroll locked while a sheet is modal, Escape / scrim / swipe by purpose. Maintainers lose two mechanisms (useFocusTrap and useScrollLock in sheets) and gain one 68-line CSS module. The three deltas a consumer can observe are listed under "Behaviour deltas".

Problem and solution fit

Today every sheet dialog is presented with showModal() and then re-implemented in script: a focus trap on top of the browser's inert page, a body pin on top of the top layer, a captured trigger on top of the browser's own focus record. #7706 kept all of that when it moved the dialog to each sheet. This PR removes what the browser already does and keeps the rest, so the remaining script is the exact list of things a per-sheet dialog cannot know on its own: that it is one step of a flow.

How far native goes

Mechanism	Today's custom code (main)	Native / CSS replacement tried	Chromium result	WebKit result	Kept custom? Why
Page scroll lock	useScrollLock: body { position: fixed; top: -scrollY; overflow: hidden }, a lock counter, a measured scrollbar-gutter hold, scrollTo on release	One rule, :root:has(dialog[data-sheet-dialog]:modal) { overflow: hidden; scrollbar-gutter: stable }, installed once per document from a layout effect (the way Layer installs its @position-try rules; StyleX cannot style the root)	Holds: at scrollY 300, wheel 400 → 300, real finger drag 300px → 300, wheel/drag inside the sheet content → 300 (no chaining), through a handoff with two open dialogs → 300, under another showModal() dialog → locked, released only after the exit motion; scrollY never moves so nothing to restore; viewport width 390→390	Same for wheel in every state. Touch cannot be injected into Playwright WebKit, and desktop WebKit is not iOS Safari: whether iOS honours root overflow: hidden for a finger is unverified	No custom code. Caveats: iOS touch unverified; a page that scrolls a nested element rather than the viewport is no longer pinned (main pinned <body>); scrollbar-gutter: stable would open a gutter on a classic-scrollbar desktop page that had no scrollbar (headless Chromium has overlay scrollbars, width 0, so unmeasured)
Focus containment	useFocusTrap on the switcher (keydown Tab wrap + focusin redirect)	showModal() makes everything outside the dialog inert	Tab ×10 inside a step never reaches the page or the retained dialog; the cycle has one extra stop (browser UI / nothing) per lap where the JS trap wrapped directly	Same; two extra stops per lap (browser UI, then the dialog element)	No custom code. Same behaviour as Core Dialog, which already relies on it. The retained dialog is also natively inert (a lower modal dialog is inert while a higher one is open)
Escape and platform close	cancel handler + the shared dismissal stack's document Escape listener + the per-dialog keydown re-route	Native cancel (Escape in a modal dialog, Android back, the close watcher); preventDefault() always, then dismiss only if purpose !== 'required' and the shared stack says this layer is topmost	cancel arrives natively; the stack gates it; required blocks it	Same	Kept (unchanged from main, now on each sheet's dialog): purpose gating and the overlay-dismissal family contract are not expressible natively; non-modal show() dialogs get no cancel at all, so the stack's Escape listener stays for them
Scrim tap	onClick on the dialog where target === currentTarget	The native backdrop click already targets the dialog element, so the same handler is the native path; closedby="any" was evaluated	closedBy exists in Chromium; cancel carries no source, so it cannot distinguish a backdrop tap (info only) from Escape (info and form)	closedBy absent in WebKit ('closedBy' in HTMLDialogElement.prototype → false)	Kept the click handler (3 lines, now scoped to the active sheet's dialog). closedby rejected: no WebKit support and no per-source purpose gating
Focus return on close	Switcher: capture document.activeElement at first present, .focus() it after the last dialog closes. Standalone: capture + focusForLastInput	close() on a modal dialog focuses the element focused at showModal()	A one-sheet flow and a standalone sheet return natively (the script call is then a no-op). A multi-step flow does not: the last step's dialog recorded a control inside the previous step, which is hidden when the flow ends, so the browser's return lands nowhere. Measured: opener focused after Escape / scrim tap / swipe from step 2 on both builds	Same	Kept the flow-level capture + return (switcher), with preventScroll so an off-screen opener does not pull the page back (main's body pin masked that scroll). Standalone unchanged
Scrim	One ::backdrop on the one dialog; gesture writes --_sheet-scrim-opacity	One ::backdrop per dialog. Tried letting both paint: the entering dialog's backdrop sits above the retained panel in the top layer (dims the previous step, double-dims the page); a cross-fade dips to 0.44 mid-handoff	With the retained dialog owning the scrim and the entering dialog's backdrop at 0 until the retained one closes, then taking the value with its transition pinned off for one style flush: dim 0.5 on all 70+ frames of every switch, dipFrames=0 overFrames=0, page pixel luminance 127 mid-handoff and settled, settled screenshots pixel-identical to main	Same	Kept (same shape as #7706, ~40 lines). The browser offers no handoff between two backdrops; CSS alone either double-dims or dips. transition-duration: 0s must be an inline custom property because the stylesheet audit forbids it in a rule
Layer dismissal ordering	One useLayerDismissal registration for the flow	Nothing native orders a sheet against a Popover or Dialog opened inside it	Registration points at the active sheet's dialog; a showModal() dialog opened during a handoff hit-tests on top; Escape with a nested Popover closes the Popover first (existing browser specs pass)	Same	Kept (unchanged): the family contract requires it, and it is the only thing that knows about non-dialog layers
Top-layer order on reversal	n/a (one dialog)	None: the top layer orders by showModal() call, and a modal dialog beneath another modal dialog is inert	Going Back to a sheet still fading beneath the current one: close() + showModal() of the returning sheet (two extra native calls), dim 0.5 throughout, focus on the returning panel	Same	Kept (same as #7706): a native constraint, reached only by reversing within ~400ms
Focus during a handoff	n/a (one dialog; focusPanel moves focus)	showModal() of the next step records the Back/Continue control that presented it; close() of the retained dialog later tries to focus it	When the control is in a hidden step the return is a no-op; when it is in the now-active step (Back) the browser would move focus from the panel to that control a fade later. closeKeepingFocus (8 lines) re-focuses whatever was focused before the close	Same	Kept: smaller than #7706's blur-before-present and robust to how an engine implements the return

Replaced outright: useScrollLock (sheets) and useFocusTrap (switcher), plus the blur() workaround of #7706. The deprecated read-only hasActiveFocusTrapEscape() signal keeps reporting a modal flow (useFocusTrapEscapeCompatibilitySignal, one line) so the released compatibility contract holds.

Before / after structure

Before (main): one switcher-owned <dialog>; steps are panels inside it.

<dialog aria-label="Active step" aria-modal="true">   <!-- switcher: showModal() once; body pinned by script -->
  <div hidden …><div class="astryx-bottom-sheet">step A</div></div>
  <div><div class="astryx-bottom-sheet">step B</div></div>
</dialog>

After: the switcher renders no element; each child renders what a standalone sheet renders, and a <style id="astryx-bottom-sheet-scroll-lock"> in <head> holds the one lock rule.

<dialog data-sheet-dialog aria-label="Step A" inert aria-hidden="true">  <!-- retained: open until its panel fades -->
  <div inert aria-hidden="true"><div class="astryx-bottom-sheet">step A</div></div>
</dialog>
<dialog data-sheet-dialog aria-label="Step B" aria-modal="true">          <!-- active: showModal() on entry -->
  <div><div class="astryx-bottom-sheet">step B</div></div>
</dialog>

Behaviour deltas

Area	main	This PR	Note
Native calls per switch	0	showModal(next) at click, close(previous) ~650ms later	measured, both engines
Open dialogs during a handoff	1	2 (retained one inert, aria-hidden); 1 at rest
Page dimming through a switch	0.5 constant	0.5 constant, no dip, no double dim	70+ samples per switch
Scroll lock mechanism	inline styles on <body>, scrollTo on release	root overflow: hidden + scrollbar-gutter: stable while any sheet dialog is modal	scrollY never changes; <body> untouched
Page that scrolls a nested element, not the viewport	pinned (body position: fixed)	not pinned	observable change; the sheet's own content still contains its scroll
Page scroll on iOS by touch	pinned	root overflow: hidden	unverified on hardware
Tab inside a modal sheet	wraps directly within the dialog	native containment: one (Chromium) or two (WebKit) browser-UI stops per lap, never the page	same as Core Dialog
Focus after a switch (modal)	new step's panel	new step's panel	identical traces
Focus after a switch (non-modal)	lost to <body>	new step's panel	improvement (native show() focusing steps)
:focus-visible on the focused panel after a pointer switch	false	false in Chromium; true in WebKit	the panel has outline: none, so nothing paints; next Tab behaves the same
Focus return on close	opener	opener	identical; now with preventScroll
Escape / scrim tap / swipe	close via onActiveSheetChange(null) per purpose	same	identical
Rapid reversal (Back while fading)	panel z-index swap	close() + showModal() of the returning sheet	2 extra native calls; no dip, no focus jump
ref	one persistent dialog	the dialog hosting the active sheet; moves on handoff; null while closed	observable change
onCancel, dialog props	one dialog	every sheet dialog (handlers reach the active one)	observable change
Accessible name / role	precedence rule across children	each dialog named by its own sheet; consumer aria-label still wins	equivalent output
SSR markup	one closed <dialog>	one closed <dialog> per sheet; the lock stylesheet is installed client-side before the first showModal()	hydration-safe (no open server-side)
A Core Dialog opened from a step	above the flow	above the flow; lock held under it and after it closes	top layer is showModal() order

Non-modal flows (hasScrim={false}) still use show() per sheet with z-index 1000/1001; nothing is locked or contained, as on main. This stays inside the non-modal path that AST-027 already lists as a conformance gap.

Comparison with #7706

#7706 workaround	Here
JS focus trap kept in the switcher	Gone — native containment
JS scroll lock kept in the switcher and standalone sheet	Gone — one CSS rule
blur() of the presenting control before showModal()	Gone — replaced by closeKeepingFocus on the retained dialog's close (8 lines, undoes a wrong native return instead of pre-empting it)
Scrim handed to the entering dialog without a fade	Remains — no native handoff between two backdrops
Re-present a sheet that is still fading on reversal	Remains — top-layer order and inertness are native constraints
Flow-level trigger capture and focus return	Remains — a multi-step flow's last dialog recorded the wrong element

Test edits

Existing suites ran unedited first against the new code: 6 failures, all assertions about removed internals or the shared-dialog structure. No behaviour assertion was weakened.

Test	Edit	Reason
BottomSheet.test.tsx — dialog stubs	showModal/close stubs also record a modalDialogs set	jsdom cannot match :modal; the set stands in for it
BottomSheet.test.tsx "keeps a standalone modal sheet presented through its exit animation"	body.style.position === 'fixed' → isScrollLockEngaged() (a modal dialog[data-sheet-dialog][open] exists)	asserted the removed body pin; the behaviour (lock held through exit) is asserted through the rule's precondition
BottomSheet.test.tsx "keeps a standalone non-modal sheet visible until its exit ends"	body.style.position → isScrollLockEngaged() === false	same
BottomSheetSwitcher.test.tsx helper getSharedDialog()	renamed getActiveDialog(); added getSheetDialog, openDialogs, isScrollLockEngaged, recordDialogCalls (whose close returns focus like a browser)	the role query resolves to the active sheet's dialog (the retained one is aria-hidden)
"keeps the previous sheet stationary until the new entrance finishes, then fades it"	dialog[open] count 1 → [details, confirm] during the handoff, [confirm] after; shared identity → active identity; retained dialog asserted inert	two dialogs are open through a handoff by design
"replaces an unfinished outgoing sheet during rapid navigation"	count 1 → 2; asserts a third showModal()	the returning sheet re-presents
"ignores late scrim updates from an outgoing sheet gesture"	scrim variable read from the retained (owner) dialog; entering dialog asserted 0	scrim ownership moved to the retained dialog
"dismisses the flow from the one shared scrim", "releases the shared modal layer…", "can coordinate a non-modal flow…"	body.style.position → isScrollLockEngaged(); titles	removed body pin
"keeps focus in a modal sheet that has no tabbable controls"	now "focuses the panel … and leaves Tab to the native modal dialog": asserts the panel is focused, the dialog was presented with showModal() and is aria-modal, and that no script prevents Tab	the old assertion (keyDown Tab returns false) tested the JS trap's preventDefault, which jsdom was the only place to observe; the behaviour moves to BottomSheetNativeModal.a11y.browser.spec.ts ("Tab stays inside the active sheet"), which passes in Chromium and WebKit
"forwards … to the shared dialog", "keeps the shared dialog inline…"	titles and helper name only	wording

New tests: sheetScrollLock.test.ts (3: rule text, idempotent per-document install, iframe copy); BottomSheet.test.tsx "native modal host" (4: attribute + stylesheet installed once before presenting with <body> untouched; lock held under a Core Dialog and released only with the sheet; native cancel by purpose with preventDefault always; backdrop click by purpose); BottomSheetSwitcher.test.tsx "one native dialog per sheet" (13: per-sheet structure; native call order; lock across a handoff and through the exit; lock under a Core Dialog opened from a step with the Dialog presented on top; ref movement; dialog-prop forwarding; one scrim with instant takeover; focus into the entering sheet with the retained dialog inert; focus kept when a retained dialog's close returns focus natively; cancel by the active sheet's purpose after a handoff; cancel/backdrop click on the active dialog only; non-modal show() per sheet; re-present on reversal); __tests__/BottomSheetNativeModal.a11y.browser.spec.ts (5, Chromium + WebKit: CSS lock through a handoff, under another modal dialog, released after exit, with <body> untouched and no chaining; non-modal locks nothing; Tab containment; Escape + backdrop tap + focus return; steady scrim). Against the main build the lock test fails as expected (root overflow stays visible).

Records touched

BottomSheetSwitcher.spec.md and BottomSheet.spec.md are draft records. They were edited only where they describe the host and the mechanisms behind it (Intent, Ownership, FR2, FR5, FR8, FR9, representative states, PR1, design relationships, theming-anatomy key Shared dialog → Sheet dialogs, verified_by). No decision was added; the decision logs are unchanged.

Writing rubric results

• Present tense, describes the system: yes.
• No proposals, history, pull requests, or research inside the records: yes (this PR body carries them).
• Authority unchanged: draft stays draft; no claim was promoted.
• Ownership boundary still names a single owner per fact: yes (dialog lifecycle, scrim ownership, flow focus return → switcher; the dialog element and the scroll-lock rule → BottomSheet; panel anatomy → BottomSheet).
• Verification map names a real test per contract: yes (BottomSheetSwitcher.test.tsx, BottomSheetNativeModal.a11y.browser.spec.ts, sheetScrollLock.test.ts).

Authority

• component:BottomSheetSwitcher (draft) FR1–FR9, PR1, AR1–AR3 — preserved; FR5 and PR1 now name native containment and the CSS lock.
• family:overlay-dismissal — one registration for the flow; native cancel routed through shouldDismissOnCloseRequest(); existing Layer suites pass unedited.
• architecture:layer-runtime and spec:AST-027 — native showModal() hosting; the non-modal show() + z-index path remains the recorded gap.
• spec:AST-002 / architecture:public-component-api — props, defaults, and exports unchanged; ref semantics documented in the prop doc and the Changeset.

Change and evidence

Files (all under packages/core/src/BottomSheet/ plus the Changeset): BottomSheet.tsx (624 → 804 lines: the switcher item now renders its own dialog, plus closeKeepingFocus, the lock install, and the attribute), BottomSheetSwitcher.tsx (660 → 730 lines: no element, no trap, no scroll lock, but the per-sheet coordination — presentation order, scrim ownership, dialog-prop forwarding — moved in; 8 lines fewer than #7706's switcher), BottomSheetSwitcherContext.ts, new sheetScrollLock.ts (68 lines, the only "shared CSS plumbing"; it is sheet-scoped by attribute and lives in the BottomSheet directory, so no file outside it changed), sheetScrollLock.test.ts, __tests__/BottomSheetNativeModal.a11y.browser.spec.ts, the two .doc.mjs, the two draft specs, the two test files, and a [feat] Changeset.

API audit. Core built on origin/main and on this branch; .d.ts files and the exports map diffed. exports map identical. BottomSheet/index.d.ts and the root index.d.ts byte-identical. BottomSheet.d.ts and BottomSheetSwitcher.d.ts differ only in doc comments and the destructuring pattern in the declared signature (no type change). BottomSheetSwitcherContext.d.ts changes and sheetScrollLock.d.ts is new; neither is reachable from a public entry.

Browser (Chromium 149 and WebKit "Safari 26.5" under Playwright 1.61, 390×844, touch, static Storybook of main and of this branch, stories hug-content, capped, no-scrim, and the standalone showcase). Native call log, per-frame backdrop sampling, focus traces, hit tests, Tab traces, and for the lock: page made 3000px tall and scrolled to 300, flow opened from an off-screen opener, then wheel (both engines) and real finger drag (Chromium, CDP) over the backdrop and inside the sheet content, through a handoff, under a probe showModal() dialog, and after close.

• Lock: scrollY 300 before, during (wheel, drag, chaining, mid-handoff, under a probe dialog), and after; root overflow hidden exactly while a sheet dialog is modal, including the exit motion; <body> carries no inline style; viewport width 390→390. main: body pinned, scrollY reads 0, restored to 300 by script on release.
• Scrim through every switch, both directions: dim 0.5 [A:1.00] → 0.5 [A:1.00 B:0.00] → 0.5 [B:1.00]; dipFrames=0 overFrames=0; luminance 127 mid and settled. main: 0.5 constant.
• Focus: panel → panel on modal switches; opener after Escape, scrim tap, swipe; non-modal switches land on the new panel (body on main).
• Tab ×10 in a modal step never reaches the page or the retained dialog in either engine.
• Escape, scrim tap, swipe on step 2 close the flow; the scrim tap is a no-op in no-scrim.
• Rapid reversal: 4 native calls, dim 0.5 throughout, focus on the returning panel.
• Probe showModal() dialog during a handoff hit-tests on top in both engines.
• No horizontal overflow; no console errors or warnings on either build in either engine.
• Settled screenshots pixel-identical between builds in both engines (24 of 24 settled pairs); mid-handoff frames differ by 0–5% of pixels from animation timing.
• Pre-existing, not a delta: a process-cold WebKit leaves the first backdrop's entrance transition at 0.94–0.97 opacity when the story's play function opens the flow at page load — on main and on this branch alike (6/6 each), never on a warm engine or a reopen.

WebKit under Playwright approximates Safari but not iOS hardware, and cannot be driven by native touch; the iOS touch-scroll answer is still open.

Recommendation

Native-first is viable: the browser carries focus containment, inertness, the per-dialog focus record, cancel, the backdrop, and — with one CSS rule — page scroll lock, and every measurement matches main or improves on it. The smallest custom set that must stay in a per-sheet design is: (1) the flow-level opener capture and return, (2) the scrim handoff between two backdrops, (3) closeKeepingFocus on a retained dialog's close, (4) the re-present on reversal, and (5) purpose gating of cancel plus the shared dismissal stack — items 1–4 exist only because two dialogs stand in for one sheet; item 5 exists on main too. The open risk is iOS Safari's handling of root overflow: hidden for touch, which this environment cannot measure; if a device check fails there, useScrollLock would return for sheets and the rest of this PR still stands.

Scope

• This PR has one primary intent.
• Independently removable deltas were removed or split.
• Public text and artifacts contain no internal Meta context.

Testing

pnpm exec vitest run packages/core/src/{BottomSheet,Dialog,AlertDialog,Selector,DropdownMenu,ContextMenu,DateInput,DateTimeInput,TimeInput,Tokenizer,Layer,Layout,CommandPalette,MoreMenu,Popover,Lightbox,MobileNav}/ packages/lab/src/MobileTokenizer/
  main:   Test Files 85 passed, Tests 2451 passed
  branch: Test Files 86 passed, Tests 2471 passed   (+20 new, 1 new file)
  existing suites unedited against the new code: 6 failed (all structural, listed above), 574 passed
pnpm -F @astryxdesign/core typecheck                      ✓
ASTRYX_STRICT_LINT=1 pnpm exec eslint packages/core/src/BottomSheet/   0 errors (1 pre-existing warning in an untouched spec)
pnpm check:sync · check:knowledge · check:package-boundaries · check:changesets   ✓
node scripts/sync-exports.js --check                      ✓
git diff --check                                          ✓
pnpm exec vitest run --project node scripts/build-css.test.mjs   11 passed (Core rebuilt on the branch first)
node scripts/release/changeset-coverage.mjs pr --base origin/main --head HEAD --released v0.7.1   ✓ ships to @astryxdesign/core, each named
pnpm storybook:build                                      ✓ (main and branch)
pnpm exec playwright test packages/core/src/BottomSheet/__tests__/ --project chromium   43 passed (38 existing + 5 new)
pnpm exec playwright test … BottomSheetNativeModal, BottomSheetFocusReturn, BottomSheetKeyboard --project webkit   35 passed (30 existing + 5 new)

[vercel]

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

Project	Deployment	Actions	Updated
astryx	Ready	Preview	Oct 20, 2026 12:47am UTC

[astracat-bot]

AI review status for this pull request.

Review status	Updated
❓ Needs maintainer input (for maintainers only)	Oct 20, 2026, 4:41 AM UTC

[github-actions]

PR Analysis Report

Preview availability: Storybook and Sandbox are not both ready on the exact-head Vercel preview.

No new or modified components detected.

Bundle Size Summary

Package	Size (ESM)	Size (CJS)	Gzipped
@astryxdesign/core	N/A	4.9KB	1.2KB

Accessibility Audit

Status: No accessibility violations detected.

Visual Regression

40 of 482 shot(s) changed.

A change here is a question, not a failure: check whether the after is the
picture you intended. Record that review on the PR. Baseline maintenance is an
explicit dispatch of CI; this report never rewrites the baseline or adds a release gate.

component	story	theme	mode	pixels
Selector	Indicator Space Bottom Sheet Narrow End RTL	neutral	light	754,077
Selector	Indicator Space Bottom Sheet Wide Start	neutral	light	740,032
Chat	Full AI Chat	probe	light	230,823
Chat	Full AI Chat	probe	dark	229,805
RichTextEditor	Markdown parity: side by side	probe	dark	94,866
RichTextEditor	Markdown parity: side by side	probe	light	89,762
Collapsible	Dividers	probe	dark	66,445
Collapsible	Dividers	probe	light	66,445
Chat	Default	neutral	dark	36,304
Chat	Default	neutral	light	35,767
Chat	Empty State	probe	light	23,531
Chat	Empty State	probe	dark	22,622
ChatAutoScroll	Scroll Behavior Comparison	neutral	dark	15,378
ChatAutoScroll	Scroll Behavior Comparison	neutral	light	14,107
Chat	With Attachments	probe	light	13,406
Chat	With Attachments	probe	dark	11,232
ProgressBar	With Target Mark	probe	dark	7,895
ProgressBar	With Target Mark	probe	light	7,895
Markdown	Default	neutral	dark	7,573
Markdown	Default	neutral	light	7,566

and 20 more.

---

_{Generated by PR Enrichment workflow | View full report}

[astracat-bot]

Approval waits on a maintainer decision. This prototype rewrites the draft BottomSheetSwitcher requirements (FR2, FR5, FR8, FR9, PR1) and BottomSheet's ownership for a dialog-per-sheet model with a CSS scroll lock, and it changes what the switcher ref points at, so a maintainer needs to approve that direction and decide its release tier. Whatever they decide, these need changes:

• scrollbar-gutter: stable in sheetScrollLock.ts reserves a gutter the scrim does not cover, so on desktops with classic scrollbars a short page shows an undimmed 15px strip and the sheet shifts; reserve the gutter only while a scrollbar is showing.
• The switcher's dialog props now land on every sheet dialog, so a forwarded id is duplicated, and assignRef drops React 19 callback-ref cleanups (utils/mergeRefs.ts already handles them).
• The non-modal path adds a z-index 1001 band, which spec:AST-027 FR1 and FR4 rule out.
• spec:AST-013 FR10 needs a real iOS Safari check of the touch scroll lock, and at 2,239 authored lines the change should be split, for example with the scroll lock first.

[Automated review]

[astracat-bot]

scrollbar-gutter: stable reserves a gutter even when the page has no scrollbar. With classic scrollbars, on a page that does not scroll, the ::backdrop does not cover that gutter: the right 15px stays undimmed and the sheet re-centres 7.5px left (1280px, headless Chromium). holdScrollbarGutter only reserved it when window.innerWidth > documentElement.clientWidth. Keep that condition, or drop the gutter from this rule.

[astracat-bot]

Spreading the switcher's dialog props onto every sheet dialog duplicates identity attributes: <BottomSheetSwitcher id="checkout"> with three sheets renders three elements with id="checkout", where there used to be one. Forward id and data-testid only to the dialog the ref resolves to.

[astracat-bot]

This drops React 19 callback-ref cleanups. A ref that returns a cleanup never has it called, and receives null instead; the previous useMergedRefs → mergeRefs path ran the cleanup. Keep the returned cleanup and call it on detach, as utils/mergeRefs.ts does.

[astracat-bot]

spec:AST-027 FR1 and FR4 rule out depending on a page-wide band such as "overlays at 1000", and DEC-4 rules out adding another one. Here 1001 makes the step order depend on that band. The base ordered steps locally (positionerTop: 1) inside one 1000 shell. An isolated owner with local values, or the Layer route AST-027 names, avoids the new band.