[vjeux]

Maintainer impact

The Popover rendering proof fails under load: 2 Layouts at +417 ms in #10169's CI.

Intended invariant

component:Popover PR4: "On a page with no other frame work, the touch growth and shrink do no Layout or Paint per frame".

Cause

• Each switch's motion engine sets an inline --astryx-motion-will-change-after: opacity at story mount and removes it 500 ms later, about 285 ms into the growth, laying out the anchored layer (the CI shape); late timers move it past 400 ms.
• Each motion's start also measures the touch hit areas (layout reads, then inset writes); a 360 ms stall would move that in too.
• Product impact: invisible. The demote costs about 2 ms of main thread, changes no pixel (0 of 31,000, growth held), and delays no compositor frame (longest gap 19 ms vs 18 without).

Change

One test file, four commits, +123 −11:

• Before the open, the proof polls (50 ms, up to 10 s) until no content element holds an inline motion promotion; computed will-change stays opacity under touch.
• Each window starts at 400 ms, or after the frame drawing the last of the hit areas' new insets if later, and must span 300 ms.

Review bar

first determine whether the 2 Layouts at +417 ms are a real forced layout from the Popover group's new code (a layout read during the touch motion), which makes it a product fix, or test timing (an unrelated layout counted in the window), which makes it a test fix that attributes layouts to their source. Red-first either way; never relax "no Layout or Paint per frame".

• Popover forced layout: ruled out; a test-only fix removes both sources; red-first: Evidence; toBe(0) unchanged.

Evidence

• Failure before the change: arms fail 3 of 3: 466ea9b (a 450 ms block 100 ms in) on main, Layout 2; b90572b (the block after the first frame) on the first commit, Layout 6; 01c589c (that block, one control a frame) on the third, Layout 6.
• Success after the change: at the head, those blocks pass (5e5848d, 8c8ad10 3 of 3; b292681 5 of 5), the proof passes 10 of 10 under each of three loads, Chromium 33 of 33; WebKit 26, 7 skipped, twice.
• Strictness: 86f45f3 (a switch style write 600 ms in) and 9342b76 (a 900 ms early block, on the floor) fail 3 of 3.
• No product or CI changes.

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

pnpm build && pnpm storybook:build
pnpm exec playwright test --config playwright.interaction.config.ts --project=chromium packages/core/src/Popover/__tests__/PopoverTouchMotion.interaction.browser.spec.ts -g "no Layout or Paint per frame" --repeat-each=10
taskset -c 2,3 pnpm exec playwright test --config playwright.interaction.config.ts --project=chromium --workers=1 packages/core/src/Popover/__tests__/PopoverTouchMotion.interaction.browser.spec.ts -g "no Layout or Paint per frame" --repeat-each=10

[vercel]

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

Project	Deployment	Actions	Updated
astryx	Ready	Preview	Nov 6, 2026 7:23pm UTC

[github-actions]

PR Analysis Report

🔎 Preview · 51a3e06

Storybook · Sandbox · built from this PR's head

Modified components

Component	Size	LoC	Visual · A11y · RTL
Popover · Storybook	78.7 kB (0) · 13.4 kB CSS (0)	1,488 (0)	✓ · ✓ · ✓

Package size: no published package changed; no new dependencies.

API changes: none in this pull request

Committed API report behind this build: cli 19 rows, core 24 rows; run pnpm api:report.

Touched components checked; add ci:run-visual, ci:run-a11y or ci:run-rtl to run the full suites.

---

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

[cixzhang]

/approve-spec dbaa9a4

[astracat-bot]

AI review status for this pull request.

Review status	Updated
🟢 Reviewed (for maintainers only)	Nov 6, 2026, 7:27 PM UTC

[astracat-bot]

Changes requested: one fix in the new window-start logic.

• The rendering windows wait only for the first hit-area inset write (PopoverTouchMotion.interaction.browser.spec.ts:726-729), so when a stall delays the hit-area pass and its 4 ms frame budget (utils/hitArea.ts:642) splits the three Switches across frames, the later writes still count in the presentation window (reproduced: Layout 6, Paint 2); extend the start through each write that lands before the previous one has drawn.

Otherwise the proof stays strict, since both zero assertions and the frame-loop control are unchanged, and a stall that only delays the pass now leaves the window clean where the base version fails. Minor: the first commit's message still describes a computed will-change check, but the code reads inline styles. CI has not run yet because this pull request is a draft.

[Automated review]

[cixzhang]

/approve-spec 51a3e06

[astracat-bot]

Approved pending CI: CI has not run on this head yet because the pull request is a draft, and marking it ready for review starts it.

The rendering windows now start after the last inset write of the hit-area pass, so a stall that spreads the pass over several frames leaves the presentation window at 0 Layout and 0 Paint, where the base version counts 10. Both zero assertions and the frame-loop control are unchanged, so the proof stays strict.

[Automated review]