[performance] Optimize useHover hooks and safePolygon hot paths

[atomiks]

• I have followed (at least) the PR section of the contributing guide.

Purely perf-related change for bundle size + runtime cost

[pkg-pr-new]

• base-ui-tanstack-start

• vite-css-base-ui-example

  pnpm add https://pkg.pr.new/@base-ui/react@4199
  

  pnpm add https://pkg.pr.new/@base-ui/utils@4199
  

commit: 9ccef50

[mui-bot]

Bundle size report

Bundle	Parsed size	Gzip size
@base-ui/react	▼-36B(-0.01%)	▼-86B(-0.06%)

Details of bundle changes

---

Check out the code infra dashboard for more information about this PR.

[netlify]

✅ Deploy Preview for base-ui ready!

Name	Link
🔨 Latest commit	87abf70
🔍 Latest deploy log	https://app.netlify.com/projects/base-ui/deploys/69a8f1fb9e08790008771b38
😎 Deploy Preview	https://deploy-preview-4199--base-ui.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[flaviendelangle]

Only one concern from Claude Code
Nice to have perf improvement AND bundle size reduction in the same PR 🔥

[atomiks]

Codex Review

Overview

This PR refactors hover and safePolygon hot paths, extracts shared hover delay helpers, and adds targeted tests around wrapper-trigger handling and safePolygon geometry. I reviewed the full upstream/master...head diff and the latest commit on this branch.

Findings

No blocking issues found in this patch.

1. ℹ️ [Informational] safePolygon hot path is faster than master in the focused micro-benchmark

Impact: The main performance-sensitive path reviewed here shows a measurable win relative to master, so the scalar geometry rewrite does not introduce a runtime regression in this check.

Evidence: packages/react/src/floating-ui-react/safePolygon.ts plus the new safePolygon tests (packages/react/src/floating-ui-react/safePolygon.test.ts) were benchmarked with the same harness on both refs (120 samples, 2500 events/sample, safePolygon({ requireIntent: false }) hot path).

Ref	Mean (us/event)	p50 (us/event)	p95 (us/event)	Samples
upstream/master (0824de08c)	0.1393	0.1136	0.2006	120
PR head (87abf70dd)	0.0755	0.0540	0.1600	120

Recommendation: No change required for this path based on current evidence.

2. ℹ️ [Informational] DRY simplification in hover delay utilities is clean

Impact: Delay resolution and pointer-type gating are now centralized, which lowers maintenance surface without changing public behavior.

Evidence: packages/react/src/floating-ui-react/hooks/useHoverShared.ts centralizes resolution in resolveValue, and packages/react/src/floating-ui-react/hooks/useHoverFloatingInteraction.ts now reuses getDelay(..., 'close', ...) instead of a separate close-only helper.

Recommendation: Keep this structure as-is.

Confidence: 4/5

High confidence from a full-branch diff review, targeted behavioral tests in touched modules, and a direct head-vs-master micro-benchmark on the main hot path.

Notes

• Review scope covered all files changed in this branch, not just the latest commit.
• CI bundle delta should be read from the latest bot run on current head (87abf70dd).