Repository navigation
feat(dashboard): scroll every tab strip sideways with one wheel hook - #25680
Conversation
Only the side panel tab row uses useWheelToHorizontal. The follow-up bar cancels every wheel event over its row, even at either end, so the page cannot scroll while the pointer is over the chips. The embed strip adds the raw wheel delta from a passive handler, so the page scrolls with the strip. The session, terminal and File Explorer tab strips ignore the wheel. Bind useWheelToHorizontal on all five strips and delete the two handlers it replaces. The follow-up bar and the embed strip now let the page scroll at the ends of the row, and they ignore a wheel with a modifier key held.
UX Review (Opus 5.5, fork) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing to check. The diff only binds the existing wheel hook to five tab and chip strips. It adds or changes no control, label or state that a screenshot could show. [UX-REVIEWED] b3a9c54 Description read[DESCRIPTION-READ] 1d5f71d8c9396899c8e282a915ffa0b77db7badaf9ade0be39389c0e362e95d9 That sha256 names everything this verdict read: the PR description, plus the 3 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Design Review (Opus 5.5, fork) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the five strips now use the existing shared wheel hook, so they all follow the same rules, and the two hand-written handlers that broke the page scroll are deleted. [DESIGN-REVIEWED] b3a9c54 Description read[DESCRIPTION-READ] 2e23b990c829d800a2a8f9c6a3e8609e3003e0cf768910b5197c7b430b183d9c That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5, fork) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing to check. Every strip now uses the existing What this change shipsInventory (7 items) — 7 justifiedIntent: let mouse-wheel users reach clipped tabs and chips in every strip, with the same wheel rules on each one. This is a FIX (Closes #22154).
[FIRST-PRINCIPLES-REVIEWED] b3a9c54 Description read[DESCRIPTION-READ] 8fd6ee82e8e58e63ba17164343fb9aaa6e8f506b05c927cb0ee10ebf3bb143f8 That sha256 names the PR description this verdict read. Recompute it with |
Opus 5.5 Review (fork) — ✅ no blocking findingsReviewed |
GPT 6.1 Review (fork) — ✅ no blocking findingsReviewed Review detailsNo findings. |
iamwhatever
left a comment
There was a problem hiding this comment.
Approving b3a9c54.
Each strip now binds the existing useWheelToHorizontal hook from its scroller's ref (hook). The two hand-written handlers are deleted: the follow-up bar's handler always cancelled the wheel, and the embed strip's passive handler let the page scroll too. The shared test helper checks scrolling while clipped, the hand-off at the far end and a row that fits, and every strip calls it (helpers.tsx#L125). All fork AI lanes passed without notes, checks are green, and there are no other reviews.
— crew-pr-reviewer (r135)
Problem / Motivation
Goal: A vertical mouse wheel scrolls every tab and chip strip sideways, with the same rules on each one.
Only the side panel tab row uses the shared wheel hook from #22024. The other strips differ:
Why it matters
With a mouse wheel and no trackpad, a user cannot reach a tab or chip that is clipped off the edge of these strips.
Not a goal
What changed (motivation → approach → change)
The hook already has the rules we want. The five strips needed to use it.
Each strip binds the hook from its scroller's ref. The follow-up bar and the embed strip already had a ref callback, so the hook goes in there and their own wheel handlers are deleted. The other three strips (
SessionTabStrip, the terminal strip inBottomTerminalPanel, and the File Explorerapps/file-explorer/TabStrip.tsx) get a small ref callback that calls the edge-cue hook and the wheel hook.A wheel over the follow-up bar or the embed strip now behaves like the side panel: the page scrolls once the row is at an end, and a wheel with Ctrl, Shift, Alt or Meta held is left alone.
Backwards compatibility
Compatible: no props, keys or APIs change. Only the wheel behaviour above changes.
Reader: website/src/components/FollowUpBar.tsx:setScroller -- chat -- the hook binds from the same ref callback |
FollowUpBar.scrollEdges.test.tsx"scrolls sideways on a vertical wheel only while the chips are clipped"Reader: website/src/components/EmbedTabStrip.tsx:setStrip -- app -- the hook binds from the same ref callback |
EmbedTabStripCoverage.test.tsx"turns vertical wheel movement into horizontal scrolling"Tests
One shared helper,
expectWheelToScrollRow, checks that a strip scrolls sideways while clipped and leaves the wheel to the page at its far end and while the row fits. Each of the five strips has a test that calls it. Removing the wheel binding from any one strip fails that strip's test. I deleted the embed strip's "keeps the wheel translation" test, because the handler it guarded is gone.Manual verification
I ran each strip's real component in Chromium 145, 320 px wide with ten tabs, over a tall page. All 37 checks pass here, and 19 fail on main.
Not checked: touchpad, Windows, macOS, Firefox, WebKit, page zoom and display scaling.
Screenshots / video
Why no screenshot: the change binds a wheel listener and moves no pixel.
Related Issues
Closes #22154
Part of #23982
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)