[teamleaderleo]

Leo, 2026-11-04:

hovering across the session rail gets a bit laggy and the missing chunks of the wave bother me :( it needs to get faster

also the left side session rail should just be having a nicer max height instead of going all the way up and down. make it look like codex's size.

also the height of the left scroll rail should never be affected by anything

Leo-ask: ask-0eea9d95e5

What changes

• One height for every chat, the Codex rail's size. The prompt rail's box was max(count × 10, 320) capped at 92% of the window, so in a long chat it ran from the top of the pane to the bottom. It is now 240 px for every chat. The prompt count, the window, the pane's height, the prompts on screen (the band) and the bottom stack don't change it. Past 24 prompts its list scrolls and keeps the band in view (railScrollTop). Only a view too short for 240 px clips it (railRoom). A composer or tray that grows tall enough to reach the rail lays over its foot: the rail neither moves nor shortens (Leo, 2026-11-03: "i just don't like it when the thing shifts"). The other arms keep the old foot-shortening (railClearHeight). height.experiment.ts gains fixed-200, fixed-240 (default) and fixed-280 (fixed: true in RAIL_HEIGHTS), and the agent-pane.thread-minimap-height gallery entry's notes describe them beside the old arms.
• No holes in the wave on a sweep. In the zipper (the shipped wave), every tick that was not the hovered one eased to its new length over 180 ms. So during a sweep the hovered tick snapped long, but the rows ahead of it were still short, and the trailing rows were still long. The bump had holes on its leading side. Now growing ticks ease over the quick step (80 ms), like the hovered tick does on arrival, and shrinking ticks keep 180 ms as the zipper's tail. The hovered tick is still the single longest every frame.

The hover lag itself (React work per row change, the preview card) is measured separately, and its fix follows in a separate PR.

Tests

Red first (fa0b8ea), then the fix:

• bun test test/thread-minimap-wave-model.test.ts: new "a sweep keeps the bump whole". On base, after one frame of a 60 Hz sweep, the row ahead has 0.18 of its length; it needs more than 0.35. Passes on head.
• bun test test/thread-minimap-rail.test.tsx: new "the rail is one height, the Codex rail's, whatever the prompt count or the pane's height" (1 to 1,000 prompts, 400 to 2,100 px views → 240). Fails on base (320, 450, 662, …), passes on head.
• bun test test/thread-minimap-rail.test.tsx: the band (1, 24, 50 and 1,000 prompts on screen) keeps 240. Fails at 8a21b18 (500 for 50), passes on head.
• bun test test/thread-rail-layout-box.test.tsx: the rail's box is identical with a 120 px, a 188 px and a rail-reaching stack. Fails at 8a21b18 (the height shrank), passes on head.
• test/thread-rail-full-height.test.tsx was updated to the fixed box: the rows not drawn are counted through the list's gaps.
• All test/thread-minimap* and test/thread-rail* jsdom suites pass locally.

Observed / Not run

• Observed: the bun suites above, red on base and green on head; vp fmt --check on changed files. scripts/web-lint.sh fails only on base files (pages/changelog/updateSurfaces.css design token, App.tsx and others' format).
• Not run: the type-aware lint/type check (left to CI on this shared host), and gallery captures.

Follow-ups

Review follow-ups: #35239.

Spec

design/session-rail.md "Height and overflow" (leo-workspace) should read: the compact rail is one fixed height, 240 px (the Codex rail's size), for every chat. Prompt count, window, pane height, the band and the bottom stack never change it or move it (a stack tall enough lays over its foot), and a view shorter than that clips it. Past 24 prompts its list scrolls. This replaces "starts at 320 px, grows toward 92%" (last good commit d53567b). This host can't publish to leo-workspace, so that edit follows from a lane that can.

Changelog

• Changed: The agent chat's prompt rail is one compact height for every chat, and its hover wave keeps up with a quick sweep.

risk:normal

🤖 Generated with Claude Code

Agent session: agent session

[teamleaderleo]

Review (subagent claude-opus-5-5 medium): SHIP at 8a21b18

Visual verdict: no images; code verdict only [follow-up].
Checked railHeight/railTop/overflow and wheel ownership with the fixed 240 px box, zipper invariants (no overshoot, hovered single longest), other readers of the height arms; all rail suites pass.

Follow-ups:

• [follow-up] the band rule can still grow the rail past 240 px when >24 prompts are on screen at once
• [follow-up] restore direct railHeight coverage for the initial-* arms
• [follow-up] reword the height gallery note "tall initial rail"
• [follow-up] chats with 25-66 prompts now scroll the rail list, so the wheel over the rail scrolls it (intended)

[github-actions]

cmux status _{(one comment, edited in place; each line is updated by the job that owns it)}

Review c55d741: HQ feed card · gallery before/after runs · checks

App of this head (app.zip, kept 1 day)

gh workflow run reload-build.yml -R manaflow-ai/cmux --ref feat-cmux-next -f tag=pr34666 -f ref=lane/rail-wave-fast-codex-height -f platform=macos -f backend_mode=local

Stale: these images are of 8a21b18; the head is now c55d741 and a new capture is running.

Gallery before/after for 8a21b18: 16 states rendered, 32 images differ, 3438 more touched states not rendered. Images are in the run's artifact, not on this public PR; the independent review of this head reads them on its own, so the PR needs no hand-made screenshots of these states.

gh run download 41763960516 -R manaflow-ai/cmux -n gallery-advisory-34666-8a21b182423257daf96c68753e1fb0b94da8633e (then read gallery-review.json)

States

State	Theme	Status	Diff	Picked by
agent-pane.chat-minimap/streaming	Apple System Colors Light	changed	0.016%	imports model.ts (1 hops)
agent-pane.chat-minimap/streaming	Apple System Colors	changed	0.016%	imports model.ts (1 hops)
agent-pane.chat-minimap/at-rest	Apple System Colors Light	changed	0.016%	imports model.ts (1 hops)
agent-pane.chat-minimap/at-rest	Apple System Colors	changed	0.016%	imports model.ts (1 hops)
agent-pane.chat-width/standard	Apple System Colors Light	changed	0.0032%	imports model.ts (1 hops)
agent-pane.chat-width/standard	Apple System Colors	changed	0.0032%	imports model.ts (1 hops)
agent-pane.chat-width/wide	Apple System Colors Light	changed	0.0032%	imports model.ts (1 hops)
agent-pane.chat-width/wide	Apple System Colors	changed	0.0032%	imports model.ts (1 hops)
agent-pane.chat-width/wide-blocks	Apple System Colors Light	changed	0.0032%	imports model.ts (1 hops)
agent-pane.chat-width/wide-blocks	Apple System Colors	changed	0.0032%	imports model.ts (1 hops)
agent-pane.composer-edge/short-edge	Apple System Colors Light	changed	0.0286%	imports model.ts (1 hops)
agent-pane.composer-edge/short-edge	Apple System Colors	changed	0.0286%	imports model.ts (1 hops)
agent-pane.composer-edge/wide-edge	Apple System Colors Light	changed	0.0286%	imports model.ts (1 hops)
agent-pane.composer-edge/wide-edge	Apple System Colors	changed	0.0286%	imports model.ts (1 hops)
agent-pane.composer-edge/no-edge	Apple System Colors Light	changed	0.0286%	imports model.ts (1 hops)
agent-pane.composer-edge/no-edge	Apple System Colors	changed	0.0286%	imports model.ts (1 hops)
agent-pane.composer-floats/scrolled-to-end	Apple System Colors Light	changed	0.0714%	imports model.ts (1 hops)
agent-pane.composer-floats/scrolled-to-end	Apple System Colors	changed	0.0714%	imports model.ts (1 hops)
agent-pane.composer-floats/composer-grown	Apple System Colors Light	changed	0.0714%	imports model.ts (1 hops)
agent-pane.composer-floats/composer-grown	Apple System Colors	changed	0.0714%	imports model.ts (1 hops)
agent-pane.composer-rest-gap/gap-40	Apple System Colors Light	changed	0.0495%	imports model.ts (1 hops)
agent-pane.composer-rest-gap/gap-40	Apple System Colors	changed	0.0495%	imports model.ts (1 hops)
agent-pane.composer-rest-gap/gap-56	Apple System Colors Light	changed	0.0495%	imports model.ts (1 hops)
agent-pane.composer-rest-gap/gap-56	Apple System Colors	changed	0.0495%	imports model.ts (1 hops)
agent-pane.composer-rest-gap/gap-72	Apple System Colors Light	changed	0.0495%	imports model.ts (1 hops)
agent-pane.composer-rest-gap/gap-72	Apple System Colors	changed	0.0495%	imports model.ts (1 hops)
agent-pane.continuation-spacing/shipped	Apple System Colors Light	changed	0.0075%	imports model.ts (1 hops)
agent-pane.continuation-spacing/shipped	Apple System Colors	changed	0.0075%	imports model.ts (1 hops)
agent-pane.continuation-spacing/two-thirds	Apple System Colors Light	changed	0.0075%	imports model.ts (1 hops)
agent-pane.continuation-spacing/two-thirds	Apple System Colors	changed	0.0075%	imports model.ts (1 hops)
agent-pane.continuation-spacing/half-tucked	Apple System Colors Light	changed	0.0075%	imports model.ts (1 hops)
agent-pane.continuation-spacing/half-tucked	Apple System Colors	changed	0.0075%	imports model.ts (1 hops)

_{Pick other states with a Gallery: entry.id, entry.id/variant line in the PR body.}

Unclear: web / react-apps-check (1) failed: no known signature; failed step: Fail on the gate checks. This PR merged before its CI finished: fix forward on main.

CI failed on c55d741296 (run 41804363657 attempt 1): 2 unknown.

Job	Verdict	Why
web / react-apps-check (1)	unknown	no known signature; failed step: Fail on the gate checks
web / react-apps-check	unknown	no known signature; failed step: Check webview test shards

Not re-run automatically: web / react-apps-check (1), web / react-apps-check are not machine failures.

Written by scripts/ci/classify_failures.py (ci-failure-attribution.yml); signatures are its SIGNATURES table. A machine verdict is the runner's fault, not this PR's; yours means the failing file is one this PR changes, also red on main that main's latest full suite fails the same way, seen on other PRs that it failed on another pull request's run lately.

red: cmux-next checks (god files, concurrency, crash safety, l10n); do not merge until green; see lane-rules
red: web / Web status; do not merge until green; see lane-rules
red: web / react-apps-check; do not merge until green; see lane-rules
red: web / react-apps-check (1); do not merge until green; see lane-rules

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.4182499138501 (e0aa712b3b)
• cmux NEXT nightly 1.0.0-nightly.4184087390301 (08052d331e)
• cmux NEXT nightly 1.0.0-nightly.4184224384601 (a3abfc4160)
• cmux NEXT nightly 1.0.0-nightly.4184409870901 (78ee0e108e)

[teamleaderleo]

Review (subagent claude-opus-5-5 medium): SHIP at c55d741

Delta since the SHIP at 06971b0 (itself a delta from 8a21b18): a fixed arm (default fixed-240) is one height whatever the band or bottom stack, and a grown stack lays over its foot instead of moving or shortening it. The jsdom suites test/thread-minimap and test/thread-rail pass (163). The new tests fail at 8a21b18 (500 for a 50-prompt band; the height shrank under a grown composer).

Follow-ups:

• F1 [follow-up] a tall stack can cover the fixed rail's lowest ticks, including the active one; consider clamping the list's scroll to the live room without moving the box.
• F2 [follow-up] unit test for railScrollTop with a band longer than the 240 px rail.

[blacksmith-sh]

Found 8 test failures on Blacksmith runners:

Failures

Test	View Logs
/tmp/tmp0bt1dkje/summary-fail.json (live):/summary-fail.json (live):	View Logs
/tmp/tmp1swy6th1/summary-missing.json (live):/summary-missing.json (live):	View Logs
[new_workspaces] debug.new_workspace timed 0 of 50 new workspace rows/ [new_workspaces] debug.new_workspace timed 0 of 50 new workspace rows	View Logs
[new_workspaces] the window draws no frames/[new_workspaces] the window draws no frames	View Logs
[tabs] debug.tab_switch timed 0 of 40 switches/ [tabs] debug.tab_switch timed 0 of 40 switches	View Logs
[tabs] the window draws no frames/[tabs] the window draws no frames	View Logs
nightly cmux-tui resolve/nightly cmux-tui resolve	View Logs
the short nightly decide gate must use the pooled Linux route with a Blacksmith fallbac k/ the short nightly decide gate must use the pooled Linux route with a Blacksmith fallbac k	View Logs

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[teamleaderleo]

Finding outcome: comment-6627083606/F1 filed-follow-up #35239
Finding outcome: comment-6627083606/F2 filed-follow-up #35239