[teamleaderleo]

live-qa on nightly 3a63d6d (Home/Chief pane 655-1100 px wide): a 1 px full-height line ran through the Chief chat at the list's old 320 pt place, black in light and dark, while the real divider sat at the clamped width.

Why

HomeSidebarSplitView.layout() calls super.layout() first, and NSSplitView places its divider's layer there, at the current position. Then the split clamps the list (place(target) → setPosition), which moves the columns but not the divider layer until a later layout. The chat column is see-through, so the stale layer showed through it.

Change

After place(target) in layout(), lay out once more so the divider layer follows. Widths, clamps and behavior don't change. (setWidth and resetWidth go through a normal layout and were already correct; a test pins both.)

Tests

HomeSidebarSplitDividerTests reads the split's divider layers (the split view's own sublayers) after a clamp, and after a width set outside layout, in light and dark.

• Red at the test-only commit a2bf972: cmux-ci run --class light --script scripts/ci/package-test-lane.sh --ref a2bf9726ead37c9e9c2ef32a12401d3143b5cfbf --arg=suite --arg=Packages/macOS/CmuxNext --arg=HomeSidebarSplitDividerTests: aClampedDividerLeavesNoLineAtItsOldPlace fails in light and dark (a divider layer stays at 320 pt and isn't at the list's width); the width-set test passes.
• Green at the head a1b4d1b: the same command with --ref a1b4d1b641b96519e1a695e0bacb4e8498b594cc --arg=HomeSidebarSplitDividerTests,HomeNarrowSidebarTests: both suites pass (aws-m4pro-7).
• cmux-ci compile-check --ref a1b4d1b641b96519e1a695e0bacb4e8498b594cc --app: passed.

Observed / Not run

Observed: the red and green runs above on the aws fleet (Xcode 26.6), and the app compile. Not run: the app on screen; a capture-mini before/after follows in a comment.

Changelog

Fixed: no stray line through the Chief chat when its pane is narrow.

No capture: the fix is a layer position, which the divider-layer frame tests read directly in light and dark; the real-app before/after is being taken on a capture mini and will be added as a comment.

live-qa report: the stray 1 px line at x=974 (feat-cmux-next nightly 3a63d6d).

Agent session: cosmic-gray

🤖 Generated with Claude Code

[github-actions]

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

Review a1b4d1b: 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=pr23998 -f ref=chief-ui-split-divider -f platform=macos -f backend_mode=local

red: cmux-next checks (god files, concurrency, crash safety, l10n); base red: the same failure on the base 5363032; does not block

[teamleaderleo]

Owned by chief-ui on cosmic-gray · session

[blacksmith-sh]

Found 12 test failures on Blacksmith runners:

Failures

Test	View Logs
/tmp/tmpb3fj3rfk/summary-missing.json (live):/summary-missing.json (live):	View Logs
/tmp/tmpkgois0tx/summary-fail.json (live):/summary-fail.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
convention violations found (ERROR lines above)./ convention violations found (ERROR lines above).	View Logs
nightly cmux-tui resolve/nightly cmux-tui resolve	View Logs
test_every_submodule_free_checkout_on_an_owned_runner_resets_submodules (main.Reuse dWorkspaceSubmodules.test_every_submodule_free_checkout_on_an_owned_runner_resets_submo dules)/ test_every_submodule_free_checkout_on_an_owned_runner_resets_submodules (main.Reuse dWorkspaceSubmodules.test_every_submodule_free_checkout_on_an_owned_runner_resets_submo dules)	View Logs
test_every_submodule_free_checkout_on_an_owned_runner_resets_submodules (main.Reuse dWorkspaceSubmodules.test_every_submodule_free_checkout_on_an_owned_runner_resets_submo dules) (job='web-bundles')/ test_every_submodule_free_checkout_on_an_owned_runner_resets_submodules (main.Reuse dWorkspaceSubmodules.test_every_submodule_free_checkout_on_an_owned_runner_resets_submo dules) (job='web-bundles')	View Logs

...and 2 more test failures. View all on Blacksmith

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

[teamleaderleo]

cmux-next crash-class rules / cmux-next crash-class rules is red on 11 of 11 fresh feat-cmux-next PR heads in the last 2:00:00, so it is most likely a base red, not this PR's. The base-reds lane has it; a base merge picks up the fix once it lands.

[teamleaderleo]

Review (claude-reviewer-23998): SHIP at a1b4d1b

Correctness first, read-only review of 0ba602a4...a1b4d1b6 (no build run here).

• HomeSidebarSplitView.swift:90: the second super.layout() is a static call to NSSplitView's implementation, so it can't re-enter our layout(). It runs only inside the !placed || mismatch branch. If AppKit marks the view as needing layout again, the next pass finds sidebarWidth == target and does nothing more, so it can't loop. No new loop risk beyond what place() already had.
• splitViewDidResizeSubviews (:151-157): the extra pass runs after placing is reset. The width is still not saved, because the guard needs NSSplitViewDividerIndex, and a layout or adjust pass doesn't set it. The first super.layout() at :77 already ran on every pass outside placing, so the save behavior is unchanged. ResizeInvariantTests.theHomeTranscriptKeepsItsMinimumThroughResizes (clamp to 76, then grow back to 320) would catch a clamped width being saved by mistake. That test should be green on this head.
• Behavior: widths, clamps, holding priority and drag paths are unchanged. A drag runs no layout pass, and setWidth/resetWidth don't touch the new line.
• Test: the !frames.isEmpty check stops it from passing when no divider layers exist. If a future macOS stops hosting dividers as split-owned sublayers, the test fails loudly rather than silently passing, which is acceptable. #if DEBUG around setWidth matches the source and is fine, since tests build Debug. Swift 6: $0.delegate === split (CALayerDelegate? vs NSView) and \.frame on @MainActor are fine. The local split shadowing Self.split is legal.

No blocking findings. CI on this head was still running at review time.

[github-actions]

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.3973137732601 (15186f150b)
• cmux NEXT nightly 1.0.0-nightly.3973389583201 (48a69e9f66)