fix(menu): re-measure merged-menu rows when a tab swap changes card height

[zenibako]

Fixes #3549.

Switching merged-menu tabs could leave a large gap around a short card or clip content and actions. The cached swap preserved an attached AppKit row while replacing its payload with a differently sized card; AppKit retained the row's old measurement.

The in-place swap now requires matching measured card heights. Different-height cards use the existing insert-before-remove path, so AppKit measures the incoming row. Equal-height swaps retain their attached views. Nearby tail, highlight, and metadata reconciliation is simplified; the complete PR is net -2 production lines.

Thanks @zenibako (Chandler Anderson) for isolating the cause, implementing the fix, and supplying the native diagnostic. The diagnostic now requires materialized card geometry, validates the intended tab selection, and compares row height against both the view frame and independently measured intrinsic height. Offscreen virtualized rows remain valid.

Native before/after proof

Same six-provider synthetic fixture on macOS 26.6.2, opened through the production status-item menu with cost sections and a scrollable Overview. Both test bundles were signed with the matching Developer ID. The before control disables only the height check; the final source was restored and rebuilt afterward. No real accounts or provider credentials were used.

Overview → Claude	Height check disabled	Fix enabled
Table / menu height	735 / 516 pt	516 / 516 pt
Card row / content height	341 / 122 pt	122 / 122 pt
Native assertions	Fail on stale geometry	All 11 transitions pass

Before	After

Validation

Reviewed head: 5af73dc90f6c655927f1a41b8781f6a061cf4f69.

• Signed native diagnostic: 38.25 seconds, all 11 phases, zero failures. Negative control: 44.28 seconds, fails on the original 735/516 and 341/122 measurements.
• make test-fast FILTER='StatusMenuTests|MenuRowContainerSizingTests': 182 tests in two suites passed.
• make check: zero violations in 2,276 files. Independent review and isolated Codex review completed; no accepted actionable findings remain.
• make test was attempted locally: 47 groups passed before an unrelated CostUsageStoreScaleProofTests 180-second timeout stopped the run. This is not a full local-suite pass.
• Exact-head CI: passed: both macOS test shards, all three Linux builds, lint, and the aggregate gate.

[clawsweeper]

🦞👀
ClawSweeper picked this up.

Pull request received. I will update this pull request when review starts.

ClawSweeper review complete

ClawSweeper finished reviewing this revision. The review result is being finalized.

View the workflow run.

[clawsweeper]

Codex review: needs maintainer review before merge. Reviewed September 13, 2026, 11:36 AM ET / 15:36 UTC (Revision 7).

ClawSweeper review

What this changes

Replace cached menu cards when their measured heights differ, preserve equal-height row reuse, simplify reconciliation, and add regression tests, native diagnostics, and a release note.

Merge readiness

✅ Ready for maintainer review

This remains a useful, well-supported fix: current main and v0.60.1 retain the affected swap behavior. Native before/after evidence supports the repair, the previous diagnostic finding is resolved, and no actionable patch defects remain.

Priority: P2
Reviewed head: 5af73dc90f6c655927f1a41b8781f6a061cf4f69

Review scores

Measure	Result	What it means
Overall readiness	🦞 diamond lobster (5/6)	A small runtime repair has strong native before/after evidence, focused regression coverage, and no remaining actionable findings.
Proof confidence	🦞 diamond lobster (5/6) ✨ media proof bonus	Sufficient (screenshot): The signed macOS production-menu diagnostic exercises cached row reconciliation with unequal-height cards; inspected screenshots show corrected spacing and restored actions, while the captured head-specific results report matching row/content geometry across all 11 phases.
Patch quality	🦞 diamond lobster (5/6)	No actionable review findings were identified.

Verification

Check	Result	Evidence
Real behavior	Verified	Sufficient (screenshot): The signed macOS production-menu diagnostic exercises cached row reconciliation with unequal-height cards; inspected screenshots show corrected spacing and restored actions, while the captured head-specific results report matching row/content geometry across all 11 phases.
Evidence reviewed	9 items	Applicable repository guidance: Read the complete root AGENTS.md; no nested AGENTS.md or maintainer-notes directory was present. Applied guidance on small changes, native AppKit validation, isolated fixtures, and provider-data separation. No builds or tests were executed during this read-only review. Introduced production behavior: The verified base-to-head diff adds a measured-height condition to cached row reuse. Differently sized cards take the existing insert-before-remove path; tail handling and metadata simplifications preserve their prior behavior. Fix remains necessary on main: Fetched main still swaps compatible card payloads without comparing heights. The caller subsequently refreshes heights only when row width changes, so that existing helper does not replace this repair.
Findings	None	None.
Security	None	None.

How this fits together

CodexBar’s merged menu displays cached usage cards when users switch provider tabs. Its reconciliation layer decides whether to reuse attached AppKit rows or insert replacement rows, affecting card sizing and access to menu actions.

flowchart TD
 A[Provider tab selection] --> B[Cached usage cards]
 B --> C[Compare row compatibility and height]
 C -->|Matching heights| D[Reuse attached row]
 C -->|Different heights| E[Insert replacement row]
 D --> F[Content-sized menu and actions]
 E --> F

Loading

Before merge

None.

Agent review details

Security

None.

Review metrics

Metric	Value	Why it matters
Production versus test delta	Production −2 lines; tests +432 lines	The runtime repair stays small; added coverage includes two regressions and an opt-in native diagnostic.
Native geometry	11 phases pass; stale height mismatch 219 → 0 pt	The reported negative control and after-fix measurements exercise the changed AppKit row-reuse path.

Root-cause cluster

Relationship: fixed_by_candidate
Canonical: #3549
Summary: This PR is the focused candidate fix for the reported merged-menu height defect; related account-switching work remains separate.

Members:

• canonical: Merged menu keeps a stale height after switching provider tabs (macOS 26 table-backed menus never shrink) #3549 - The linked open report documents the same unequal-card-height trigger and explicitly identifies this repair.
• adjacent_distinct: feat(menu): unify Claude and Codex account switching (view-only segments, shared System Account submenu) #3550 - Account-switching behavior is separate, and its body explicitly excludes the menu-height fix.
• partial_overlap: chore: collect provider credit notes and Tahoe diagnostics #3572 - The merged native diagnostic investigated the same report but made no production sizing repair.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Technical review

Best possible solution:

Keep content-sized provider tabs by replacing differently sized cached rows while preserving equal-height reuse and the existing menu cache.

Do we have a high-confidence way to reproduce the issue?

Yes: switching between unequal-height cached cards has a concrete native reproduction and negative control in the supplied evidence, supported by current-main source. This reviewer did not execute AppKit.

Is this the best way to solve the issue?

Yes: the guard uses existing measured sizes and replacement behavior, preserving equal-height reuse and avoiding the previously rejected uniform-height padding.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning medium; reviewed against 62a7fa1c409c.

Labels

Label justifications:

• P2: The fix addresses blank space and clipped menu actions during provider switching, with impact limited to merged-menu layout.
• rating: 🦞 diamond lobster: Overall readiness is 🦞 diamond lobster; proof is 🦞 diamond lobster and patch quality is 🦞 diamond lobster.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (screenshot): The signed macOS production-menu diagnostic exercises cached row reconciliation with unequal-height cards; inspected screenshots show corrected spacing and restored actions, while the captured head-specific results report matching row/content geometry across all 11 phases.
• proof: sufficient: Contributor real behavior proof is sufficient. The signed macOS production-menu diagnostic exercises cached row reconciliation with unequal-height cards; inspected screenshots show corrected spacing and restored actions, while the captured head-specific results report matching row/content geometry across all 11 phases.
• proof: 📸 screenshot: Contributor real behavior proof includes screenshot evidence. The signed macOS production-menu diagnostic exercises cached row reconciliation with unequal-height cards; inspected screenshots show corrected spacing and restored actions, while the captured head-specific results report matching row/content geometry across all 11 phases.

Evidence

What I checked:

• Applicable repository guidance: Read the complete root AGENTS.md; no nested AGENTS.md or maintainer-notes directory was present. Applied guidance on small changes, native AppKit validation, isolated fixtures, and provider-data separation. No builds or tests were executed during this read-only review. (AGENTS.md:1, 5af73dc90f6c)
• Introduced production behavior: The verified base-to-head diff adds a measured-height condition to cached row reuse. Differently sized cards take the existing insert-before-remove path; tail handling and metadata simplifications preserve their prior behavior. (Sources/CodexBar/StatusItemController+MenuReconcile.swift:128, 5af73dc90f6c)
• Fix remains necessary on main: Fetched main still swaps compatible card payloads without comparing heights. The caller subsequently refreshes heights only when row width changes, so that existing helper does not replace this repair. (Sources/CodexBar/StatusItemController+MenuReconcile.swift:125, 62a7fa1c409c)
• Latest release comparison: The v0.60.1 source has the same unconditional compatible-card reuse and lacks the proposed height guard. Local historical-object retrieval returned HTTP 403; the read-only GitHub contents API supplied the release source successfully. (Sources/CodexBar/StatusItemController+MenuReconcile.swift:125, 0b79c8edec8b)
• Native before/after proof: Inspected both downloaded screenshots from the captured PR body: https://github.com/user-attachments/assets/76516072-f854-4969-afe4-6eff712e1f22 shows oversized gaps and clipped actions; https://github.com/user-attachments/assets/2582f24a-0e63-4ac4-8d4f-7dc832988187 shows a compact Claude card with About and Quit visible. The captured body ties them to signed production-menu runs on macOS 26.6.2 and reports table/menu heights improving from 735/516 to 516/516 points, with all 11 phases passing at the reviewed head. (5af73dc90f6c)
• Regression coverage and review continuity: The two focused tests verify equal-height identity preservation and different-height replacement in both directions. The complete native diagnostic exercises the production status-item menu, checks selected tabs and materialized row geometry, and uses an isolated synthetic fixture. GitHub commit inspection confirms the earlier two-provider diagnostic mismatch was removed; all scripted providers are enabled now. (Tests/CodexBarTests/MergedMenuScrollingSwapNativeProofTests.swift:166, 5af73dc90f6c)

Likely related people:

• unknown: The claimed source-line change could not be verified from bounded local history. (role: source history unknown; confidence: low)
• unknown: The claimed source-line change could not be verified from bounded local history. (role: source history unknown; confidence: low)

Rating scale

Score	Internal tier	Crab rank	Meaning
6/6	S	🦀 challenger crab	Exceptional readiness
5/6	A	🦞 diamond lobster	Very strong readiness
4/6	B	🐚 platinum hermit	Good normal PR; ordinary maintainer review
3/6	C	🦐 gold shrimp	Useful, but confidence is limited
2/6	D	🦪 silver shellfish	Proof or implementation needs work
1/6	F	🧂 unranked krab	Not merge-ready
N/A	NA	🌊 off-meta tidepool	Rating does not apply

Overall follows the weaker of proof and patch quality.
Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

Workflow

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

History

Review history (6 earlier review cycles)

• reviewed 2026-09-13T13:15:06.888Z sha 82c325f :: needs real behavior proof before merge. :: [P3] Use a matching swap script for the two-provider diagnostic
• reviewed 2026-09-13T13:36:26.454Z sha 0d48ba9 :: needs maintainer review before merge. :: none
• reviewed 2026-09-13T14:15:20.362Z sha 0d48ba9 :: needs maintainer review before merge. :: none
• reviewed 2026-09-13T14:45:29.225Z sha f68cbe1 :: needs maintainer review before merge. :: none
• reviewed 2026-09-13T15:05:20.484Z sha f68cbe1 :: needs maintainer review before merge. :: none
• reviewed 2026-09-13T15:11:18.494Z sha 5af73dc :: needs maintainer review before merge. :: none

[steipete]

Landed as caad1ca. Thanks @zenibako for the root-cause isolation, fix, and native fixture.

The final head passed 182 focused tests (make test-fast FILTER='StatusMenuTests|MenuRowContainerSizingTests'), make check, and both complete macOS CI shards plus Linux builds in https://github.com/steipete/CodexBar/actions/runs/34764650337. The local full-suite attempt stopped at an unrelated scale-test timeout; that limitation is recorded in the PR body.

The matching Developer ID-signed macOS 26.6.2 diagnostic passed all 11 transitions in 38.25 seconds. Disabling only the height check reproduced the original table/menu mismatch (735/516 points) and card row/content mismatch (341/122 points), failing the diagnostic in 44.28 seconds. The inspected before/after screenshots in the PR contain only synthetic fixture data. Production changes are net -2 lines, and the Unreleased changelog credits the contributor.