Dead code: sidebarMarkup() and badgeFor() in ui/pages/index.js have no callers

[richard-devbot]

Found while promoting Approvals into the nav (#537).

sidebarMarkup() in src/observability/dashboard/ui/pages/index.js renders a legacy flat sidebar — one nav-link per page, with an approvals badge via badgeFor(id). Neither function has a caller anywhere in src/. The live navigation is the #278 destination shell (navigationGroups/desktopNavigationMarkup in ui/navigation.js), which renders .destination-link + .secondary-link and filters on !child.hidden.

pageMarkup() in the same file is live — only sidebarMarkup and its badgeFor helper are orphaned.

Why it is worth removing rather than leaving: it is actively misleading. It renders every page unconditionally, ignoring the hidden flag, and carries badge-approvals/badge-alerts element ids that look like the live badge wiring. While investigating #537 it read as evidence that Approvals was already reachable — it is not.

Small and self-contained: delete both functions, confirm no test references sidebarMarkup or the badge-approvals/badge-alerts ids it emits (the live topbar chips are separate elements — check before deleting), run the gates.

[richard-devbot]

CLAIM — starting (Claude Main, 2026-08-01). Will verify no test references sidebarMarkup/badgeFor or the badge-approvals/badge-alerts ids before deleting, then run the full gate.

[richard-devbot]

Shipped via PR #543. As instructed, checked for test references before deleting — found one (dashboard-e2e-96.test.js), which turned out to be silently validating the dead sidebarMarkup() this whole time, giving false confidence about nav/container parity against the wrong navigation implementation. Retargeted it at the real live nav (desktopNavigationMarkup) instead of just dropping the coverage — mutation-tested against two real regressions, both caught.

Also found the chain didn't stop at sidebarMarkup: client.js's setBadge('badge-approvals'/'badge-alerts', ...) calls target ids only sidebarMarkup() ever rendered, so they were silently dead in production too (not just tests) — setBadge no-ops when the element is missing. Removed those calls and the now-fully-unused setBadge() helper itself. Full detail in the PR.

[richard-devbot]

Shipped via PR #543, merged in commit 8c869b2.

Deleted sidebarMarkup()/badgeFor() as asked. Following the issue's own "check before deleting" instruction surfaced a connected chain: the one test that referenced sidebarMarkup() was silently validating dead code, giving false confidence about nav/container parity — retargeted at the real live nav (desktopNavigationMarkup) instead of dropping the coverage, mutation-tested against two real regressions. Also found client.js's setBadge('badge-approvals'/'badge-alerts', ...) calls were themselves dead in production (no live element ever rendered those ids), removed them, and removed the now-fully-unused setBadge() helper itself.

Qodo's review caught 2 real issues on the first cut — a shallow existence-only nav assertion and a brittle exact-HTML-substring check — both fixed and mutation-tested.

1950/1950 core tests, 43/43 browser tests, typecheck/lint/security-audit/validate all clean.