mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-15 23:31:29 +00:00
57894b5b5e
Per-tab path routing for Models made each tab a separate route parsed out of the pathname, which is fragile under a static export mounted at a runtime-variable server-root prefix. Revert the tabs to in-memory state: a single /models-and-endpoints route renders an antd Tabs whose active tab is React state, and each tab body moves from its own page.tsx into a non-routed panel component under panels/. Role-gating (which tabs show), the refresh control and the header are unchanged. The ?model= / ?team= query drill-in stays: it is query-param based (read via useSearchParams, written via history.pushState), so it is unaffected by the server-root prefix and remains shareable. The shared tab-routing helpers (createTabRoutes / useTabRouting) are untouched; the other four pages still use them. Removes the per-tab route dirs, layout.tsx and tabRoutes.ts (+ their path-routing tests) and replaces the layout's coverage with a page test for in-memory tab switching, the drill-in overlays and role-gating. |
||
|---|---|---|
| .. | ||
| litellm-dashboard | ||
| Dockerfile | ||
| nginx.conf | ||