mirror of
https://github.com/stablyai/orca.git
synced 2026-10-19 16:03:26 +00:00
Brennan Benson
a125662b10
Fixes #10757. A managed Codex account launches against its own self-contained CODEX_HOME, which broke three things at once. **History disappears (`/resume` is empty).** Codex's own resume picker only lists rollouts under the launch CODEX_HOME, and nothing bridged history into a per-account home — only the AI Vault's discovery scan knew about the other homes. Every other Orca-visible home's rollouts are now hardlinked in, on selection and again at launch, so one physical log is listed everywhere. **The restart panel sticks and "nothing happens".** A queued restart is only executed by a mounted TerminalPane, but the prompt covers every stale pane in the worktree — including tabs that are parked, cold-deferred, or otherwise unmounted. Those never drained, so the prompt returned with a button that now did nothing. Requesting a restart now answers the prompt immediately while the pane keeps its pending restart, and a pane drains that request when its reconnected PTY binds. The notice record survives so A -> B -> A still collapses. **Stuck on the previous account after a restart.** CODEX_HOME is fixed in a shell's environment at spawn, and the daemon keeps those shells alive across app restarts while the restart notices (renderer state) are discarded — so reopening Codex silently reused the old account with no way to escape. Each PTY's launch account is now recorded on disk and compared against the current selection at startup, which re-raises the prompt the app restart dropped. Unrecorded PTYs are never reported: a wrong prompt is worse than none. codex-home-paths.ts crossed its max-lines budget, so the resource copy-marker helpers moved to their own module rather than taking a suppression.