Skip to content

manual pane rerun retains stale failed job state and output #55

Description

@tomfordweb

Observed after recovering a Vite server in an existing sidemux-managed pane.

Reproduction:

  1. Start a background job and interrupt it.
  2. Rerun the saved command manually in the same pane through send_keys.
  3. Wait until the new Vite process reports ready and begins listening.
  4. Call list_panes, read, and wait for the pane or previous job.

Observed:

  • The manually started server was healthy and listening.
  • Sidemux continued to expose the previous job's failed state.
  • read and wait returned stale output from the previous failed run.
  • wait could time out despite the new process already reporting ready.

Expected:

A command started manually in a managed pane should either create a new tracked execution context or make the pane explicitly report that its foreground process is untracked. A previous job's exit status and output cursor should not be applied to the new process.

Related but distinct: #27 covered stale pattern matches within one tracked job, and #28 covered a tracked foreground job remaining active after the prompt returned. This case starts a new command in the same pane after the old job ended.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions