Files
iptvnator/apps/electron-backend-e2e/src
4grayandClaude Opus 5 1ef75b6392 fix(e2e): make the single-instance coverage survive CI
Two fallouts from adding the lock, both in test scaffolding:

- The competing-instance helper spawned Electron without the
  --no-sandbox/--disable-gpu flags launchElectronApp adds on Linux CI, so
  the process died on a signal (exitCode null) instead of being refused.
  Both paths now build argv through one shared helper. The helper also
  captures stderr and reports the exit signal, so a future CI-only
  failure says why instead of just "not 0".

- requestSingleInstanceLock creates SingletonLock/SingletonSocket
  (lockfile on Windows) in userData and holds them for the process
  lifetime. Windows releases those handles asynchronously as the process
  dies, so the dataDir fixture's immediate rmSync started hitting EBUSY
  in unrelated suites. Removal now retries and, failing that, warns
  instead of failing a green test — it is throwaway temp state.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-07-27 20:08:31 +02:00
..