Testing and Reproduction

Testing and Reproduction

Run the software suite

From the repository root, install Python 3 and Node.js. CI uses Python 3.12, Node 22, and Playwright 1.63.0. Install Chromium for the browser portion.

npm i --no-save playwright@1.63.0
npx playwright install chromium
npm test

The runner checks manifest integrity, Python and shell code, launch helpers, API behavior, the Decky panel, and browser smoke coverage. The local browser test may print a skip and still exit successfully when Playwright or Chromium is missing. Run CI=1 node scripts/test_e2e_app.js to make those prerequisites mandatory. Hosted CI separately rebuilds the plugin archive and compares its contents. Test runner · Browser test · CI workflow

Investigate a real launch

1. Record the exact revision deployed to the site and installed on the Deck, Deck model and SteamOS version, Proton version, storage backend, and whether the game is a ready folder or a local install job.
2. Run systemctl --user status deckstream and python3 ~/.config/deckstream/doctor.py. Match the queue state to the daemon's recent logs and the game's stream report. Doctor · Launch report implementation
3. Label the cache state before pressing Play. For a timing comparison, distinguish an empty cache from a previously profiled cold run and a warm run. Record profile state, network conditions, bytes fetched, enabled flags, and times to Proton start, title menu and usable controls where observed.
4. Share the first failing stage or the actual observed milestone. If you create a report using doctor.py --bundle, review it before sharing. Remove credentials, private hosts and identifying game paths from any attached output.

A passing suite reproduces software checks. A device result needs a Steam Deck, an authorized game folder, and observed game behavior. Evidence and Limits