Contributing

Thanks for helping Linux streamers! Every distribution, desktop and app combination is a little different, so reports from your setup are as valuable as code.

The quickest contributions

• Your doctor output. Run tuxcast doctor and open an issue when a check is wrong or misses a problem you had.
• An Elgato action tuxcast does not convert. The conversion report names its UUID. Open an Elgato action issue with that UUID and, if you can, the key's Settings from the profile's manifest.json.
• A guide. The guides live in docs/guides/. Fixes and new sections are welcome, especially for desktops and GPUs we don't use.

Development setup

You need Python 3.11 or newer and nothing else: tuxcast has no runtime dependencies.

git clone https://github.com/achedon12/tuxcast.git
cd tuxcast
make install   # .venv with tuxcast in editable mode, ruff, pytest and MkDocs Material
make check     # lint, tests and the documentation build, like CI

make help lists every target. .venv/bin/tuxcast runs your working copy.

Project layout

src/tuxcast/
├── cli.py             commands and their output
├── system.py          the fakeable view of the machine every probe goes through
├── doctor/            one module per doctor topic
├── convert/           Elgato reader, action mappings, keyboard layouts
├── bridge.py          Discord IPC bridge
├── audio.py           PipeWire sinks
├── deck.py            Stream Deck hardware
└── patch.py           patch keeper
docs/                  the website, also published as the wiki
tests/                 pytest suite, with synthetic Elgato profiles and fake machines

Adding an Elgato action mapping

Mappings live in src/tuxcast/convert/actions.py. A one-to-one mapping is a line:

MAPPINGS["com.elgato.twitch.marker"] = simple(f"{TWITCH}::Marker")

When settings need translating, register a function with @maps("<elgato uuid>") that returns StreamController actions. Then:

1. add a test in tests/test_convert.py,
2. add the row to docs/reference/elgato-actions.md (a test checks that every mapping is documented),
3. note it under Unreleased in CHANGELOG.md.

Adding a doctor check

Each topic is a module in src/tuxcast/doctor/. A check receives a System and yields findings (ok, info, warn, fail, skip); a warning or failure should always say how to fix it. Read files with system.read()/system.path() and run commands with system.run(), never directly: that is what lets tests fake the machine (see tests/test_doctor.py). Checks must only read: the doctor never changes anything.

Pull requests

• One topic per pull request, with tests for behaviour changes.
• make check passes.
• User-facing changes are noted in CHANGELOG.md and, when relevant, in docs/.
• Commit messages describe what changes for users, in the imperative ("Detect Vesktop's socket", not "fixed stuff").

By contributing, you agree that your contributions are licensed under the MIT license.

---

_{Generated from docs/reference/contributing.md. Edit it there: this page is overwritten at each change.}