btd700linux/docs/VALIDATION.md

Validation — 5 September 2026

This document separates observed hardware behavior from simulated tests. The project was developed using AI (OpenAI Codex); these results are not a warranty.

Environment

Bazzite / GNOME, Python 3.14, GTK 4.22, libadwaita 1.9. Connected BTD 700: USB 3542:3001, firmware 3.11.0.

Observed on the real device

• Correct control-interface discovery and opening without detaching audio drivers.
• Successful state, mode, codec, quality, transport, Auracast configuration/name and firmware-version reads. See PROTOCOL.md for response bytes.
• Reported live state: Gaming, aptX Adaptive, 24-bit / 48 kHz, audio playing.
• Native GTK window and registered GNOME tray item with a live status tooltip.
• A second USB client is rejected with a clear message while the app owns the device.
• The device owner subsequently confirmed that the app works. The individual functions exercised by the owner were not enumerated.

Automated verification

• Protocol/controller/transport tests: command allowlist, malformed frames, response correlation, event acknowledgements, readback, capability checks, input validation, reset confirmation and password handling using a simulated device.
• Localization tests: English/German detection, explicit CLI override, message placeholders, localized status/errors and unchanged numeric protocol data.
• Desktop integration tests: isolated installation/uninstallation and autostart behavior without changing the real user's launchers.
• tools/check_gui.py exercised English and German windows and actual D-Bus menu events against a demo device: modes, codec, transport, connection, broadcast discovery/quality, name/password saving, encryption, closing/reopening and preserving/discarding unsaved edits.
• Window layouts rendered and inspected at 620 × 800 and 420 × 600 pixels.
• README screenshots captured from the real English GTK window at 620 × 880 with fictional demo data. No real USB access or generated mockups were used.
• Extraction helper output matched the SHA256 of the original assembly used for protocol research. No vendor binaries were added to the distribution.

The CI workflow runs the unit tests on Python 3.10 and 3.14, in both languages. It does not run real hardware tests or claim that a particular receiver works.

Not yet systematically verified

• Hardware writes across different firmware revisions and all individual commands.
• Real Auracast receiver compatibility, audio quality and reconnect behavior.
• Other distributions/desktops beyond the tested Bazzite/GNOME setup.

tools/check_hardware.py --run has not been executed. It deliberately interrupts audio, changes settings including the broadcast password, and attempts to restore the original values. It does not perform a factory reset and cannot perform firmware updates. Unplugging midway can prevent restoration.