On Mon Aug 18 13:41:57 2025 +0000, Rémi Bernon wrote:
Yeah I think it could be good to avoid differences between all of the backends. The backend can provide a specific device timestamp, or the code here could make up one unique. To be clear, are we confident that USEC_INITIALIZED is identical across all of the Linux devices of the same physical device and hence usable as container ID, right?
`USEC_INITIALIZED` is different across the various udev devices created for the controller, but if we use the root USB udev device we can get a consistent `USEC_INITIALIZED` value. See [here](https://gitlab.winehq.org/cmcadams/wine/-/commit/426547f138195a57ab7b88ffb8c... winepulse.drv, and [here](https://gitlab.winehq.org/cmcadams/wine/-/commit/2bf60d670d2eb76b74b578995e8...) in winebus.