Spec entrenches current privacy harming getUserMedia / enumerateDevices systems

[pes10k]

The current getUserMedia / enumerateDevices system has significant privacy failings, which are only partially mitigated, but not fundamentally solved, through recent spec changes like requiring an activation in the frame, etc. But the pattern of "expose all the devices to the page and let the page choose" is incompatible with a range of needed privacy improvements in the platform (mainly related to fingerprinting).

When PING last reviewed those specs, the understanding was that getUserMedia / enumerateDevices couldn't be removed from the platform in the short term, but that the future direction of media on the web would be going in a different, more privacy preserving direction, with the browser mediating device selection etc.

Im not requesting in this issue for any changes in the spec at the moment. Im wondering if the WG / authors have talked with the getUserMedia folks to make sure the functionality described here is accounted for / compatible with the future direction of getUserMedia / enumerateDevices