- From: Jan-Ivar Bruaroey <jib@mozilla.com>
- Date: Wed, 27 Aug 2014 14:33:07 -0400
- To: Harald Alvestrand <harald@alvestrand.no>, public-media-capture@w3.org
On 8/27/14 11:24 AM, Harald Alvestrand wrote:
> Jan-Ivar, the two cases we've been discussing are:
>
> - Tell the difference between "zero, one and more" of a device class,
> and shape UI accordingly
> - Tell if the device that was used the last time is still plugged in
> ("persistence of preferences")
>
> In both cases, doing this without actually asking the user to allow
> use of a camera/microphone (so calling getUserMedia is not an answer).
Thanks for reminding me about the use-cases.
> Neither is solvable with "current API minus this feature".
>
> Both can be solved by other APIs too (of course), but both are also
> solved by the current API proposal.
These are just observations if we do care about fingerprinting:
Intuitively, the second use-case of helping a site you've granted
(presumably non-permanent) access to in the past seems like it should be
solvable without giving same-origin device ids to sites you've never
visited before. For instance, UAs could track sites it has granted
non-permanent access to in the past (much like it tracks permanent
access) and limit this information to them. This might more closely
match the rationale for why it is reasonable.
The first use-case makes sense to me, except we're exposing "how many of
each" which is slightly more bits than (three-state) "zero, one or
more", though I suppose few people have more than two devices so it
hardly matters (though tinfoil-hat users maybe tend to have more?)
On a side-note, do we plan on exposing "virtual devices" like
screensharing in the future or just hardware sources? I'm a little hazy
on the terminology, especially now that getUserMedia() hangs off
MediaDevices.
.: Jan-Ivar :.
Received on Wednesday, 27 August 2014 18:33:34 UTC