Re: noaccess / peerIdentity as constraints

On 7/11/13 6:23 PM, Martin Thomson wrote:
> On 10 July 2013 23:24, Stefan Håkansson LK
> <stefan.lk.hakansson@ericsson.com> wrote:
>> [...] if the end user can not verify that the app has only
>> "noaccess" access to the media, this has little value. The app can just
>> omit the "noaccess" constraint and do whatever it wants - the user would
>> not know.
>
> This is a tricky user interaction problem.  It's very easy to provide
> too much information.  And this particular case is very hard to
> describe simply.
>
> The problem is that the site can send the media anywhere.  So, prior
> to actual sending (and authentication), there isn't a lot that you can
> tell a user that would make a real difference.
>
> Of course, once the stream is sending over an authenticated channel,
> it's possible to identify where it is going.  Sort of.  We don't
> restrict the stream from going to multiple destinations either, so
> it's entirely possible that the stream is being recorded by the site.
> And a clever site could cease transmission during the times that the
> user is checking to prevent them from learning this fact.
>
> In short, I can't imagine a good story to tell a user about a noaccess
> stream, until it reaches a remote peer.

I agree. To me it seems we should spec up the "noaccess" constraint 
(presumably as part of the getUserMedia operation), and how 
MediaStream's with this property are special (including rejecting 
"addTrack"), and then put a note about that the main usefulness is when 
such MediaStream's are used with PeerConnection.

>

Received on Friday, 12 July 2013 07:42:07 UTC