Re: When the light goes on

On 19/05/14 22:54, Martin Thomson wrote:
> We aren't going to say anything other than:
>
> [active] The browser MUST provide noticeable indicia when actively
> capturing media from a device.
>
> [potential] The browser MUST provide indicia when a site has a nascent
> ability to capture from a device without a user consent prompt.

I like this. Simple and understandable.

>
> This has several ramifications: persistent grants of consent will show
> the second indicator persistently.  This could be used as a hook to
> enable revocation of consent.
>
> Non-persistent grants of consent can be paused somehow.  We didn't
> agree on the precise control surface.  I have proposed the use of
> "MST.enabled" for this.  That causes the active indicator to disappear
> but the potential indicator remains.  For a non-persistent grant, only
> the track ending causes the indicia to disappear.

I fail to parse the above completely. Should the first "non-persistent 
grant" say "persistent grant"?

And, for clarity, the active indicator would of course only disappear if 
all MSTs that use the source are disabled.



>
> There was a concern that the browser would be unable to reacquire a
> lock on the device once the track is resumed.  This is a valid
> concern, but one that exists any time that the hardware light goes off
> anyway and it was determined that the hardware light going off was a
> desirable property from this solution.  The browser might implement a
> browser-local lock that would prevent other sites from acquiring the
> camera, which wouldn't protect from other applications on the machine
> grabbing the device, but it would prevent some failure modes.
>
>

Received on Thursday, 22 May 2014 12:57:39 UTC