Re: [mediacapture-main] Pull Request: Extend iframe with a new allowusermedia attribute (issue: #268)

On 05/02/16 23:12, Jan-Ivar Bruaroey wrote:
> On 2/5/16 9:35 AM, Stefan Håkansson LK wrote:
>> On 05/02/16 12:29, Martin Thomson wrote:
>>> https://docs.google.com/document/d/1iaocsSuVrU11FFzZwy7EnJNOwxhAHMroWSOEERw5hO0/edit
>> It seems to me that the essence of the "floating idea" is that only the
>> top level origin should be shown in user prompts (even if the request to
>> use certain resources is made by an iFrame).
>>
>> This seems mostly in line with PR #313. IIUC, #313 adds that any page
>> embedding an iFrame must deliberately set the allowusermedia attribute
>> for that iFrame to be able to ask for user media. And if it is the top
>> level origin only that will be shown in the user prompt, it makes total
>> sense to me if the top level origin can prohibit an iFrame to ask for
>> user media.
>
> The floated idea seems to conflict with any "on-by-default" for access
> in iframes whatsoever, so I trust we don't have that in PR #313?

Unless I am totally mistaken, #313 introduces "off-by-default" for 
access in iFrames. Currently there is no way (defined in the spec) the 
parent origin to control if an iFrame can access microphones/cameras 
using gUM, #313 introduces a control (and as said permission is off by 
default).

> I
> thought I saw discussion about maybe there being a more permissive
> default in some cases, specifically, if no iframe permission parameters
> were specified at all, or did I misunderstand? I'm not very familiar
> with this part.
>
> About the floated idea: The fusing of permissions between iframes and
> their top origin sure simplifies, but the move of control from users to
> the site concerns me. I fear this will lead to demand for browser
> policy-settings to control third-party permissions the way we have today
> for third-party cookies...
>
> .: Jan-Ivar :.
>
>> There is a part in #313 that would go away: it is said that the iframe
>> "needs to identify itself in the security prompt presented to the
>> user.", and that would go away. But to me #313 seems sensible for now,
>> we can remove that part later if the "floating idea" is adopted.
>
>

Received on Saturday, 6 February 2016 12:26:13 UTC