Re: [rtcweb] WGLC for draft-ietf-rtcweb-ip-handling

[Justin Uberti <[e-mail]>]

Correct, and I think there are two distinct scenarios being conflated in
this discussion.
1) Web pages who aren't using WebRTC for communication, and therefore don't
call getUserMedia (Bernard's scenario)
2) Web pages who are using WebRTC for communication, and typically do call
getUserMedia.

This document attempts to prevent sites in scenario 1) from having access
to all IP addresses, and I believe does so successfully.

There is of course the question of whether sites in scenario 2) should have
access to all IP addresses. Some observations:
1) This problem is not specific to web browsers. Existing mobile apps
already have access to all IP addresses.
2) The 'right' behavior here is situation-dependent. As an example, should
a user on a VPN send their media traffic over the VPN or not? One can
construct a valid rationale for either position.
3) In the event where specific behavior is desired, this document does
provide specific modes to allow browsers or extensions to select the
desired behavior.

As such, I believe a full treatment of scenario 2 is out of scope of this
document, and the existing options presented in the document are sufficient.


On Thu, Mar 29, 2018 at 2:29 PM Bernard Aboba <[e-mail]>
wrote:

> Stephen Farrell said:
>
> "So it's all a bit of a fig leaf IMO."
>
> [BA] Correct me if I'm missing something, but there are a class of abusive
> applications that don't even pretend to support realtime communications and
> don't call getUserMedia (probably so as not to give the user an inkling of
> what they are up to).
>
> So while there is some leeway to argue about the meaning of "consent" for
> an application that actually utilizes the WebRTC APIs for communications
> between browsers, there are a class of abusive applications that don't
> fulfill even the most liberal definition of "consent".
>
> On Wed, Mar 7, 2018 at 2:51 PM, Stephen Farrell <[e-mail]
> > wrote:
>
>>
>>
>> On 07/03/18 22:35, Justin Uberti wrote:
>> >>> In addition, while gUM consent is given as an example, it is not
>> >> normative.
>> >>>    Mode 1 MUST only be used when user consent has been provided.  The
>> >>>    details of this consent are left to the implementation; one
>> potential
>> >>>    mechanism is to tie this consent to getUserMedia consent.
>> >> Sure. OTOH, IIUC, that is what's done in web browsers so it kind
>> >> of really is normative, in practice. Again, apologies if there
>> >> are other things done in browsers.
>> >>
>> > I believe that the Brave browser uses Mode 4 in its private browsing
>> mode.
>> > https://github.com/brave/browser-laptop/issues/260
>>
>> Interesting, didn't know that and good to see.
>>
>> OTOH, still the vast majority of web browsers will I assume
>> do mode 1 for the vast majority of webrtc calls.
>>
>> So it's a bit of a fig leaf IMO.
>>
>> S.
>>
>>
>>
>> _______________________________________________
>> rtcweb mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/rtcweb
>>
>>
>