[rtcweb] 答复: 答复: 答复: 答复: Relaying for privacy

My mistake. It should be at “peerconnection” rather than “getusermedia” in order to get what we meant.

发件人: Martin Thomson [mailto:martin.thomson@gmail.com] 
发送时间: 2012年5月18日 9:05
收件人: 邓灵莉/denglingli
抄送: rtcweb@ietf.org; Eric Rescorla
主题: Re: [rtcweb] 答复: 答复: 答复: Relaying for privacy

I do not think that any option other than what Eric described is truly feasible. That is, of the options provided, only the one where the site has to be trusted bit to reveal your IP is likely to succeed.

getUserMedia provides a cue of a fundamentally different nature to that you are looking for in this use case. getUserMedia signals that the user is ok with the site gaining access to their camera/microphone.  That is a contract between the user and the site, not the two users.

If you don't trust the site, then you are in Torbutton land. That is, specific UI for this.

I've concluded that we don't need this requirement. There is little that a browser can sensibly do beyond something like Torbutton.

And I'm not convinced that trying to replicate, or require the replication of Torbutton, is a great idea either. Tor users understand the value it provides, most other users won't care for the added complication.

On May 16, 2012 10:59 PM, "邓灵莉/denglingli" <denglingli@chinamobile.com> wrote:
>
> Sorry, by " In either cases, the setting can be done through getusermedia as Martin said." I mean " In either cases, the setting could be done through getusermedia as Martin said."
> Since according to our experience with chrome so far, media relaying through turn is not yet supported.
>
> Lingli
>
> -----邮件原件-----
> 发件人: rtcweb-bounces@ietf.org [mailto:rtcweb-bounces@ietf.org] 代表 邓灵莉/denglingli
> 发送时间: 2012年5月17日 13:51
> 收件人: 'Eric Rescorla'
> 抄送: rtcweb@ietf.org
> 主题: [rtcweb] 答复: 答复: Relaying for privacy
>
> It makes sense to me.
> However, considering the relaying burden would not be trivial, I'd further prefer that the enable/disable button be dedicatedly set for the RTCWEB instance in question, rather than the whole pack of connections of the browsers.
> In other words, it is better to be triggered at the time of page-loading for the RTCWEB app for a global configuration (keeping current IP privacy for all callers) or when called by for a specific configuration (keeping privacy for the incoming caller).
> In either cases, the setting can be done through getusermedia as Martin said.
>
> BR
> Lingli
>
> -----邮件原件-----
> 发件人: Eric Rescorla [mailto:ekr@rtfm.com]
> 发送时间: 2012年5月17日 10:07
> 收件人: 邓灵莉/denglingli
> 抄送: Justin Uberti; <rtcweb@ietf.org>
> 主题: Re: 答复: [rtcweb] Relaying for privacy
>
> Yes. Tor is too slow to do voice effectively. I would expect torbutton or a torbutton-like to route media through a turn server. This will provide worse privacy than onion routing but better than just letting the server control the routing.
>
> Ekr
>
> On May 16, 2012, at 18:06, 邓灵莉/denglingli <denglingli@chinamobile.com> wrote:
>
> > Hi, Ekr
> >
> > I agree with you about the Martin's comments.
> > For the proposal of Torbutton, I have a concern about the employment
> > of union routing (encrypting and decrypting several times between the
> > relaying nodes along the way) in Tor would be not acceptable to the
> > RTC requirements of VoIP use-cases.
> > Would this be an issue?
> >
> > BR
> > Lingli
> >
> > -----邮件原件-----
> > 发件人: rtcweb-bounces@ietf.org [mailto:rtcweb-bounces@ietf.org] 代表 Eric
> > Rescorla
> > 发送时间: 2012年5月17日 6:10
> > 收件人: Justin Uberti
> > 抄送: rtcweb@ietf.org
> > 主题: Re: [rtcweb] Relaying for privacy
> >
> > On Wed, May 16, 2012 at 2:54 PM, Justin Uberti <juberti@google.com> wrote:
> >> My view was that we need to make it _possible_ for sites to respect
> >> privacy (via the API), and provide documentation that encourages
> >> sites to do so. As you point out, I think it's difficult for the
> >> browser to automatically figure out what to do here, and when to do it.
> >>
> >> The identity of the "called" user may not be a factor in certain
> >> applications (e.g. some p2p game), making it unclear as to whether
> >> forcing use of relay is even the right thing to do.
> >
> > I concur with all this.
> >
> > I think it's also important to reiterate that this is about the site
> > cooperating to protect your privacy. If you want to protect the user's
> > IP address from the site, then the user needs to use Torbutton or some such.
> > And I would want Torbutton to arrange relay my RTCWEB traffic as well
> > (though likely that can't go through Tor for performance reasons).
> >
> > -Ekr
> > _______________________________________________
> > rtcweb mailing list
> > rtcweb@ietf.org
> > https://www.ietf.org/mailman/listinfo/rtcweb
> >
> > __________ Information from ESET NOD32 Antivirus, version of virus
> > signature database 7138 (20120515) __________
> >
> > The message was checked by ESET NOD32 Antivirus.
> >
> > http://www.eset.com
> >
> >
> >
>
> __________ Information from ESET NOD32 Antivirus, version of virus signature database 7138 (20120515) __________
>
> The message was checked by ESET NOD32 Antivirus.
>
> http://www.eset.com
>
>
>
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb
>
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb