Re: [rtcweb] Call for review

[westhawk <[e-mail]>]

> On 6 Mar 2019, at 23:55, Justin Uberti <[e-mail]> wrote:
> 
> 
> 
> On Wed, Mar 6, 2019 at 2:11 AM westhawk <[e-mail] <mailto:[e-mail]>> wrote:
> 
> 
>> On 6 Mar 2019, at 03:34, Justin Uberti <[e-mail] <mailto:[e-mail]>> wrote:
>> 
>> We'll always mask any local addresses with mDNS, since we don't know if they're public or not. We will also provide a srflx candidate with the actual address should STUN tell us that information.
> 
> So the ‘always’ should be conditional on how the addresses are ‘found’?
> Perhaps a definition of Local IP addresses would help:
> “Local IP addresses means addresses discovered by interrogating the operating system for
> a list of available interfaces and their associated IP addresses”  
> 
> Plus a clarification that “these rules do not apply to addresses that are subsequently found via
> STUN or ICE - note that this may cause an address to be listed twice - once as a host candidate with a masked mdns
> and a second time with it’s IP address as a reflex candidate”
> 
> Perhaps this could be simplified by simply referring to 'host' IP addresses, where 'host' has the same meaning (i.e. local interface) as in ICE. 

Yep, that is probably clearer.

> 
> (I’ve a feeling there is section of the ICE RFC that talks about eliminating reflex candidates that duplicate host candidates 
> but can’t find it)
> 
> There is, but I content it does not apply here, since these srflx candidates will contain a different (non-hostname) address.

Yep, but I think it might be worth a clarification note in this document - unless it is clear to everyone else that the de-dupe is done by host name not by underlying address. - Essentially this is a clarification on what order the processing is done across multiple RFCs.

T.