Re: [MMUSIC] Empty candidate lists in ice-sip-sdp offers and answers

[Flemming Andreasen <[e-mail]>]

On 4/29/19 5:27 PM, Roman Shpount wrote:
> On Sat, Apr 27, 2019 at 2:02 PM Christer Holmberg 
> <[e-mail] 
> <mailto:[e-mail]>> wrote:
>
>     ...
>
>     >NOTE: Since it is possible that no candidates were provided in
>     the offer,  or that all candidates in the offer
>     >where skipped due to unsupported address type or FQDN name
>     resolution failure, ICE nomination process
>     >can start with no remote candidates. This,  however, does not
>     indicate an immediate ICE nomination failure.
>     >See <<draft-holmberg-ice-pac>> for more details.
>     >
>     >In the Receiving the Initial Answer section I have added the
>     following:
>     >
>     >NOTE: Since it is possible that no candidates were provided in
>     the answer, or that all candidates in the answer
>     >where skipped due to unsupported address type or FQDN name
>     resolution failure, ICE nomination process can
>     >start with no remote candidates. This,  however, does not
>     indicate an immediate ICE nomination failure.
>     >See <<draft-holmberg-ice-pac>> for more details.
>
>     I think the text is confusing, because it starts talking about the
>     ICE nomination process. There are other things that happen before,
>     including the creation of candidate pairs.
>
>     So, I think the text should say that the agent will not be able to
>     create candidate pairs, and it will not be able to start sending
>     connectivity checks. Instead it will wait for the peer agent to
>     start sending connectivity checks, that the agent will then
>     process as peer reflexive candidates.
>
> I will try to come up with some text that uses the same terminology as 
> RFC 8445.
Do you have the suggested text ?
>
>     >I wanted to get the group's opinion regarding:
>     >
>     >1. Allowing offers and answers with no candidates
>
>     I am ok with the principle. There seem to be use-cases for it, so...
>
Agreed.
>
>     >2. Adding a reference to draft-holmberg-ice-pac regarding how
>     such session descriptions or session descriptions where all
>     candidates were rejected are handled.
>
>     The reference in the text above is misleading, because to me it
>     sounds like PAC describes how to process offers/answers without
>     candidates. PAC is only about a timer.
>
>
> Maybe this draft should cover the situation when session description 
> processing did not generate any candidates. The fact that ICE has some 
> minimal run timer is not enough. We still need something that 
> explicitly states that it is acceptable to run ICE nomination process 
> without candidates since, as far as I know, nothing in RFC 8445 
> specifies if this is allowed or not.
I am not in favor of adding a normative dependency to ice-pac at this 
late stage. Why do we need that ?

Thanks

-- Flemming (with chair hat on)
>
> Regards,
> ___________________________________________
> Roman Shpount | CTO | Cell: [phone]
> 4905 Del Ray Ave, Suite 300 | Bethesda, MD 20814
> TurboBridge
>
>
>
> _______________________________________________
> mmusic mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/mmusic