Re: [MMUSIC] Adam Roach's Yes on draft-ietf-mmusic-sdp-bundle-negotiation-49: (with COMMENT)

[Eric Rescorla <[e-mail]>]

On Thu, Apr 19, 2018 at 6:35 AM, Christer Holmberg <
[e-mail]> wrote:

> Hi,
>
> >>>>> This doesn't answer my question. Who, specifically, thinks that this
> terminology
> >>>>> of focusing on the address:port and not the m= section is good?
> >>>>
> >>>> Considering that the focus has been on the address and port since day
> one, for 7 years, I think it is good. After all, while there is more to it
> than the address:port, the main reason for BUNDLE is to be able to
> multiplex in a single address:port.
> >>>>
> >>>> Now, we HAVE discussed changing "address:port" to "transport", but it
> would still not be about the m= section.
> >>>>
> >>>>>> Having said that, if we can change it with a search/replace
> operation I am
> >>>>>> happy to discuss it. I THINK we could do that e.g., with
> "transport" or
> >>>>>> "3-tuple".
> >>>>>
> >>>>> But I don¹t want to make yet another re-write of the document just
> to once
> >>>>> again change the terminology.
> >>>>
> >>>> This document is going to be used by a lot of people. It's important
> that it be
> >>>> clear.
> >>>>
> >>>  I fully agree. But, I am not sure that endless re-writings the
> document is going to achieve that.
> >>
> >> I think this has turned into an argument about style. One usage may be
> more pleasing to the “ear” than the other, but I don’t see issues of
> technical correctness or clarity here. As I asked for the the “associated
> with” question, is anyone arguing that the word choice here is likely to
> cause material implementation errors?
> >
> > I don't think "address:port" versus "transport" will, but I do think
> "address:port" versus "m=section" or "transport parameters" will, and in
> fact I identified a material issue around trickle the other day, that I
> haven't seen a response to.
>
> My suggestion was to add text about the special case of port 9 and IP
> address 0.0.0.0.
>
> Because, when you use trickle, the offer may not contain any address
> information at all (you can send an offer without any candidates), so it is
> a special case no matter what.
>

Yes, but the point is that you cannot address the m=section parameters via
address:port, so every piece of text in the document that suggests that you
do so needs to be repaired.

-Ekr


> Regards,
>
> Christer
>
>