Re: [rtcweb] Minutes for the Data Channel Presentations

[Randell Jesup <[e-mail]>]

On 4/8/2013 3:13 PM, Bernard Aboba wrote:
> Ted --
>
> I have gone over the minutes from the DataChannel sessions.  It 
> appears to me that there are some issues with the minutes.  Comments 
> below on Randell's session.  More to follow.
>
> Presenter: Randell Jesep Slides:
> http://www.ietf.org/proceedings/86/slides/slides­86­rtcweb­5.pdf

Sigh.  I was waiting for that misspelling to occur.... :-(

>
> [BA] What specific drafts did this presentation relate to?  I think 
> it's quite important to be specific, since comments during the session 
> need a context.  For example, did this presentation relate to the 
> individual submission draft-jesup-rtcweb-data-protocol, the RTCWEB WG 
> work item draft-ietf-rtcweb-data-channel, or the MMUSIC WG work item 
> draft-ietf-mmusic-sctp-sdp?  Since the presentation refers to SDP,  at 
> least the MMUSIC WG work item seems relevant, no?

draft-jesup-rtcweb-data-protocol (as stated right above what you copied, 
on the other side of a page break).  It *impacts* 
draft-ietf-mmusic-sctp-sdp.

> Richard Ejzak asked what is our bottom line? He sees this as going 
> backwards
> from interim. SDP negotiation removed
>
> [BA] Randell's proposal did not propose to remove all SDP negotiation. 
>  It leaves negotiation of DTLS port and SCTP associations in SDP, but 
> removes SDP negotiation of streams. It seems important to be specific 
> about what was being discussed (e.g. modifications to 
> draft-ietf-mmusic-sctp-sdp), so as to be able to understand what 
> commenters were referring to.

Correct.  "Initial SDP negotiation of individual DataChannels removed" 
would be more accurate.

>
> Peter Thatcher commented that he thinks the presenter's proposal is great
>
> [BA] It is hard for me to understand what Peter meant without some 
> additional context (e.g. was the comment only about the proposal to 
> remove stream negotiation from SDP, or potentially a comment about 
> draft-jesup-rtcweb-data-protocol?)

The implied context in the room was that what was presented and in the 
draft (data-protocol) was great.

>
> Randell added that the solution DO have a protocol parameter with this 
> still.
>
> [BA] What "solution" is being referred to here?  The proposal in 
> draft-jesup-rtcweb-data-protocol?   A post to the mailing list?   The 
> participants were discussing how streams are set up, but it wasn't 
> clear to me what was being referred to.

I think this quote is confused.  (We're talking about the whole RTT 
thing here mostly).  I *think* I said we *do* have a protocol parameter 
on channels, and so that can be used to handle channels created after 
the initial offer/answer.  I also stated earlier that I'd responded to a 
very similar comment on the list with a more detailed response (the 
previous night, so people might not have read it yet).

>
> Harald Alverstrand stated that he Agree with this. Glad got data 
> channel negotiation out of the document.
>
> [BA] Specifically I believe Harald was referring to Randell's proposal 
> to remove stream negotiation from draft-ietf-mmusic-sctp-sdp.  Since 
> stream negotiation is still in the document, "got" is not correct.

Yes, given this wasn't a joint session and we didn't update the other 
draft in time for the cutoff, technically that's correct, but in 
practice the SDP is the way to do negotiation for DataChannels as 
defined in rtcweb, so regardless if the change had been formally applied 
yet they're linked.

>
> Salvatore Loreto stated that he like this proposal ­ should negotiate 
> only one way and should not use SDP.
>
> [BA] Again, I believe that this meant "should not use SDP for stream 
> negotiation".

Correct.  (And Salvatore is the other author on the mmusic draft.)

>
> Jerome Marcon added that the draft does not really contain all this 
> material now, therefore text needs to be added. Randell agreed that 
> there need to be added
>
> [BA] What does "the draft" refer to here?  draft-ietf-mmusic-sctp-sdp? 
>  That makes the most sense to me since stream negotiation was still in 
> there.

Unclear, but it may well refer to the mmusic stuff, and might also refer 
to the even/odd details.

>
> Chairs asked how many had read the draft, which was couple of dozen. 
> Chairs
> urge people to read and report new issues to the mailing list. Request 
> that
> Randell takes the open issues in separate emails to the list.
>
> [BA] Again what is "the draft" referring to here?  Since 
> draft-ietf-mmusic-sctp-sdp didn't contain the material that was 
> presented, was this what the chairs were referring to?  Or did they 
> mean draft-ietf-rtcweb-data-channel?  I have read all the above docs, 
> and have gone over the MeetEcho session, and still don't understand 
> what was being referred to here, (or even why it was relevant).

draft-jesup-data-protocol (again, stated above the page-break in the 
notes, and on the title page of my slides)

>
> Salvatore commented that one open issue is if we are going to define
> prioritization between channels. Randell responded that prioritization is
> tricky as it interacts with congestion control
>
> [BA] Prioritization is discussed in draft-jesup-rtcweb-data-protocol, 
> but not in the other documents, so was Salvatore referring to that 
> document here?
>
> Randell responded that he has no issue of adding this to the list of 
> open issues.
>
> [BA] Open issues on what document?
>

Ditto the comments above.  draft-jesup-data-protocol

-- 
Randell Jesup
[e-mail]