[avtext] Proceedings from IETF 91 (Honolulu)

["DRAGE, Keith (Keith)" <[e-mail]>]

(As WG cochair)

Somewhat belatedly find the proceedings from this meeting below.

Regards

Keith

Tuesday 12th November 2014, 15:20 - 17:20 HST (Room: Kahili)

Chairs: Keith Drage / Jonathan Lennox

Agenda bash, status update, and review of related protocol items (10 min)
-------------------------------------------------------------------------

Slides: slides-91-avtext-3.ppt

Admin slide (slide 2).

Note Well (slide 3).

Agenda Bash (slide 4) no comments.

Status: Splicing notification (slide 5).

One new WG document: draft-ietf-avtext-splicing-notification-00.

Two reviews received - Bo Burman, Stephan Botzko. Chairs would like to have at least one  more.

Colin Perkins (via Jabber) agreed to provide a review.

After that (assuming all issues raised by reviews are resolved) the document should be ready  for WGLC.

Status: draft-uberti-payload-vp9-00 (Slide 6).

Defines RTP payload for VP9.

Also proposes new AVPF feedback message: LIR "Layer Intra Request" for partial refresh of  SST scalable media.

Review from AVTEXT would be welcome. Likely useful as a general-purpose mechanism.

Stefan Wenger agreed to provide a review.

Taxonomy (30 min)
------------------

Presenter: Bo Burman

draft-ietf-avtext-rtp-grouping-taxonomy-02

Slides: slides-91-avtext-0.pdf

[Slide 2]

No update since IETF90

[Slide 3]

Add definition for "multimedia conference".

No objection from meeting.

New definitions for layered coding transport. seems like there is agreement on the list for  this.

[Slide 4]

No need for common term for "media source or media sink"

Jonathan: "all RTP stream sbetween media source and sink" is not too long and therefore OK.

No need for a term for "all RTP streams between a media source and media sink".

[Slide 5]

Keith: WG is satisfied that the document is complete.  
AD: IETF LC is sufficient for inter-WG review.  

Way forward: Will have a -03 version with conclusion from meeting, will be available this  week. Ready for WGLC when new document available (will have three week WGLC due to end of  meeting and bridging thanksgiving).

[Slides 6 through 10 not presented - backup slides only]

RTP Stream Pause / Resume (30min)
---------------------------------

Presenter: Bo Burman

draft-ietf-avtext-rtp-stream-pause-05

Slides: slides-91-avtext-1.pdf

Has already had WGLC.

Close all open issues to enable publication request.

[Slide 2] 

IPR disclosure from version 0.

[Slide 3]

Changes since 01

Clarification on unsolicited PAUSED / TMMBN=0.

Major re-write of SDP signaling

Change refuse to refused - indication

Emil Ivov (via Jabber): did we decide ack to unsolicited post message.

Presenter: coming to it next slide

[1st point on Slide 4] 

Open issue: Do we need ACK for unsolicited PAUSED.

Roni: two directions: just repeat message vs. ack.  Timing/ repetition rules needed
 
Mo: no ack makes more sense - no need for acknowledge, no need for a state lock. Transient  advisory message. Send periodically as update.

Jonathan: Mechanism is designed for group communication. From whom in group do you expect  ACK? Do you expect ack from the whole group? In which case how to know?

Bo: agree this is why repeat is better.

Emil (via Jabber): TMMBr has TMMBN, why not ack.

Roni: tmnbn is not really an ack, it's more a bandwidth negotiation thing.

Keith: people are happy with multiple transmissions?

Emil (via Jabber): no objection to not adding ACK.

Roni: no Ack, so now we need repetition rules.

Bo: proposal on list was to use loss rate as metric. 

Stephan: don't limit number.

Roni: likes idea of text saying tune it to loss rate (informative)

Keith as Chair: Will not have ack. Will have informative text for repetition rules  indicating "tune it to loss rate".

[2nd point on slide 4] 

retransmission interval for unsolicited PAUSED?

Jonathan: limit of RTCP timing rules as sufficient?

Emil (via Jabber): one reason to recommend.

Colin (via Jabber): every RTCP interval?

Stephan: no over specify.

Bo proposal: send with evert RTCP compound packet.

Bo proposal #2: add : "you can send as often as you wish as long as you do not exceed limit  (informative).

Action: send it with every compound RTCP packet as long as want and may send it while not  exceeding the RTCP bandwidth.

[Slide 5] 

Chair: any other open issues.

Room: no.

Action: Editor will spin a new version with these changes, then will have one week second  WGLC.

Header Extension for SDES Items (20 min)
----------------------------------------

Presenter: Mo Zanaty

draft-westerlund-avtext-sdes-hdr-ext-02 

Decide whether should be adopted as WG item.

Slides: slides-91-avtext-2.pdf

[Slides 1 through 7 presented] 

Will be used for critical SDES items required to process media immediately and fate share  with media.

It was brought in various WG (RTCWEB. MMUSIC, CLUE)

The draft is currently registering CNAME. MID is registered in the bundle draft. 

The syntax is just put the header extension in an RTP header extension.         

[Slide 8]

Propose to adopt as AVTEXT WG item.

Keith: is this the right direction forward?  Have concerns be addressed?

Harald: no concerns with present scope of draft.

Bernard: no concerns.

Rob Hansen: would like to use it.

Humm for adopt.  Unanimous in favor  Confirm on list.

Chairs will ask for milestone.

16:50 Close
-----------

Stephan: There is payload issue, H265 payload vs. updated terminology in taxonomy. People  asked to look at it.

Thanks to Stefan Wenger and Roni Even for providing notes for the meeting. Thanks to Harald  Alvestrand for acting as Jabber scribe.