Minutes of the RAI Open Area Meeting at IETF 81 =============================================== Minutes edited by Gonzalo Camarillo Based on notes by Keith Drage and Allyn Romanow Meeting chaired by Robert Sparks and Gonzalo Camarillo Slides presented included in the proceedings Tuesday, July 26th, 1710-1810 ----------------------------- Topic: Survey results so far of the process in RFC 5727 Presentation by Chairs The chairs presented a survey of the results of the DISPATCH process defined in RFC 5727. The RAI area have been using that process for around 2 years. Topic: Open discussions Discussion led by Chairs In general, the DISPATCH process seems to be working fine in the opinion of most people. People provided feedback on a few issues in order to further improve the process. The development of SDP extensions needs to be coordinated. Two options for such coordination are to have MMUSIC develop all SDP extensions or to create an SDP directorate to review all extensions, which could be created in different WGs. Having a list with all ongoing SDP extensions or having all authors of SDP extensions add SDP as one of the keywords in their drafts could also be useful. Having several mailing lists to discuss a topic while it is being dispatched can be confusing. Two options for avoiding such confusion are not to create any mailing list while a topic is being discussed in DISPATCH or to make it very clear that discussions related to chartering (e.g., about the scope of the effort) should take place on the DISPATCH list while technical discussions (e.g., the properties of different possible approaches to the problem) should take place on a separate list. Once they are chartered, some WGs do not have enough energy to progress the work described in their charters. It was mentioned that such low-energy WGs do not disrupt the operation of the area. They simply keep their responsible AD checking their status periodically. However, the chartering of those WGs can consume significant AD, IESG, and community resources. Therefore, WGs-to-be need to show the potential to have enough energy in order to be chartered. It was mentioned that virtual interim meetings help WGs make progress. A few WGs have been making good use of such virtual interim meetings. The DISPATCH deadlines are sometimes too tight. One option is to two deadlines, one for work that might need a BOF and one for other work. Some people find it difficult to keep track of all the new WGs being chartered. Two options are to automatically subscribe people who were active in the chartering discussions on the DISPATCH list (opt-out solution) and to highlight the chartering of new WGs on the DISPATCH list so that interested parties can subscribe to their lists (opt-in solution). The acronyms of new WGs need to consider the tradeoff between being meaningful and being cute. Collaborating with W3C (in particular in the area of RTCWeb) can be helpful to update some of our internal processes (e.g., registration of protocol values) to a fast-paced environment like the web. The meeting ended. End of Minutes ============== Raw notes by Keith Drage and Allyn Romanow follow: Notes by Keith Drage ==================== [slide] RFC5727 Intended to improve handling of new work proposals and ease completion of work taken on Created the DISPATCH process The whole area as its scope WG Chartered in April 2009 First official meeting in July 2009 (IETF 75) Updated the SIP Change policies P-headers were deprecated Easier registration of header fields for experimental or private use Informational RFC or external document Designated Expert Clearer registration for event packages RFC Required Designated Expert [slide] DISPATCH Structure Decisions on how to handle particular work proposals are made in DISPATCH The actual work is executed elsewhere As a WG, a BOF, an AD-sponsored RFC, ... [slide] DISPATCH Track Record 1 Post BoF charter review (atoca) 3 BoFs recommended (rtcweb, codec, e2md) 2 Dispatched to existing WG 9 New WG recommended (8 formed) 2 Progression as AD sponsored recommended 4 No further work recommended (and several ongoing conversations) [slide] Dispatched Existing working groups Sound level indication (AVT) Action referral using REFER (Splices) AD sponsored Reason in responses SIP Forum User-Agent Config No further IETF work recommended Profile datasets DTMF Info Interconnect guidelines Q4S [slide] Ongoing Conversations Session Identifier SIPSCOTCH charter not advanced, but conversations on the general problem still being discussed [slide] Groups Formed Name Formed Name Expanded sipclf Sep-09 SIP Common Log Format martini Dec-09 Multiple AoR reachabiliTy InformatioN Indication codec Jan-10 Internet Wideband Audio Codec siprec Mar-10 SIP Recording soc May-10 SIP Overload Control salud Jul-10 Sip ALerting for User Devices splices Jul-10 looSely-couPLed sIp deviCES atoca Aug-10 Authority-to-Citizen Alert cuss Aug-10 Call Control UUI Service for SIP clue Jan-11 ControLling mUltiple streams for tElepresence vipr Mar-11 Verification Involving PSTN Reachability rtcweb May-11 Real-Time Communication in WEB-browsers [slide] Message Trends [slide] Documents and Milestones [slide] Meetings [slide] Observations SIPCLF (Sep09-) Took significantly longer than expected Close to pubreq on all documents Martini (Dec09-Mar11) Heavy use of virtual interims Heavy use of trac Siprec (Mar10-) Heavy use of virtual interims SIXPAC Insufficient energy to complete forming group Atoca (Aug10-) Extremely low activity Salud (Jul10-) and Cuss (Aug10-) Low activity warnings at IETF80 Cullen: Commented on the short time to completion of MARTINI. [slide] Discussion What has worked well in the DISPATCH decision process? Where has it helped? What isn't helping as well as we hoped? Is DISPATCH helping us produce better charters? Is DISPATCH making the best decisions so that the actual work gets done effectively? Open mic: Spencer: Helpful was taking something kicking round before got to IETF and moving it to closure; getting the attention of the community in dispatch. Charter in dispatch was not charter that went forward with. Bernard: Nothing can do about groups that do not have activity. Problem is ones that have lot of activity but do not make progress. Where WG not making progress, then is that problem of dispatch. Mary: Cut down on WG shopping. Still got some rough edges. Mentioned SDP changes in lots of groups. Gonzalo: Do centralise [SDP] changes or do separate groups do it. Mary: Thinks in case of SDP should centralise. On splices [Mary clarified on list that she was referring to SALUD], do not think they have ever met [correction that have meant once] - no discussion on mailing list for it. SIXPAC was a glitch in the system. In dispatch should not create umpteen different mailing lists. Robert: While discussion is in dispatch should not spawn new mailing lists. Mary: Also thinks should not form new mailing lists before dispatched. On AD sponsored ones... Robert: Asked for response on 3GPP. Gonzalo: Process for mailing lists. Charter discussion currently go to mailing lists, technical to specific lists. AD sponsored. Mary: Has not seen slimlining make things happen in AD sponsored. Jon: Talk about low energy thing. Adam: When have strong external deadline then virtual interims are way to do things. Cannot keep up with mailing lists that well when there lots of them. Would like tools help in this. Robert: Do you think that filtering of dispatch is enough? Miguel: On SDP - Centralise things in MMUSIC. There are other documents that are more high level. Does not think MMUSIC should be core group for these. Spencer: Liked the statistics. Some of efforts that were dispatched were part of dispatch main meeting and some were adhocs; did that have any correlation. Cullen: Centralisation of things. The RAI area should be working so we do not have to do everying in one group. With SDP in particular the extensions all interract. Keith: .... Mary: +1 on things Adam: If SDP is fragile do we need an SDP directorate. Hadriel: Likes idea of page of SDP impacting. Gonzalo: New groups do take AD time when getting chartered so that is a downside. Robert: New charters do consume shocking amounts of time; do really have to have a need. Dean: Is dispatch process saying no frequently enough. Cullen: Has draft that dispatch did not say no to but wishes they did. (Adjacency stuff). Gonzalo: Are we attracting the relevant work to the area? Daryl: Wanted to kill a draft, but noone could say no. MARTINI: Cablelabs had dedicated person working on it. Most IETF drafts do not see that. Cablelabs do not like the fact of so many different standards organisations all working on the same thing. Likes what has happened with W3C and rtcweb. Need to do that more often. Mary: Directorate previously identified. There is a RAI area directorate, although has not been used well. Also did kill a couple of things. Bernard: Martini was a group with few dependencies. Spencer: Counted 21 things dispatched - think about from IESG perspective. Within 3 of number of BOFs within entire IETF. Jon: Tough to kill things. Fine with people going off and doing things. Thinks not active enough. More radical tool solutions to make our work happen faster. Sohel: Afraid that dispatch would close pretty soon when it opened. Now appears that have too many things. Hadriel: Gripe on how early agenda closed off on dispatch. Mary: IESG deadline for BOF proposals fixes this. Cullen: Wonders whether dispatch should have two deadlines, ones where they might need BOF and those where we assume they will not. Hadriel: Discussions on whether a deadline is a deadline or not. Richard: Has instituted three or more drafts, and process has worked. There can be tweeking but is happy. Sohel: Saturday meeting on webrtc - can have those important meetings on weekdays. Robert: That was not an IETF meeting. Keith... Meeting closed 18:07. Notes by Allyn Romanow ====================== Slides Overhauled RAI 2 years ago Is it doing what intended? How to improve it Dispatch Track record Observations Martini finished in a year - very successful Low activity groups - Atoca, Salud, Cuss Analysis of Dispatch- what works, what doesn't, effective? Mary's comments - there is a lot of overlap on SDP. ADs- would it be better to centralize? Mary - yes Mailing list is confusing - charter discussion go to dispatch, other to mailing list - confusing AD sponsored docs - couple that are languishing.. 3GPP - there hasn't been as good coordination, think there will be better now. John P.- one of the reasons to create dispatch was to handle.. low energy issues. So when he hears there are languishing, he feels that's fine, that's what the process was designed for Adam - When we have a strong external deadline, we can get things to be done, Martini, good, but it was a special case. Mailing lists- can't keep up with all of them. Get tools group, when dispatch sets up a new WG, automatically sign you up, or something like that. Miguel - mmusic. SDP. Incongruencies between groups changing SDP. Want to centralize, at least review any SDP. Other docs not just adding parameters, more architectural.. mmusic shouldn't be the core. Perhaps dispatch is the home for these architectural drafts. Spencer Dawkins - likes stats presented. Asks about ad hocs that were spun out of DISPATCH. Wants stats for them. What's the difference between an adhoc and BOF? Not so clear sometimes. Cullen - as to centralization. Thinks we should have our protocols in such a way that don't need centralization, but SDP is so special and fragile that should be dealt with in centralized way. Gonzalo - changes in RTP should be in AVTCORE. Everyone agrees Keith - page to see what is affecting SDP, would say tools, but that won't work Would like a summary of all docs in progress affecting SDP Robert S.- brainstorm. Consider to allow drafts when submitted to have keywords, could search by keyword. For protocols like SDP WG chairs could maintain list on the Wikis. This is doable says Gonzalo Low energy groups - Idea was to have some groups that didn't meet at ietf, but did have ad hocs, problem is didn't have ad hocs. When we have new names, can they have meaningful names so he can recall. Others. - but our names are cute. Mary - tried to get the wiki to work, but it's manual, tried to get people to do it for RAI interim meetings. So if it's manual won't be done. We should tTalk about SDP changes the RAI area meeting, is of broad interest. MARTINI had worked on the problem a lot before coming to RAI. Adam- if SDP is so fragile, we should have an SDP directorate Hadriel Kaplan - likes Keith suggestion to have a page showing what drafts affect SDP. Would like meaningful names. Martini- core group that really wanted to do it quickly Gonzalo - it takes a lot of work to charter a group, time he could spend elsewhere. The whole IESG has to do something. Takes shocking length of time to charter. Don't do it casually. Dean- Is dispatch saying no often enough. A goal was to filter groups. Cullen - wish they had say no to his draft about adjacency. They should've said no. Gonzalo - strategic, should we be reaching out more? Like RTCWEB. Daryl - recalls another time with a draft that should've been killed. It was Cullen's fault as AD. Martini - had a full time person working on it. But this is not the norm. Doesn't like the fact that there are multiple standards groups all are working on the same thing. It would be good if we could coordinate better. Likes what's happening with RTCWEB. Mary - directorate. We do have a RAI area directorate. It's former WG chairs. Not used enough. On the RAI wiki. Says we did kill some things Bernard - MARTINI had few dependencies. If going to focus on something, focus on dependencies, they fixed it someone. If could do so more, or get groups to cooperate more closely, would get more done quicker Spencer - 21 things accomplished. This is a large amount of stuff that has been dealt with. Thinks it's a beautiful thing. Recognize the quantity of work that has been looked at in DISPATCH. John P - tuff to kill things. He's ok with letting people go off to do things. But doesn't think we are competitive enough in the current environment. What Keith suggested web pages.. great idea. Innovate in this space, next big challenge. To deliver what rtcweb mindset wants. A permissive world, where things aren't shut down. Hadriel Kaplan - gripe on how early dispatch closes the agenda. Way too far in advance Mary - it's set by the deadline for IETF BOF proposals. Might be that we need to dispatch to BOF Cullen - when we started they thought that they wouldn't be able to tell if it needed to go BOF or not. But now we have more experience and can tell. How about having 2 deadlines now First one is for going to a BOF, second deadline is just for giving some time in dispatch. Mary - did anyone have something that they couldn't submit because of deadline? Hadriel did - Mary says what ?? .. no response Keith - there should be some flexibility. But he thinks it is there and the chairs exercise it well. Mary - added stuff to this DISPATCH meeting after deadline. Met the first deadline, not the second deadline. Gonzalo - people are on vacation and the deadline is sept 5, so it seems particularly short Richard Shockey - he authored a bunch of drafts, he's happy about the process. He's happy. Sohel - he is concerned that he has a job and can't attend a meeting on Saturday, like RTCWEB that met. It wasn't an IETF meeting. Keith - we HAVE had previous meetings on weekends. It's useful. Make sure advertised well before people by the ticket. In strong support of having such meetings. Often they are combined.