[15:56:04] --- Dan York has joined
[15:57:49] --- yuji has joined
[15:59:43] --- bew has joined
[16:01:31] --- spencerdawkins has joined
[16:01:48] <spencerdawkins> i will jabber scribe this one...
[16:02:13] --- agallant has joined
[16:04:04] --- deanwillis has joined
[16:04:30] --- jfischl has joined
[16:04:44] --- isudo has joined
[16:04:55] <deanwillis> thanks spencer
[16:05:27] <spencerdawkins> milestones are still agressive
[16:05:51] <spencerdawkins> ------------------- feedback of problem statement
[16:06:03] <spencerdawkins> Experience with ACH
[16:06:14] --- sal has joined
[16:06:40] <spencerdawkins> issue 1 - finding feature group - is not possible, make this a recommendation only so we don't slow down people writing drafts
[16:07:18] <spencerdawkins> people not forthcoming with detailed call flows
[16:07:51] <spencerdawkins> may be too much work, may be proprietary information. will do alternative polling techniques
[16:08:12] <spencerdawkins> problem statement is guideline, not impediment to work
[16:08:25] <spencerdawkins> jonathan - agree with "not impediment"
[16:09:53] <spencerdawkins> only about 9-10 entries for UA/proxy. some reasons for small number of submissions
[16:10:07] <spencerdawkins> 70 users have accessed results so far
[16:10:27] <spencerdawkins> robert sparks - surveys for implementers are really hard to do!
[16:10:43] --- JJ has joined
[16:12:18] <spencerdawkins> about 80 percent of results - ACH does not involve UA
[16:12:37] <spencerdawkins> 486 and 480 are most common busy codes
[16:13:05] <spencerdawkins> More response codes for DND (486 more common)
[16:13:33] <spencerdawkins> jonathan - would be interested in knowing what calling UA does with these codes (if anything different)?
[16:13:51] <Dan York> Survey slides are at: http://www3.ietf.org/proceedings/07dec/slides/bliss-2.ppt
[16:13:55] <spencerdawkins> john elwell
[16:14:22] <spencerdawkins> further feedback on mailing list? but then we can refine the questions
[16:14:33] <spencerdawkins> ACH idea was to work with a small group
[16:15:15] <spencerdawkins> francois - some questions were difficult to answer, assumed how a product would work and reflected question author's products
[16:15:34] --- bhoeneis has joined
[16:15:38] <spencerdawkins> don't just have DND, have several flavors, not sure what to map to the questions
[16:17:07] <spencerdawkins> really creative implementations might not do 486 (not acceptable) - custom features might pick obscure response codes
[16:17:39] <spencerdawkins> jonathan - survey onlist but don't respond to list (vendors in public)
[16:17:55] <spencerdawkins> make questions a bit more freeform?
[16:18:09] <spencerdawkins> do people want another survey around ACH?
[16:18:20] <spencerdawkins> ask at the end of the meeting
[16:19:23] <spencerdawkins> jonathan - great work for getting survey out and getting results that show whether we have a problem - we do
[16:19:46] <spencerdawkins> not sure we need to do it - ask about calling UA reaction but we may not need a lot more information
[16:19:55] --- ttfr has joined
[16:20:05] <spencerdawkins> francois - not sure if doing something different means we have a problem
[16:20:17] <spencerdawkins> wait for jonathan's slides
[16:21:57] <spencerdawkins> "john's slides"
[16:22:18] <spencerdawkins> ------------------------------ Autoamted handling - John Elwell
[16:22:37] <spencerdawkins> "automated" - sheesh! need another jabber scribe? :-)
[16:23:12] <spencerdawkins> ACH-analysis draft identifies a number of problem areas
[16:24:12] <spencerdawkins> interaction between proxy and UA - often UA isn't even involved, so your UA never sees the call
[16:24:37] <spencerdawkins> conflict between UAs (serial or parallel forking)
[16:24:47] <spencerdawkins> parallel less deterministic
[16:25:54] <spencerdawkins> not sure response codes are precise enough for proxies to know what happened (DND as example)
[16:26:07] <spencerdawkins> interaction with privacy on what you tell calling user
[16:27:21] <spencerdawkins> do proxies wait for all UAs to respond (with lots of variations, including scope of 6XX)
[16:27:33] --- john.zhao has joined
[16:27:39] <spencerdawkins> 6XX includes voice mail?
[16:27:52] <spencerdawkins> different ways of configuring proxies
[16:28:58] <spencerdawkins> lots of reasons to do ACH at proxy but UA-specific ACH may also make sense - answer may differ between enterprise and residental environments
[16:29:28] <spencerdawkins> problems with inconsistent configurations - how to avoid this
[16:29:58] <spencerdawkins> tell UA proxy is responsible (based on proxy)?
[16:30:38] <spencerdawkins> Jonathan - may also need to tell proxy UA is responsible ;-) in peer to peer environment
[16:32:08] <spencerdawkins> maureen - don't want my phone to ring with DND, don't want my phone arguing with my proxy with DND either
[16:32:24] <spencerdawkins> UA should win
[16:33:26] <spencerdawkins> solve what the user wants, not what the UA wants...
[16:33:33] <spencerdawkins> enterprise might have other ideas
[16:33:53] <spencerdawkins> paul - not necessarily one feature, not all devices support all variants
[16:34:32] <spencerdawkins> think configuration framework will help with inconsistent configurations
[16:35:58] <spencerdawkins> jonathan - terminating feature, on behalf of called party, but called UA isn't even involved
[16:36:30] <spencerdawkins> this is the kind of stuff where mismatches break features
[16:37:30] <spencerdawkins> configure UA with policy about retargeting, etc.
[16:38:54] <spencerdawkins> use provisioning framework as part, but not all, of the problem
[16:39:16] <spencerdawkins> make recommendations about 6XX?
[16:39:33] <spencerdawkins> (will be going question by question here)
[16:40:19] <spencerdawkins> DND draft will depend on general ACH but isn't a total overlap due to presence aspects, etc
[16:41:25] <spencerdawkins> * need to make any recommendations on proxybased vs UA-based ACH?
[16:42:01] <spencerdawkins> can't make one recommendation - explain consequences. Some later questions only make sense in one model or the other
[16:43:15] <spencerdawkins> if we don't make a specific recommendation, will thinks break? ACH is broad enough that it could be done either way, recommendation could be for either way based on targeted environments, sufficient for interoperability
[16:43:33] <spencerdawkins> if we don't say anything, that's what we've always done, that's why we're hear
[16:43:52] <spencerdawkins> christer - don't think we can do that. we don't do service specification here
[16:44:55] <spencerdawkins> are there any mechanisms to put in synchonization ?
[16:45:32] <spencerdawkins> jason - as long as UA has a way of figuring out what the proxy might do, it can tell the user - specify as little as possible
[16:46:03] <spencerdawkins> paul - interoperability between what? between caller/callee, callee could be decomposed into multiple parts
[16:46:26] <spencerdawkins> (including proxies)
[16:47:00] <spencerdawkins> jonathan - phone and proxy from different vendors work - that's the goal
[16:47:28] <spencerdawkins> dean - how do you have these features work in P2PSIP? Either someone is a proxy or you put out policy
[16:48:51] <spencerdawkins> jonathan - provisioning is crux of the problem, entirely proprietary, XCAP isn't even implemented
[16:49:15] <spencerdawkins> can't end up where something is implemented in NEITHER place - need minimum recommendation
[16:49:48] <spencerdawkins> P2PSIP should assume called UA acts like terminating proxy
[16:50:32] <spencerdawkins> need to make normative recommendation on at least one place where functionality exists
[16:51:50] <spencerdawkins> these are "must-implement" recommendations for people who want to be compliant
[16:52:04] <spencerdawkins> "if you implement RFC" - "BCP"
[16:53:31] <spencerdawkins> dean decoding what jonathan says - some discussion about wording ... don't lock text at the microphone. features that don't even interrupt the UA have to be in the proxy. If there is no proxy, behave as if the proxy is in the UA
[16:53:49] <spencerdawkins> paul - where does this get configured?
[16:54:42] <spencerdawkins> if there's no UI on the phone, configure through web interface, no phone interaction with configuration
[16:55:12] <spencerdawkins> jonathan - DND requires a protocol to turn feature on/off
[16:55:37] <spencerdawkins> * way forward - have draft, adopt as starting point?
[16:58:52] <spencerdawkins> * make recommendations on response codes?
[16:59:17] <spencerdawkins> francois - if there are better response codes, send it to SIP, but we need to look at this
[16:59:38] <spencerdawkins> assumptions on network topology (interaction with forking)?
[17:00:10] <spencerdawkins> ** when john says "proxy", he means "proxy for my domain"
[17:01:49] <spencerdawkins> dean - domain proxy plus proxy that does 302 recursion... feature doesn't work the way proxy thinks it's working
[17:02:09] <spencerdawkins> paul - need to be able to signal as many cases as there are cases, may not have enough codes
[17:02:55] <spencerdawkins> jonathan - not worried about forking and HERPF
[17:03:46] <spencerdawkins> christer - what about application server?
[17:03:58] <spencerdawkins> john - draft also covers B2BUA
[17:04:11] <spencerdawkins> christer - entity between proxy and UA
[17:04:39] <spencerdawkins> paul - interaction in forwarding, whose voice mail gets the call after forwarding?
[17:05:26] <spencerdawkins> do we need recommendations about 6XX responses?
[17:05:36] <spencerdawkins> consensus of room is "yes"
[17:06:34] --- agallant has left
[17:06:34] <spencerdawkins> * default method for UA to configure its proxy/find out how its proxy is configured?
[17:08:57] --- john.zhao has left: Computer went to sleep
[17:09:39] <spencerdawkins> jonathan - have seen anonymous call rejection as need for keeping calling UA informed about what has happened - less is more - general rule ACH is invisible to calling user
[17:10:14] <spencerdawkins> keith - stimulus via text strings?
[17:10:46] <spencerdawkins> jonathan - doesn't matter why called party rejected a call (no answer after three rings)
[17:11:00] <spencerdawkins> keith - diversion, needed new codes for voice mail
[17:11:54] <spencerdawkins> francois - if there's a reason, define a code, but the default is that there is no reason. don't want signaling via text strings
[17:13:38] <spencerdawkins> ** adopt ACH-analysis as WG draft (just next step forward), needs to become/support a proposal
[17:14:21] --- lminiero has joined
[17:14:38] <spencerdawkins> jonathan - current charter says analysis may/may not go forward for publication - adopt because we're we're working on it whether we deliver it or not
[17:15:17] <spencerdawkins> strong consensus in the room to adopt this draft - chairs will work with ADs to get the right milestones
[17:16:00] <spencerdawkins> ------------------------------- call completion draft ---------------------------------
[17:16:33] <spencerdawkins> issue - explicit call queue management, compared with dialog event package
[17:16:38] --- john.zhao has joined
[17:16:53] --- bew has left
[17:17:20] <spencerdawkins> Roles, which may be decomposed or integrated
[17:18:50] --- lminiero has left
[17:19:34] <spencerdawkins> how to deal with queue abuse? limit of queue lengths per caller, interaction with anonymous callers
[17:21:13] <spencerdawkins> jonathan - idea is that state could come back, add header for 486 about subscription
[17:21:49] <spencerdawkins> only releasing one new call when callee becomes free - interaction with anonymous callers
[17:22:20] <spencerdawkins> how to identify callers? caller identity? possibly using cookies
[17:22:56] <spencerdawkins> how to transfer call queue management support indication (allow-events, callee-capability, other ideas)
[17:23:31] <spencerdawkins> * adopt poetzl and merge with worley call completion draft?
[17:23:43] <spencerdawkins> milestone is for call queuing
[17:24:02] <spencerdawkins> hmmm - start with this draft as WG item?
[17:24:38] <spencerdawkins> keith - this isn't call queuing, as I understand it
[17:24:49] <spencerdawkins> hijacking this milestone
[17:26:09] <spencerdawkins> we've deleted call flows from poetzl draft, only analysis, worley draft has protocol recommendations
[17:26:30] <spencerdawkins> jonathan - also support adopting both drafts as a combination
[17:27:03] <spencerdawkins> also - "call-info" header field with purpose parameter (register new "queue" value and you're done)
[17:27:46] <spencerdawkins> keith - please prioritize requirements for any protocol work from other groups
[17:28:09] <spencerdawkins> ** adopt - big hum with no one opposed
[17:28:23] <spencerdawkins> francois - changing the name? that's what keith was asking about
[17:29:11] <spencerdawkins> cullen - fix the milestone, get things right, move on
[17:29:21] <spencerdawkins> design team?
[17:30:38] <spencerdawkins> -------------------------- multiple line appearances, Alan Johnston
[17:30:50] <spencerdawkins> have not revised spec, have been talking to chairs for guidance
[17:31:28] <spencerdawkins> add discussion of NOTIFY vs forked INVITE, add details of existing implementations, include call flows (anil-sippint-bla-04)
[17:31:57] <spencerdawkins> discuss flavors of feature and how they can be invoked, discuss provisioning issue
[17:33:11] <spencerdawkins> if you do something with MLA that's not in the draft, please tell authors
[17:33:36] <spencerdawkins> christer - includes service description and requirements - do we do that here? in this group?
[17:34:04] <spencerdawkins> jonathan - same answer as last IETF, not normative descriptions
[17:34:18] <spencerdawkins> will also form design team for this work
[17:34:38] <spencerdawkins> not ready for adoption yet - will ask about next version when it's available
[17:34:51] <spencerdawkins> keith - ask other SDOs for design team volunteers
[17:35:38] --- JJ has left
[17:35:46] --- jfischl has left
[17:36:06] --- Dan York has left
[17:38:09] --- isudo has left
[17:41:39] --- deanwillis has left
[17:42:45] --- kamaji has joined
[17:43:10] --- yuji has left
[17:43:33] --- spencerdawkins has left
[17:48:25] --- sal has left
[17:54:38] --- bhoeneis has left
[17:57:55] --- ttfr has left
[18:10:20] --- kamaji has left
[18:20:10] --- bdr has joined
[18:21:07] --- bdr has left
[18:24:13] --- john.zhao has left
[19:16:24] --- spencerdawkins has joined
[19:31:12] --- spencerdawkins has left
[20:09:42] --- spencerdawkins has joined
[21:08:46] --- spencerdawkins has left