speechsc@conference.ietf.jabber.com - 2002/11/21
[12:26] %% dburnett has arrived.
[12:27] %% dburnett has left.
[12:33] %% qzhou has arrived.
[12:35] %% qzhou has left.
[13:05] %% qzhou has arrived.
[13:05] %% qzhou has left.
[13:22] %% mrose has arrived.
[13:31] %% davidoran has arrived.
[13:32] <davidoran> meeting will start on time at 3:30
[13:32] %% qzhou has arrived.
[13:35] %% dburnett has arrived.
[13:39] <qzhou> [Eric] we need a scribe
[13:40] <qzhou> [slide] Note well
[13:41] <qzhou> speechsc meeting is on
[13:41] <qzhou> agenda on slide
[13:43] <qzhou> [eric] doc status
[13:43] <qzhou> [eric] goal and milestones
[13:43] <qzhou> [eric] time line
[13:44] <qzhou> [eric] requirement document, anyone notiice RFC 5?
[13:44] <qzhou> [eric] protocol discussion
[13:46] <qzhou> [eric] protocol evaluations
[13:46] <qzhou> [Jerry Carter] BEEP slides
[13:47] <qzhou> slides are located at http://flyingfox.snowshore.com/i-d/speechsc/BEEP.ppt
[13:47] <davidoran> for those not in the room: slides cn be accessed at http://flyingfox.snowshore.com/i-d/speechsc/
[13:47] %% qzhou has left.
[13:47] <dburnett> yes, got them.
[13:48] %% qzhou has arrived.
[13:48] <qzhou> Sorry I was dropped by bad connection
[13:49] <dburnett> what slide are we on?
[13:49] %% qzhou has left.
[13:50] %% qzhou has arrived.
[13:50] <davidoran> just finished beep
[13:50] <dburnett> thanks
[13:51] <davidoran> starting SIP - Rajuiv just got up to start his slides
[13:58] <dburnett> is there an RTSP link for when Dave gets up?
[13:59] %% qzhou has left.
[14:01] <davidoran> edric forgot to load it. I'll email it to you right now...a bit late
[14:02] <dburnett> are you done with your prez?
[14:03] <davidoran> yup.
[14:03] <davidoran> Sarvi just started in on MRCP
[14:03] %% qzhou has arrived.
[14:03] <qzhou> Sorry folks I am in very bad connection Now is in MRCP slides
[14:05] <qzhou> The slides should be http://flyingfox.snowshore.com/i-d/speechsc/MRCP_OpenIssues.ppt
[14:06] <qzhou> But I find a broken link
[14:07] <davidoran> i pulled them just fine... do a reload on the speechsc page first
[14:08] %% qzhou has left.
[14:08] %% qzhou has arrived.
[14:10] <qzhou> {sarvi] SIP allows session negotiation. RTSP address session id.
[14:11] <qzhou> [sarvi] need to work to find our which one address the problem better
[14:12] <qzhou> now slides to Web Services (Stephane Maes)
[14:13] <qzhou> [eric] pleas explan the ancroyms
[14:14] <qzhou> [stephane] {out of slides] explan SOAP, WSDL, UDDI etc.
[14:15] <qzhou> [stephane] evaluation criteria
[14:19] <qzhou> [Dave O] support of of RTP channel?
[14:20] <qzhou> [Dave O] in web service framework?
[14:20] <qzhou> [stephane] should be done in web framework.but not there yet?
[14:22] <qzhou> [mike] any web service has been used as protocol control level?
[14:24] %% davidoran has left.
[14:25] %% qzhou has left.
[14:25] %% qzhou has arrived.
[14:28] %% qzhou has left.
[14:29] %% qzhou has arrived.
[14:29] <qzhou> [mike] not clean on how RTP with session control
[14:30] <qzhou> [dave] we are working on it, example: barge-in
[14:31] <qzhou> [mike] SIP is a session setup protocol. It may not fit for control protocol
[14:32] <qzhou> [(Sarvi ] SIP and RTSP/MCRP are complimentary
[14:36] <qzhou> [Sarvi] SIP at session control level, RTSP/MRCP at fine control
[14:36] <qzhou> [eric] both sip and rtsp are good to set up session.
[14:36] <qzhou> [mike
[14:37] <qzhou> [mike] how about media negotiation
[14:37] <qzhou> [dave] construct critizism and suggestions?
[14:38] <qzhou> [erric
[14:38] <qzhou> [eric] need protocol consistence analysis
[14:38] <qzhou> [eric] different requirement against protocol.
[14:39] <qzhou> [stephane] you have T everywhere? and there are F everywhere.
[14:40] <qzhou> [dave o] how to evaluation of the framework against protocol
[14:40] <qzhou> [dave o] protocol analysis give guidance to people design protocol
[14:42] <dburnett> does the analysis have to be apples to apples comparison in order to be useful to the designers?
[14:42] <dburnett> are we trying to choose a protocol or figure out the good parts of each to use in developing our own?
[14:43] <qzhou> no. this is still evaluation. I don't think decision will be made today.
[14:44] <qzhou> [stephane] what is the outcome of evaluation?
[14:46] <qzhou> [mike] protocol choice for none-workstation design. Simplicity of design is important.
[14:46] <qzhou> [stephane] another step of analysis: list of the features which is strong from different solution
[14:47] <qzhou> [marry] explan experience from midcom
[14:48] <qzhou> [marry] some of the requirements are different to meet, therefore dimed.
[14:48] <qzhou> [marry] evaluate things to fit or not.
[14:50] <qzhou> [dave o] need to cross educate people from different world (IETF and W3C, for example -- qzhou0
[14:52] <qzhou> [dave o] poll: on how people want the protocol or not
[14:52] <qzhou> [mike] good to have it and make a guideline to help people to understand how much work have to be done
[14:52] <dburnett> please rephrase -- not clear what the poll is
[14:53] <qzhou> dave asked how useful on evaluate for different protocols/frameworks
[14:53] <dburnett> thanks
[14:54] <qzhou> [marry] from midcom example the evaluations are very subjective.
[14:55] <dburnett> sounds like discussion is being based on which protocols specific people want rather than on requirements. if so, evaluations are not useful
[14:55] <qzhou> [marry] the bottom line is the performance
[14:55] <qzhou> dave explained the evaluation as the first step.
[14:57] <qzhou> [jerry c] try extending existing protocol(s) if doesn't work, try to design something new.
[14:58] <dburnett> good suggestion. has best chance of being acceptable within IETF community
[14:58] <qzhou> [stephane] even patch/extending is not simple
[14:59] <qzhou> [dave o] we use engineering judgement.
[15:00] <qzhou> [dave o] if the requirements can be met by a protocol, even it is suboptimal on performance. do we accept it?
[15:01] <qzhou> [mike] there are long time on internet that people accept suboptimal solution, efficiency is not the optimal goal.
[15:02] <qzhou> [rajiv] from vendor
[15:02] <qzhou> [rajiv
[15:02] <qzhou> point of view, there is not much time to do a lot of evaluation.
[15:03] <qzhou> [rajiv] web framwork needs more work than SIP type of protocol
[15:05] %% Magnus has arrived.
[15:07] <qzhou> [dave o] W3C are already liason with SPEECHSC.
[15:08] <qzhou> [eric b] could web service can run on a 200 Mhz PC with 256 MB ram?
[15:08] <qzhou> [dave o] yes.
[15:09] <qzhou> the above was address to an earlier question on footprint of web service
[15:11] <qzhou> [stephane] the importance is to carry on a familiar framework to industry. the engineers will optimize that to make it workable.
[15:11] <qzhou> [piyush modi] vendors need to implement the protocol. The question will be on # channels, memory requirement, and latencies
[15:12] <qzhou> [piyush] web service is a great way to do things we couldn't do before
[15:15] <qzhou> [mike] confused on draft. it looks like speechsc is for internet servers, not terminals.
[15:15] %% mrose has left.
[15:15] <qzhou> [eric b] the initial focus is on servre to server.
[15:17] %% Magnus has left.
[15:17] <qzhou> [stephane] the performance guesses are most based on certain implementation. it may not be true if an efficient implementation comes out as HTML as example.
[15:19] <qzhou> [stephane] speechsc can cover both server and terminal
[15:20] <qzhou> [qzhou] need to be efficient and light weight
[15:20] <qzhou> [eric] IBM PDA demonstration is an example.
[15:21] <qzhou> [eric] what to do next: start protocol design along with evaluation?
[15:21] <qzhou> [stephane] can be started along with evaluation
[15:22] <qzhou> [eric] anyone object only one team.
[15:23] <qzhou> hum: majority want to have one team, and start design now.
[15:23] <qzhou> [mike] start up date
[15:24] <qzhou> [eric] worth to finish protocol evaluation document?
[15:25] <qzhou> [mike] useful as input to design team
[15:25] <qzhou> [eric] december is the reasonable time to get eval document done.
[15:35] <dburnett> did we decide on design team?
[15:35] <dburnett> (sorry, had to step away)
[15:43] %% dburnett has left.
[15:50] %% qzhou has left.