XCON Minutes IETF-70 Minute Taker: Dave Morgan Comments from the Chairs: Agenda Bash - none * IESG comments have come back on the Framework doc, we will be revising * Data model, many good comments were received on the mailing list and revisions have been made * The Conference Event Package has been adopted as a WG item Common Data Model - Oscar Novo presenting Draft-ietf-xcon-common-data-model-07 Summary: Oscar Novo: The reviewers commented on -06, version -07 was issued and reviewers said the changes were ok. The first issue that came up was confusion around a "persistent list." It was unclear what this referred to or why it is being proposed. There was discussion with Roni Even, Mary Barnes, Adam Roach and Dave Morgan at the mic. It was soon determined that this was added to accommodate the chat room analysis that Adam performed and posted to the mailing list. Adam said his straw man was trying to capture the case where the user wanted to add themselves to an active conference and the system already had information on that user that could be provided without the user having to reenter it for the new conference. This "persistent" data could be store in a "persistent user list." Adam commented that this may not even be implementable. ** Resolution: Back up and look more closely at the use case identified in the chat room doc/analysis around user registration before joining, before adding that to the data model. Oscar to notify the mailing list of how he proposes to proceed. The second issue that came up was confusion around the language identifier. Henning Schulzrinne, Mary Barnes, Adam Roach, and Migel Garcia come to the mic to ask for clarification. Henning asked if the language tag meant: 1) What language the participants will be speaking, 2) That the translation of this conference will be available in this language, or 3) That the machine prompts and instructions for this conference will be in this language. Oscar stated that the language tag define the language that is going to be used in that conference (option 1). Adam clarified that this is just a subject tag, there's no enforcement and people can switch to other languages, it only indicates what the user can expect. Adam also asked Oscar to make sure this is consistent with the use of the language tag in other RFCs. ** Resolution: Oscar will clarify the language tag more and specify that it is to be used for informational purposes that it is the language to be used in the conference. This will be posted to the list for discussion The third issue revolved around roles. Oscar reminded the group that due to confusion over the semantics of roles, they were removed from the -07 draft and that we left it to a future draft addressing policy. Mary Barnes, Adam Roach, Henning Schultzrinne, and Migel Garcia came to the mic and had a give and take discussion about putting roles into the model without any structure or semantics associated with them. Mary pressed that roles are fundamental to conferences and must be included in the data model even just as a basic definition. The group seemed to come to a conclusion that we could come to agreement on a set of pre-defined nouns. ** Resolution: Mary to summarize the issue of including roles as pre-defined nouns for the list. Adam then stated that when the above issues are resolved he will put the draft to a WGLC, and ask for a 4 week WGLC due to holidays. ++++++++++++++++++++++++++++++++++++ Conference Event Package - Roni Even presenting draft-xcon-event-package-extensions-02 Roni reiterated that in the last meeting there was agreement to take partial notification out of the data model and put in the event package. He said that the plan was to use the data model schema to update the package to enable partial notifications, that has to be done in the draft. Adam Roach asked if we want to modify schema to accomplish partial notification or do we want to come up with other methods. Options include Version tags, etc. Roni suggested that we try and follow the same methodology as RFC 4575. Adam said the granularity in 4575 may not work well with all our additions to the data model. Roni confirmed that a drawback of 4575 is that as the conference size grows, there's a lot of messaging, such as when new user joins meeting. We might have to propose something with more granularity, it's a real issue. Adam also brought up the following example: how do you know you have most recent view of the conference as a user. All agreed that this area needs more work/discussion. Another issue that came up is that Sre Srinvasan no longer has the time to be the editor of this draft. ** Resolution Adam said that he had already spoken to Gonzalo Camarillo and he will be the new editor responsible for this working group submission. +++++++++++++++++++++++++++++++ Conference Control Manipulation Protocol - Mary Barnes presenting draft-barnes-xcon-ccmp-03 Mary Barnes opened the discussion by stating the document proposes WDSL and she wants to know if the group still favors this approach. There was some discussion with Adam Roach and Henning Schultzrinne at the mic about the area directors favoring this approach or not. Henning said we need the schema and WDSL only provides a boiler plate around the operations you can use. ** Resolution: Mary/Adam will talk to the area directions and forward their findings to the mailing list. For now we will go forward with WDSL. Mary then noted that the schema is lacking detail. She asked for a show of hands on who has read the document (1/3 of room) and whether she should continue along these lines (it appeared most of those people raised there hands again). Simon Romano came to the mic and expressed his frustration with the lack of effort in this area. He said that, as a general comment, the problem with the CCMP document is it disappears and reappears years later. He built a web services based implementation but he's not using it any more. They addressed roles and policies as well. When he posted it to the mailing list there was no interest. Mary responded he had a valid point. Roni Even came to the mic to discuss difference between the "change" and "add" operators. Mary said changing does include adding. Jonathan Lennox then came to the mic and brought up the issue of endpoint hosted conferences and if one endpoint was acting as the conference server behind a firewall. A discussion ensued with Henning, Jonathan, Roni, Mary, Gonzalo and Francois Audet around discovering the conference server address. Henning said that XCON has not spent cycles on endpoint hosted conferencing and has kept the scope to centralized conferencing.. But Jonathan and Gonzalo came to agreement that we should do what BFCP does, to use mline. ** Resolution: Gonzalo Camarillo will volunteer to write the text The last issue revolved around the use of error codes. Henning stated that we have to make sure the extension model is correct for error codes. Mary suggested that we take it up in the working group item when new additions to error codes are made, the chairs nodded in agreement. ++++++++++++++++++++++++++++++ Text Chat with Centralized Conferencing - Mary Barnes presenting draft-boulton-xcon-session-chat-00 Mary started by clarifying that this document replaces the MSRP thread of documents. She asked how many people (show of hands) had read Adam's chat summary and this document (not many). There was some discussion at the mic among Dan York, Brian Rosen, Roni Even, and Adam Roach around what the scope of this document was. Dan asked if we were trying to determine how other systems will interact with centralized conferencing and asked whether we were actually specifying an IM protocol. Brian said that we are trying to add IM to an existing multimedia conferencing. If you have MSRP chat this would give you more capabilities. Adam said this is supposed to be an implementer's guide, to achieve the same level of functionality in centralized conferencing as most IM systems have today. ** Resolution: Mary will add more clarification around scope to the document. ++++++++++++++++++++++++++++++++++++++++++++++ Wrap Up Adam Roach: Every agenda item had one or two issues that should come out on the list. We're looking forward to your participation ============================================================= VERBATIMS Common Data Model - Oscar Novo presenting Draft-ietf-xcon-common-data-model-07 Oscar Novo (ON): reviewers commented on -06, version -07 was issued and reviewers said the changes were ok ON: asked about persistent list and any alignment with SIMPLE, Mary Barnes (MB): not clear on issue ON: how do you map nickname of user in a list to user in the conference MB: need to have user names as persistent data ON: don't want user names to be temporal Adam Roach (AR): is persistent list unique per conference, is this nickname scoping? MB: are we talking about conference user id. ON: should we use conference user ID Roni Even (RE): the application server needs to capture the user ID AR: Work with Mary to figure MB: Adams chat room analysis is where this came from AR: concept in some systems that you have provide info to get into a conf. His straw man may not be implementable - add yourself to an active conference MB: (confused) ** AR: Resolution: back up and look more closely at the use case identified in the chat room doc/analysis around user registration before joining, and add that to the data model. AR: standardize language to the standard :lang specification in other RFCs This user will be communicating in this lang, vs XML lang identifier says this text is in this lang ON: if you pick up the phone and call this conf, this is the lang you will hear, that is what we are specifying Migel Garcia: what if I want to switch the languages Henning Schulzrinne (HS): three types of language, 1) what the user speaks 2) you can find the translation in this lang 3) prompts of the conf are in this lang 3) ON: we are defining the language that is going to be used in that conf AR: We have a subject tag, no enforcement, can go to other langs - only indicates to the user what they should expect HS: need to be clear on language of prompts and expected language of discourse ** ON: Will clarify the lang tag more and specify that it is to be used for informational purposes that it is the language to be used in the conference. This will be posted to the list for discussion ++++++++++++++++++++++++++++++++++++++++++++++ Roles --- ON: comments on the semantics of the roles, we removed in -07 and policy will not be defined in this model, maybe in the future when we define policy we can put in roles MB: roles are fundamental, framework has references to roles, realizes we're missing the policy part. MB had problem with the hierarchy of roles, not the existence ON: feels roles are more suited to policy, they imply policy MB: don't think we can wait to define roles AR: because we've serialized work in group, no issues leaving it as is, or a set of enumerations with nothing associated with it. Later have rules to bind roles to policies MB: want to define enumeration of roles now AR: enumeration without permissions HS: 1) roles as a string without structures, 2) multiplicity of roles, define too few you constrain problem MB: all I want is basic definition HS & MB: can we agree on pre-defined nouns MG: problem when reading this. What is an observer? Problem with no associated semantics. HS concerned, about specific extensible set. If we don't define what it does, we're not going to agree on what it does. Why can't we just define tokens AR: resolution: Mary to summarize the issue of including roles as pre-defined nouns for the list AR When above issues are resolved put it to WGLC, ask for a 4 week WGLC due to holidays ++++++++++++++++++++++++++++++++++++++++++++++ Conference event package - Roni Even presenting draft-xcon-event-package-extensions-02 RE: agreement was to take partial notification out of the data model and put in the event package RE: use data model schema to update package enabling partial notifications, that has to be done in the draft AR: Do we want to modify schema to accomplish partial notification or do we want to come up with other methods. Options include Version tags, ... RE: suggestion was to keep the same as 4575 AR: granularity in 4575 may not work well with all our additions to the data model RE: agree, 4575 drawback, as the conference size grows, there's a lot of messaging, such as new user joins meeting. Might want something with more granularity, it's a real issue AR: how do you know you have most recent view of the conference as a user. This area needs more work/discussion ** RE: Sri can't be editor, Gonzalo will be new editor responsible for WG submission ++++++++++++++++++++++++++++++++++++++++++++++ Conference Control Manipulation Protocol - Mary Barnes presenting draft-barnes-xcon-ccmp-03 MB: document proposes WSDL AR: Asking about CAMEL, some confusion MB: related to OASIS AR: are Area directors trying to get us to avoid it, no it's not proving useful HS: We need the schema, WSDL only provides boiler plate around operations you can use, seems like Paris fashions (don't be caught with last year's hot items). MB: Not getting rid of SOAP or schema AR: Will talk to Area Directors ** MB: for now we will go forward with WSDL, will forward comment to the mailing list MB: Schema is lacking detail, looking for show of hands on who has read (1/3 of room) and whether she should continue (most of those people raised there hands again) RE: discussing change vs. add operations, add and change are not the same, MB: changing does include adding Simon Romano SN: General comment, the problem with here is the document disappears and reappears years later, he had a web services based implementation but not using it any more. Roles and policies, they did that as well. Posted to the mailing list before on this topic but got no response MB: You have a valid point. Jonathan Lennox (JL): problem with SOAP, assumptions, such as endpoint hosted conferences. What if one endpoint was acting as the conference server behind a firewall HS: XCON has not spent cycles on endpoint hosted conferencing. JL: Propose what BFCP does, use mline JL: How do you discover conferencing server address ---- discussion ensued on ways to discover address of endpoint hosted conference server which included Francois Audet ---- ** Gonzalo Camarillo: will volunteer to write the text HS: Keep scope centralized conferences. RE: http option is valid, have problem traversing NAT. Mention retrieve can be accomplished with a subscribe, Mary will add comment GC: resolution does not see problem with using mline, he will provide text MB: error codes HS: have to make sure extension model is correct for error codes MB propose as a working group item when new additions are made - chairs nodded ++++++++++++++++++++++++++++++++++++++++++++++ Text chat - Mary Barnes presenting MB: this doc replaces the MSRP thread of docs MB: Who read Adam's chat summary and this doc (not many) Dan York (DY): are you trying to determine how other systems will interact with centralized conf - we're not specifying IM protocol - are we? MB: we're using centralized server to provide this functionality instead Brian Rosen: key thing is to add IM to an existing multimedia conf. If you have MSRP chat going, this would give you more capabilities. RE: agreed MB: will have more clarification around scope DY: agree a good way to provide another communication mode. What does this doc do? Are you going to describe protocol. MB: How you can use elements we built around media types and just use text as the media. Adam's doc is really a reqs doc AR: This is supposed to be an implementer's guide. To get the same functionality as most IM systems have today RE: We are addressing here the interaction between user and conferencing server. We are not trying to do the IM protocol. The decomposition is the media control ++++++++++++++++++++++++++++++++++++++++++++++ Wrap Up AR: every topic had one or two items that should come out on the list. Looking forward to your participation