[9:43] Dale Innis: ( Ah people just rezzing now :) )
[9:43] Tess Linden: we're thinking about getting appearance and avatar data nailed down for login and teleport as one of the next design steps (which includes trust since content is involved) but not sure if there are other avenues that'll be more useful
[9:43] Dale Innis: looks to see if he's sitting on anyone.
[9:46] Saijanai Kuhn: Tess, I've been wondering if getting a group IM model might be an easier and less controversial next step at least for public testing
[9:47] Tess Linden: JayR: sure, I think we should have meetings everywhere :)
[9:47] Tess Linden: we often attend the Opensim office hours on Tuesdays at Wright plaza
[9:47] JayR Cela: and they have quite a few people on their in-world staff
[9:47] Dale Innis: Yeah we can definitely have other meetings elsewhere. :)
[9:48] Saijanai Kuhn: if the AD can act as a central dispatcher for each virtual world youve visited (recently?) that would allow some sense of community between worlds even before inventory is available
[9:48] Tess Linden: would be good to see how well the Sun protocols match up w/ OGP
[9:48] Dale Innis: Sai/Tess: IM as a next POC is a good idea. I do think we should be thinking about the asset issues at the same time, though, and doing some design.
[9:48] Saijanai Kuhn: dispatcher for the group IM, that is
[9:48] Dale Innis: And we should bug the RealXtend folks to use the AD model!
[9:54] Teravus Ousley: I've seen some issues occasionally with teleporting from and to regions with the old patch. Usually they're regionhandle issues.
[9:54] Dale Innis: (So far it doesn't actually look like the trust part of the asset part of the OGP will have to be very complicated; most of the hard stuff is domain-internal. I think!)
[10:01] Teravus Ousley: isn't sure how the Linden sims are configured.. however.. OpenSimulator has always been configured where a gridspace is 256m2. That means configuration for a sim in opensimulator is a gridspace x, y meters / 256
[10:02] Dale Innis: Has anyone signed up to start a page / discussion on IM interop and design and requirements? I'd be glad to start the page, but I don't know enough to get actual content right.
[10:02] Dale Innis: Those "256"s shouldn't be baked into the protocol, though, presumably?
[10:03] Tess Linden: in the OGP draft 3, there's a 'region/info' resource that will give you back information on the region, of which there are x and y fields, and second Life & opensim should both support the same format
[10:03] Teravus Ousley: notes that often times it isn't something people are familiar with.
[10:05] Teravus Ousley: So bonghit: if you teleported from a region running the older patch and the region X and Y were not multiples of 256, then you will teleport to a void.
[10:05] Tess Linden: not released yet, coming soon! probably this week
[10:09] Teravus Ousley: ah, well, currently, "I think" the client uses it to place objects and avatar in the visual space.. as well as cache objects. It gets sent out via the UDP protocol in many places.. and if the client doesn't understand the regionhandle that the simulator sends, you get message like 'object update from unknown region'.
[10:10] Dale Innis: ( The AWG Groupies page still has a pointer to OGP draft 1... fixing... )
[10:10] Teravus Ousley: in opensimulator we made the choice to do that so that configuring for the end user is easier.
[10:11] Teravus Ousley: (though, I want to add an option to specify units per grid space)
[10:12] Dale Innis: would love to see WIki pages on region handles, and on requirements / design for IM interop.
[10:13] Mahakala Omegamu: Does anybody know if there will be any restrictions on the number of opensim grids LL will provide access to?
[10:14] Dale Innis: The number? I doubt that decision's anywhere having been made yet.
[10:14] Dale Innis: Probably one to start with. :)
[10:14] Tess Linden: right now you just have to sign up for the beta and connect to aditi
[10:14] Teravus Ousley: it seems to be the other way. You type in an address to a region domain, and LL will send you there. If the region domain trusts LL, then it sends you off... once you're a gridnaut.
[10:14] Dale Innis: Oh right sorry, yeah for the beta anyone signed up.
[10:15] Saijanai Kuhn: I don't see what there would be alimit. Its a clientside issue to request a connection to a specific grid. The AD just takes the client request and forwards it to the grid. The AD doesn't have to store every grid in the universe
[10:15] Dale Innis: Well, once there are things with more value than AV names involved, it will get a little more complex than that. :)
[10:17] Mahakala Omegamu: I was talking more about real world deployment of OGP by LL. I'm curious about business opinions LL has on how big this might get for my clients.
[10:17] Dale Innis: Yeah we're not the right group to ask about Official Opinionsn on that. :) I personally would expect it to be huge.
[10:17] Saijanai Kuhn: For pure Ruth Goodness (TM), there's no practical limit. You could TP everywhere. For assets, and the like, there would be limits bound by trust
[10:18] Mahakala Omegamu: So, if LL no longer trusts you, your out?
[10:18] Dale Innis: No one's going to want to TP anywhere nude and inventoriless. :) So it's the latter that matters.
[10:18] Saijanai Kuhn: its a matter of who trusts LL at least as much
[10:18] Dale Innis: No, more that if LL doesn't trust some other grid, LL won't release any of the assets in its care to that grid.
[10:18] LaPiscean Liberty: that also means agreeing to SL TOS if Truswted
[10:19] Dale Innis: Not necessarily exactly the SL TOS.
[10:19] Teravus Ousley: Right. The region domain must trust LL's agent domain... since it doesn't provide any authentication information besides the fact that it says the person logged into there.
[10:19] Dale Innis: That's for the lawyers to figure out. :)
[10:19] Saijanai Kuhn: or creating a category of inventory permissions where "Wild Wild West Grid" is OK
[10:22] LaPiscean Liberty: Like i see if your a grid useing SL money ot econimy then you may be trusted and if your na grid with your own economy then you may not be
[10:22] Saijanai Kuhn: well it becomes an issue of trusting the AD to forward the data to the right person
[10:23] Dale Innis: LaP: it's not really about which economy is being used, I don't think.
[10:23] Saijanai Kuhn: though, I'm wondering if the caps model Graph is familiar with could allow for a second level of securty for messages...
[10:23] Dale Innis: Sai: and to not save all IM passing through and post it to the web... :)
[10:23] Teravus Ousley: at least for OpenSimulator... it doesn't really care about if an IM is from a valid user currently.. it uses XMLRPC between the regions independant of the grid (besides using the grid to find out which region the destination user is in) to send the messages.
[10:23] LaPiscean Liberty: im saying its a matter of signing into the tos of sl and agreeing before anything of this nature can exist
[10:24] Saijanai Kuhn: or the equivalent for the open grid
[10:24] Dale Innis: Money interop hasn't been worked on much yet. That'll be another very interesting question!
[10:24] Tess Linden: I don't think everyone's read all the wiki's yet, its a lot to gulp down
[10:24] Saijanai Kuhn: could have more than one possible trust agreement and more than one trust agreement between the same two parties for different things
[10:24] Teravus Ousley: Warthog: talk to Which linden for more information on that. He/She's been working on some great ideas in that space.
[10:24] Graph Weymann: it seems to me that if IM requires a "trust model" you're doing it wrong
[10:25] Teravus Ousley: Warthog: Which has office hours.
[10:25] Dale Innis: Graph, how would you do IM with no trust required?
[10:25] Graph Weymann: all you need is a strong identifier for (the proxy for, i.e. the agent domain(?) of) the target user
[10:25] JayR Cela: Linden $ are worthless in other domains
[10:25] Saijanai Kuhn: Graph how would you set up group IM between grids, given the AD is doing the authentication otherwise?
[10:25] Dale Innis: Right, but you have to trust the AD of the target user, yes?
[10:25] Bartholomew Kleiber: I see an OGP Stock Exchange coming for currencies ...
[10:25] Graph Weymann: no, yuou don't think of it that way :)
[10:25] Mahakala Omegamu: I think looking to jabber for a cross server IM architecture would be beneficial
[10:27] Graph Weymann: SL already has calling card objects, that is the basis of such a system
[10:27] Saijanai Kuhn: the way I'v been thinking of it is that a user IM's a group or user via his own AD and include the other user's AD as part of hte target. That other AD forwards the message
[10:27] Dale Innis: That's fine, Graph; it's still a trust model. :) You believe that AD x is really the AD for a given agent. Maybe that's all we need.
[10:28] Graph Weymann: the calling card can strongly identify the target AD
[10:28] Mahakala Omegamu: And jabber can form connections among disperate IM protocols already , "bridges" I think they are called
[10:28] Dale Innis: JayR: yeah, Jabber might be a very good base. Someone should compare what it can do with the OGP IM requirements. Make a Wiki page! :)
[10:28] Saijanai Kuhn: JayR maybe. I kow that Zha and Zero have discussed this quite a bit. No existing protocol does quite what they need, quote both of them
[10:29] Mahakala Omegamu: XMPP is the generic messeging architecture used by jabber and cloud computing
[10:30] Teravus Ousley: doesn't have an issue with XMPP except that it relies on appropriating a central XMPP server to a domain. If you want to have a noncentral system, you need several domains.
[10:30] Dale Innis: ( Maha: ty. I know Jabber uses XMPP; I'm not aware of which actual cloud-computing installations currently active use it. Do you have an example? THis is a tangent :) )
[10:31] Mahakala Omegamu: Yah, one second, theres a distributed rendering project that uses it, let me dig it up
[10:32] Teravus Ousley: notes that additionally XMPP was designed with the idea that the client would manage some options with regards to the list of contacts/friends.
[10:32] Tess Linden: Sai: were there any docs written up on the this stuff that Zha & Zero talked about?
[10:32] Teravus Ousley: notes that the simulator currently does a lot of that in the 'Linden' scenario
[10:33] Saijanai Kuhn: not that I know of. Zha just commented to me one day that she and Zero had talked about it, and she'd checked what zero was saying and he's right: nothing that exists does quite what they want
[10:33] Dale Innis: Maybe we can ask Zero today at his Office Hours
[10:33] Teravus Ousley: Barthomew, potentially. I was looking at using XMPP over BOSH for it.
[10:34] Bartholomew Kleiber: Sai: I didnt expect to find one product fitting it all. But if it's 80% and OpenSource it's still better than writing from scratch.
[10:34] Teravus Ousley: The current implementation came down to time for me. My time is limited and the community needed working instant messages, so I did a quick implementation in XMLRPC
[10:35] Dale Innis: XMPP isn't a product, it's a protocol.... :)
[10:36] Dale Innis: So if we want to try IM next, we should start writing down ideas about how.
[10:36] Saijanai Kuhn: so, I was thinking, as a near term thing, we could implement an IM forwarding mechanims in the AD just to give a feel for cross-grid community, and start workign out the details from what pops up by trying to implement the CURRENT design using the AD
[10:37] Dale Innis: How many ADs have to know about group membership?
[10:38] Saijanai Kuhn: Zha and I talked about it as sorta the live equivalent of email. You rout your email and eventually it ends up with your ISP and then to you when you get online
[10:38] Tess Linden: groups are tricky... can you be in the same group but in different agent domains?
[10:38] Saijanai Kuhn: that's an interesting question... can you?
[10:38] Dale Innis: But with email every message contains the entire To: and CC: lists. We probably don't want to do that with IMs?
[10:40] Saijanai Kuhn: well, for group IM to behave like SL group IM, you would either need a central server to track every person's online status, or just forward it to their AD which would already know