2007-09-13 Arch WG Minutes

From Second Life Wiki
Jump to: navigation, search

This is a partial attempt to clean up the minutes from our 2007-09-13 meeting of the Arch WG. Feel free to perform further cleanup.


Introductions

  • [09:54:46] Architecture presentation ArchWG Mtg 1 Agenda (section III)
  • [09:59:01] Zero Linden: today's architecture: region manages agents on the region
  • [10:00:05] Zero Linden: fundamental change we're planning: having an "agent domain" separate from "region domain"
  • [10:01:50] Zero Linden: the reason: scaling. otherwise, all agent information can be on any region
  • [10:03:36] Tree Kyomoon local chat would be handled by region domain, but IM would probably interact more with the agent domain
  • [10:06:06] Simon Gibbs: is avatar location in region domain or agent domain?
  • [10:06:45] Zero Linden: both. regions need a list of agents, and agent domain needs to know where the agent is
  • [10:07:41] Zha Ewry: ....and region domain will no longer be queryable for "where is avatar X?"
  • [10:09:58] Zero Linden: looking at how objects get rez'd
  • [10:11:33] Zero Linden: agent domain manages inventory, makes decision about permissions/etc, agent domain transmits info to region domain, region domain then makes its checks
  • [10:11:47] Zha Ewry: where does a HUD live ?
  • [10:11:56] Zero Linden: right now, in region domain
  • [10:13:31] Zero Linden: that's for this group to decide. right now, there's lots of stuff we've bent over backwards to make backwards compatibility work
  • [10:15:34] Zero Linden: this group is to decide what the protocols between the three major blocks: viewer, agent domain, region domain
  • (next slide)
  • [10:15:56] AdamMeeti: HTTP/REST XML web services for communication to the agent domain
  • [10:15:57] Zha (Some comments about the red lines being the key issue, but we also need to talk about the buildable stuff)
  • [10:16:32] Zero Linden: Public HTTP/REST services, or behind capabitlities facade
  • [10:16:57] robla: agent domains: 100% stateless
  • [10:16:59] Zha: the boxes are 100% stateless
  • [10:17:01] Zha: REST (see Roy Fielding's thesis)
  • [10:18:18] Zero Linden: because the agent domain web service machines are stateless, we get to take advantage of systems to load balance REST systems
  • [10:19:03] Zha: http://www.ics.uci.edu/~fielding/pubs/dissertation/top.ht
  • [10:19:12] Dr Scofield: :-)
  • [10:21:54] Dr Scofield: is that using something like SIP or is that something roll-your-own
  • [10:21:55] Dr Scofield: ?
  • [10:23:10] Zero Linden: an example of where things are broken: every time you ask about online status of friends, it queries the region domain (expensive operation)
  • [10:23:32] Dr Scofield: couldn't you turn this around?
  • (next slide)
  • [10:23:54] Dr Scofield: have the region domain update a presence server (SIP or XMPP)
  • [10:23:57] robla: DrScofiel, can you ask questions in world?
  • [10:24:03] Dr Scofield: sure, sorry
  • [10:24:09] robla: (this channel is primarily for notetaking
  • [10:24:49] Zha: (It's slightly tricky for us, in the conf room, as we can't see the in world questions)
  • [10:25:03] Zha: (Behind Zero's slides)
  • [10:26:04] Zero Linden: if this is to be a global network, we need to make it possible for many different implementation strategies
  • [10:27:01] robla: mic: how much of this arch that you are presenting should be impact our thinking about this
  • [10:28:39] Zha: Zero; Try to get everything int a collection of web services
  • [10:28:51] Zha: Zero: Get rid of mySQL services on wire as part of the definitino of the system
  • [10:30:07] Zero Linden: the unknowns... e.g. .where's the one place you can ask about the IP addr handling a particular in world item?
  • next slide: agent login
  • [10:32:58] Zha: zero: is login part of the architcture...
  • [10:33:27] Zha: Zha: I want every service to be one of several, no singletons
  • [10:33:28] robla: step 1: viewer logs in via agent service, step 2: service contacts agent store step 3: agent service contacts agent hosts, and step 4: viewer is connect to correct agent host
  • [10:51:29] gillian: hullo
  • [10:51:36] bushing: hiya
  • [10:52:45] robla: next slide: region domains
  • [10:53:05] Tao_T: ok, I got region hosts, what are the other names? :)
  • [10:53:26] robla: first slide: region hosts (top left)
  • [10:53:45] robla: second slide: region stores
  • [10:54:07] robla: (oops....first slide was region services (top right))
  • [10:55:05] robla: next slide: login, part 2
  • [10:55:21] AdamMeeti: Looks a lot like kerberos
  • [10:55:33] robla: 1. agent host contacts region service
  • [10:55:55] robla: 2. region service queries region store
  • [10:56:10] robla: 3. region service contancts region host
  • [10:56:37] robla: 4. agent host put in direct contact with region host
  • [10:57:00] Zha: Zero: (agent domain, knows truth about the agent's desires)
  • [10:57:11] Zha: Zero: Viewer doesn't coordinate this, gets informed
  • [10:57:14] Tao_T: does 1 not need to be established by the viewer?
  • [10:57:34] robla: 5. agent host returns region host to viewer
  • [10:57:47] robla: 6. viewer now can talk to correct region host
  • [10:58:14] Tao_T: or this is the extension of part1, I guess
  • [10:58:38] robla: john: can you log into agent domain without ever logging into a region?
  • [10:58:41] Zero Linden: yes
  • [11:00:11] Zha: Zha: Hold for late: Why aren't mappings of Sim or domain server to machine, just manaeged by URLs
  • [11:00:12] robla: gigs: is the agent host a proxy for the viewer?
  • [11:00:15] Zha: *later
  • [11:01:30] Zero Linden: everything will be through capabilities (explained later). short answer: that's an implementation choice
  • [11:01:53] Zha: Zero, needs that list
  • [11:02:02] Zha: 500 or so contracts
  • [11:02:34] Zero Linden: 400 or so in message template, others are exclusively web services (not visible yet)
  • [11:03:14] jhurliman: i think the number is closer to 565 in the frozen template
  • [11:03:29] Zero Linden: at some point, we need to divvy these responsibility up between the differnent domains (where "we" is "this group")
  • [11:03:40] Zha: WORKITEM for tao:
  • [11:03:50] Zha: requiremnts for login (well, maybe)
  • [11:04:05] Tao_T: well, before you design a protocol one might need some requirements for it at least a rough list
  • [11:04:09] Zha: Zha laughts loudly
  • [11:04:13] Tao_T: and LL might know better than us right now
  • [11:04:22] robla: next slide: multiple region domains
  • [11:04:46] Zha: Zero: break apartr the conflated axes of scaling
  • [11:05:04] Zero Linden: what started as scaling exercise turns into distributed authority exercise
  • [11:05:25] Zha: <a href="http://dictionary.reference.com/browse/conflate">http://dictionary.reference.com/browse/conflate</a>
  • [11:05:48] robla: next slide: multiple agent domains
  • [11:06:12] Tao_T: with reqs I mean things like shown right now. It might be helpful to have a list of what a protocol for login needs to achieve right now
  • [11:07:03] jhurliman: how do i paint stuff on a mac. where is paint
  • [11:07:24] robla: next slide: multiple both agent domains and region domains
  • [11:07:42] Zero Linden: some companies may not want their agents leave specific region domains
  • [11:10:03] robla: john: in a world of multiple agent domains, how do region domains control who gets on the network
  • [11:10:41] robla: john: e.g. BlackHat Agent Domain logging into XYZ corp region domain
  • [11:11:18] Zero Linden: XYZ corp only allows agents from trusted agent domains
  • [11:12:22] Zero Linden: identity is *very* important
  • [11:13:33] robla: next slide "visiting other domains"
  • [11:15:27] Zero Linden: agent domains may have separate capability polices for different regions, where agent domains only allow inventory to be passed to some regions, but not others
  • [11:15:37] robla: (while allowing logins to all)
  • [11:16:01] Zha: Zha: Requirement: be able to carry a sheaf of identities, rather than a single one, eg, my identity on Public Domain, and in my corporate domian
  • [11:16:45] robla: next slide: it scales
  • [11:17:01] robla: we can throw hardware at the problem
  • [11:17:05] Zha: Zha: Question - When i rez an object, which of my possibly many set of rights does it have, by default
  • [11:17:28] Dr Scofield: dr scofield: req to be able to issue regon capabilties to agents while they are in-world (ad-hoc) to invite them
  • [11:18:11] Zero Linden: suspicion that there are more region domains than agent domains
  • [11:18:23] robla: next slide: central utilities
  • [11:18:49] robla: there is some centralization today on the Internet (e.g. DNS)
  • [11:19:00] robla: in Second Life:
  • [11:19:03] robla: 1. identity
  • [11:19:26] robla: 2. location
  • [11:22:36] robla: (pardon me....I'll be right back)
  • [11:25:13] robla: next slide: running off the net
  • [11:26:36] Zero Linden: elephant in the room: where are the assets?
  • [11:28:18] robla: answer: temptation is to abstract it away and punt
  • [11:28:22] robla: probably not practical
  • [11:28:42] robla: regions might ask agents for assets
  • [11:29:05] robla: probably doesn't scale
  • [11:29:28] robla: all domains will probably end up needing asset stores
  • [11:30:10] robla: we'll discuss this afternoon; not a solved problem
  • [11:30:28] robla: end of slides: more to discuss, though
  • [11:31:12] Zero Linden: first big arch decision: this will be REST
  • [11:31:20] Tao_T: good :)
  • [11:31:23] robla: ...not SOAP, not WS-*
  • [11:31:27] Tao_T: maybe CORBA ;-)
  • [11:31:37] Tao_T: heh :)
  • [11:32:25] Tao_T: it really needs to be something very simple and so REST is a good decision
  • [11:32:27] Zha: Representational State Transfer
  • [11:32:45] Zha: Idempotent
  • [11:33:06] Tao_T: <a href="http://en.wikipedia.org/wiki/Representational_State_Transfer">http://en.wikipedia.org/wiki/Representational_State_Transfer</a>
  • [11:33:20] robla: Adam asks: why use LLSD and not JSON?
  • [11:33:34] Tao_T: good question
  • [11:33:41] Zero Linden: historical decision. up for discussion
  • [11:34:05] Tao_T: I would go with known standards wherever possible
  • [11:34:11] Tao_T: gets more participants
  • [11:34:27] robla: Tao_T: should put commentary inworld rather than here
  • [11:34:33] Tao_T: right :)
  • [11:35:08] robla: (though you can make sure your commentary makes it in the notes by putting it here....it just won't be seen by as many people in real time)
  • [11:35:56] robla: another arch decision: certified http
  • [11:36:16] DrScofiel: zero: CHTTP
  • [11:36:20] DrScofiel: zero: POE
  • [11:36:45] DrScofiel: zero: POE seems dead or dying, so CHTTP
  • [11:37:05] robla: chttp implementation either is or will be very soon publicly available
  • [11:38:33] robla: another arch decision: capabilities
  • [11:39:09] Tao_T: what is CHTTP again, I think I missed that
  • [11:39:34] robla: chttp allows ensuring exactly one post occurs
  • [11:39:37] Tao_T: Xan Linden: LLSD libraries are there for most programming langs, php/perl(python
  • [11:39:41] Tao_T: ah, ok
  • [11:39:50] Tao_T: c stands for? cool? :)
  • [11:39:55] robla: certified
  • [11:40:04] Tao_T: thanks
  • [11:40:08] robla: on to capabilities
  • [11:40:21] Zha: Seperate out request from authorization form use
  • [11:40:43] Zha: Can pass the rights downstream to other tools
  • [11:41:22] robla: capabilities have "been the future" for many many years, unfortunately not getting wide use, EXCEPT
  • [11:41:37] robla: cookies are an example of real world capabilities
  • [11:41:50] robla: new SL architecture is another example
  • [11:43:34] Dr Scofield: comment: would be good to condition a cap on the existence of a second cap: pass on a cap to a helper app, but it has to show its cap to make my cap valid
  • [11:43:35] robla: capabilities server allows cryptographic unguessable url which allows for a specific function for a specific purpose
  • [11:44:00] robla: capabilities must be passed securely
  • [11:44:19] Tao_T: are they bound to some identity or can be?
  • [11:44:26] AdamMeeti: See capability discussion at Zero Linden's office hours (chatlog availible)
  • [11:44:30] Dr Scofield: comment: bind it to presence
  • [11:44:34] AdamMeeti: 4-5 weeks ago
  • [11:44:47] robla: domains routinely revoke capabilities
  • [11:44:54] Tao_T: I should visit Zero's office hours more often again anyway
  • [11:46:55] robla: about to break for lunch: goal of discussion for afternoon.
  • [11:47:08] robla: 1. get feedback on what we're missing
  • [11:48:46] robla: 2. figure out what *we* (everyone at this meeting) are going to work on, and how we're going to do it
  • [11:48:58] Tao_T: so do you have some list of use cases internally? IMHO it's really hard to find missing things if you cannot check against some list of what it should be able to do
  • [11:49:12] Tao_T: and I mean some more top-level use cases, not every protocol function
  • [11:49:23] robla: Tao: please ask those q in world
  • [11:49:39] robla: (though it may be best to wait until after lunch)
  • [11:49:56] Tao_T: I might be gone then as it's my gf's bday party today
  • [11:50:01] Tao_T: and I asked in-world ;-)
  • [12:09:05] jhurliman: hey christian
  • [12:09:09] CPrior: howdy
  • [12:09:14] Tao_T: Hi CPrior
  • [12:09:27] jhurliman: i mean, wello
  • [12:09:47] jhurliman: use case 1: attach an xcitecock
  • [12:09:59] jhurliman: use case 2: camp
  • [12:10:16] Tao_T: maybe we really should look at the most common use cases and throw away the rest ;-)
  • [12:10:36] CPrior: (jh: heh, nice)
  • [12:10:52] Tao_T: use case 3: grief
  • [12:11:05] jhurliman: exactly
  • [12:11:11] Tao_T: griefing domain is missing
  • [12:11:19] jhurliman: i brought it up already
  • [12:11:40] Tao_T: very good
  • [12:11:56] Tao_T: not that we lose the majority of SLers by such mistakes
  • [12:13:49] Tao_T: tell those people there they should talk in sequence so we can understand more ;-)
  • [12:14:14] jhurliman: talk in sequence? like go around the table?
  • [12:14:17] Tao_T: most of them are devs so they should know what a queue is ;-)
  • [12:14:53] jhurliman: so far only a fraction of the people have been asking questions, since we only covered the very general architecture ideas
  • [12:15:16] Tao_T: as said I wish we first would have collected use cases
  • [12:15:30] Tao_T: as it's a bit hard to understand what it needs to solve when just presented with a solution
  • [12:15:47] Tao_T: or some start of a solution but I guess they are further than the idea
  • [12:15:58] Zha: Tao, while I am inclined ot agree, I think the problem is that, to the Lindens, the use cases are sort of obvious
  • [12:16:12] Tao_T: sure
  • [12:16:27] Tao_T: and that's the problem most of the times as we need a common basis
  • [12:16:46] Tao_T: to me they are not really obvious and they also seem to have come up with additional ones than we right now implement
  • [12:16:51] Tao_T: like the company domains etc.
  • [12:17:13] gillian: Could someone bring me back up to speed -- use cases of what?
  • [12:17:20] Tao_T: now that I try to write them down the problem might be though how to write them down
  • [12:17:27] Tao_T: of how to use the system
  • [12:17:29] Zha: And, in some sense, I think the other problem is that the use cases alone for each contract is sort of not going to help
  • [12:17:40] Zha: them solve the whole problem. I think frankly, we need to sets of use cases
  • [12:17:51] Zha: Those per contract, and then the overall shape
  • [12:18:04] Tao_T: whatever, at least some sort of things the system needs to be able to fulfil
  • [12:18:12] gillian: (sorry, which system?)
  • [12:18:17] Tao_T: Second Life ;-)
  • [12:18:19] gillian: (the whole futar internets?)
  • [12:18:22] gillian: Oh
  • [12:18:45] gillian: SL has had "make user personas" on its plate for quite a while
  • [12:18:51] Tao_T: like - support disconnected service
  • [12:18:59] gillian: is that related?
  • [12:19:01] Tao_T: support home-hosted regions
  • [12:19:17] Tao_T: support different currency mechanism
  • [12:19:53] Tao_T: and so on
  • [12:19:53] gillian: *shuts up*
  • [12:20:17] Tao_T: and yes, there will be different sets or scopes
  • [12:20:22] Tao_T: like the company thing is one
  • [12:20:34] Tao_T: although I'd more think of restricted domains than of organization domains
  • [12:20:40] Tao_T: we also need to define names for everything
  • [12:24:11] Zha: The naming of names -)
  • [12:24:30] jhurliman: Tao_T, i agree! when will you have the naming done? :)
  • [12:25:55] Zha: LOL
  • [12:26:05] Zha: Well, we know that *every* thing will start LLxxx, right?
  • [12:26:14] Tao_T: we simply number it
  • [12:26:17] Tao_T: LL1, LL2
  • [12:26:28] Tao_T: or give them UUIDs
  • [12:26:36] Tao_T: LLUUIDs
  • [12:26:40] gillian: jeebus
  • [12:26:49] gillian: sounds welsh
  • [12:26:50] jhurliman: ah yes, RFC 4122
  • [12:26:55] Tao_T: long standing Linden tradition
  • [12:27:06] Tao_T: they do that with all my stuff
  • [12:27:09] Tao_T: or they name it "Object"
  • [12:28:06] Zha: Execept thet don't follow 4122
  • [12:28:11] Tao_T: so I started to use the term "restricted <agent|region> domain", does that make sense?
  • [12:28:24] Tao_T: it might be restricted in many ways though
  • [12:28:37] Tao_T: so a list of possible restrictions might be nice :)
  • [12:29:13] Zha: If they followed 4122, they'd have keep the reserved bits reserved
  • [12:30:40] jhurliman: and not use md5 as a final generation step, yeah
  • [12:31:55] Zha: Right
  • [12:32:44] Tao_T: hey otakup0pe
  • [12:32:46] otakup0pe: O_o
  • [12:32:58] AdamMeeti: zomgwtfhai
  • [12:33:01] otakup0pe: indeed.
  • [12:33:08] otakup0pe: jhurliman: did you hug a linden for me like i asked
  • [12:33:18] otakup0pe: i'm heading home but i guess i'll keep checking in here.
  • [12:33:19] AdamMeeti: Rob's logging this chat - so be warned, anything here will be posted.
  • [12:33:36] jhurliman: i'm working up the courage
  • [12:33:56] otakup0pe: AdamMeeti: esepcially since that time last november, i assume that's always the case ;)
  • [12:33:59] Zha: Chuckle. Just Zero's office hours on steroids
  • [12:34:25] Tao_T: and it's all finished in Jan 2008 ;-)
  • [12:34:35] otakup0pe: ya rly.
  • [12:45:35] Tao_T: who is a dork? :)
  • [12:48:02] Gigs: Ping
  • [12:48:08] Tao_T: yo Gigs
  • [12:48:41] Gigs: Yo got irc on my blackberry hehe
  • [12:48:50] Tao_T: cool :)
  • [12:49:06] Tao_T: actually I should have installed some IRC gateway on my cell phone so I can stay in touch later ;-)
  • [12:49:46] Tao_T: maybe somebody can call me and stream to voice to me ;-) It's only an international call for 4 hours or so ;-)
  • [12:50:05] Zha: Skype!
  • [12:50:14] Gigs: I've been taking notes on my BB here .. Hope zero doesn't think I'm sitting here Texting people hehe
  • [12:50:17] Tao_T: right, might be cheaper
  • [12:50:59] Tao_T: now the question is where to put my questions/notes
  • [12:51:12] Zha: Well, eventuallym, in the wiki
  • [12:51:43] Tao_T: is there some area for these things already?
  • [12:51:56] Tao_T: for now it might be sufficient to have it somewhere on the web probably
  • [12:52:10] Tao_T: not sure if somebody wants to use this for the following discussion though
  • [12:52:49] Tao_T: *waiting for the wiki*
  • [12:54:46] Gigs: Hi rob
  • [12:55:07] Tao_T: hope lunch was good, Gigs :)
  • [12:56:21] Gigs: Yep but you all got sushi hehe
  • [12:56:52] Tao_T: feel free to login and take some, too :)
  • [12:57:05] Zha: We have to get soeone to teleport us to super secret linden land
  • [12:57:41] Gigs: Hehe all I have is my blackberry
  • [12:57:56] Dr Scofield: back again
  • [12:58:46] Tao_T: Hi Dr
  • [12:58:47] Gigs: Wb
  • [12:58:53] Tao_T: ok, here are my very raw notes about use cases etc.
  • [12:58:53] Tao_T: <a href="http://comlounge.net/slarchscribbles.txt">http://comlounge.net/slarchscribbles.txt</a>
  • [12:59:15] Tao_T: somewhat tired already though ;-)
  • [12:59:37] Zha: Not bad
  • [12:59:39] AdamMeeti: Wow, good.
  • [13:00:09] Tao_T: somewhat chaotic I guess, needs some sorting
  • [13:00:24] Tao_T: and I guess there will be a lot more open questions :)
  • [13:01:12] DrScofiel: Tao_T: wow
  • [13:01:49] Tao_T: omnioutline makes all URLs into links :-/
  • [13:01:55] Tao_T: hard to write down URL examples then
  • [13:02:32] Dr Scofield: any of the gods^Wlindens online who could give me a "lift", err, teleport? :-)
  • [13:02:53] Tao_T: godly teleports ;-)
  • [13:03:21] Tao_T: it gets handles, Dr :)
  • [13:03:23] Tao_T: handled
  • [13:03:26] Dr Scofield: thx+
  • [13:04:02] Gigs: Hehe
  • [13:04:26] Gigs: Getting ready to get back to it
  • [13:06:16] Zha: Rob: now disucssoin 20-30 minutes on each topic
  • [13:06:32] Zha: rob: first process we're going to use, who we're going to be collabating
  • [13:07:06] Zha: rob: Major architecural form, agents, regions, asset and permissions as another block
  • [13:07:18] AdamMeeti: OK, I'm taking notes--
  • [13:07:30] Tao_T: I am still recording
  • [13:07:37] Tao_T: and hopefully it won't crash while I am away
  • [13:07:48] Zha: Us geeky?
  • [13:07:51] AdamMeeti: haha
  • [13:07:56] Dr Scofield: never!
  • [13:08:14] Zero Linden: discussion needs to include some social issues alongside technological
  • [13:08:22] Zha: Some of it has to be about where we *USE* it too
  • [13:08:50] Tree Kyomoon: look at making use case scenarios
  • [13:09:24] AdamMeeti: Rob: Discussion today is groundwork - start using sldev mailing list as the primary discussion list for this.
  • [13:09:48] AdamMeeti: Rob: Will be using Wiki and Jira as well - will setup categories for it
  • [13:10:48] Zero Linden: dont expect to solve any technological problems here today
  • [13:11:13] AdamMeeti: Ben Byer: Will be writing new code for this?
  • [13:11:26] AdamMeeti: Rob: We should look at setting up a repo for it
  • [13:12:19] Zero Linden: starting at the second meeting[?] - there will be code we arnt writing at all - interoperability is important
  • [13:12:37] AdamMeeti: Rob: that is a big difference between our open source initiative -- this is about open specifications
  • [13:12:42] Tao_T: so I can finally start to hack something in Python :-)
  • [13:12:48] AdamMeeti: Rob: interoperability
  • [13:13:07] AdamMeeti: Rob: important having multiple implementations of the same blackbox
  • [13:14:53] Zha Ewry: need to coordinate equivalent of "beta grid"
  • [13:15:21] Dr Scofield: interop test fest
  • [13:15:42] Zero Linden: get together in four months[?] and we try and get a proof of interoperation going
  • [13:16:35] liana: FYI, liana moved to the conference room.
  • [13:17:39] Tao_T: and I am leaving now.. will check back later for the recording, great meeting so far! :) cya!
  • [13:17:58] Dr Scofield: cu, Tao_T!
  • [13:18:41] AdamMeeti: Rob: having this on wiki.secondlife.com is important as we can relicense for use with the IETF/W3C/etc
  • [13:19:37] Zha: You can host, Dr.
  • [13:19:52] Tao_T: I updated my notes a bit btw :) now really off
  • [13:19:54] Dr Scofield: no problem, volunteering to host the whole group
  • [13:21:51] otakup0pe: i thought rob was transcribing ;)
  • [13:21:59] otakup0pe: Tao_T: are you taking notes in the slv wiki or somewhere else ?
  • [13:22:19] AdamMeeti: Zero/Rob: what can people do?
  • [13:22:27] AdamMeeti: Adam: I'm happy to work on implementing this in OpenSim
  • [13:22:40] AdamMeeti: Zha: ditto - putting people at IBM into building working endpoints
  • [13:23:06] otakup0pe: oh i like where this is going.
  • [13:23:08] AdamMeeti: Samsung guy (forgot your name, sorry!): We are interesting in providing feedback
  • [13:23:13] Zha: (Subject to our corporate masters)
  • [13:23:21] AdamMeeti: Gigs: Clientside implementation & help getting client side code and testing done
  • [13:23:23] Dr Scofield: DrScofield: dito, need to get clearance for libsecondlife
  • [13:23:36] robla: Simon Gibbs - Samsung
  • [13:24:11] AdamMeeti: Intel guy (ditto): not much coding - but working with prototypes
  • [13:24:30] AdamMeeti: Tree/Ryan: Ideas, Jira's, Suggestions, Use cases -- particular in interactive learning
  • [13:24:55] Zha: Intel is Mic Bowman
  • [13:25:12] AdamMeeti: Jhurliman: Coding / implementation
  • [13:25:30] AdamMeeti: Jh (cont): equal time spent on client and server (java?) proof of concept
  • [13:26:21] AdamMeeti: Wyn: testing / user point of view & brainstorming
  • [13:26:37] AdamMeeti: Strife: Code or documentation - doesnt mind. Enjoys writing standards.
  • [13:27:30] liana: Samsung is Simon Gibbs
  • [13:27:46] AdamMeeti: Bushing: reverse engineering / coding - work on implementations in other languages
  • [13:28:40] AdamMeeti: Nathan (Motorola): Advocate & use cases for the mobile space / set-top space -- look at it from another angle.
  • [13:31:01] AdamMeeti: Rob: what other people should we invite
  • [13:31:10] AdamMeeti: Adam: Sean Dague (from IBM)
  • [13:31:22] Tree Kyomoon: Someone from NASA[?]
  • [13:32:03] AdamMeeti: Jhurliman: Jonathan Freedman (from Pleiades)
  • [13:32:11] otakup0pe: <_<
  • [13:32:15] AdamMeeti: Haha
  • [13:32:17] otakup0pe: Jeff Barr ?
  • [13:32:21] otakup0pe: EC2 ftw
  • [13:32:21] AdamMeeti: I'll raise it
  • [13:32:27] jhurliman: k
  • [13:32:50] otakup0pe: re: set tops. SL + Wii
  • [13:32:54] otakup0pe: just sayin'
  • [13:33:36] otakup0pe: oh and i'm curious. what percentage of people in that room is on irc.
  • [13:33:52] AdamMeeti: Someone: invite people from other virtual worlds?
  • [13:34:05] Dr Scofield: croquet?
  • [13:34:05] AdamMeeti: Rob: push/tug?
  • [13:34:17] otakup0pe: mmmmcroquet.
  • [13:34:33] AdamMeeti: Intel guy: Leave as an open question
  • [13:34:35] jhurliman: delicious squeak
  • [13:34:36] Zero Linden: agreed
  • [13:34:51] liana: Intel guy is Mic Bowman.
  • [13:35:01] AdamMeeti: Heh, liana can be my corrections. :)
  • [13:35:22] AdamMeeti: Rob: trying to fix the second life architecture so it's more pallatable for a standards body
  • [13:35:36] AdamMeeti: Rob: that's where you start talking about interconnects with wildly different systems, we have to get it to a point where it's credible.
  • [13:35:50] AdamMeeti: Mic: Be good to have someone playing the role of the radical opposition
  • [13:36:25] AdamMeeti: Zha: ought to be able to take the head of an orc from world of warcraft and slap it on your table in your social space and dance around it
  • [13:36:34] jhurliman: simple matter of programming?
  • [13:36:50] Dr Scofield: smop, 1 smop == 3 days
  • [13:36:51] AdamMeeti: Rob: interop for the sake of keeping our architecture honest
  • [13:36:57] otakup0pe: Zha: totally
  • [13:40:10] Zero Linden: seems to be three seperate groups here -- testers/users, coders and spec writing
  • [13:40:54] AdamMeeti: John: what role is linden lab going to play in these?
  • [13:41:08] AdamMeeti: Someone: introducing bugs
  • [13:41:39] Zero Linden: Linden lab is going to be putting engineers on it, writing code, etc.
  • [13:41:52] Zero Linden: thanks to your existance, your going to nag us into creating specs.
  • [13:42:09] Zero Linden: part of the reason for this is to force us into writing them
  • [13:43:54] AdamMeeti: Adam: reference implementation / server code?
  • [13:43:59] Zero Linden: not sure
  • [13:44:02] AdamMeeti: (yet)
  • [13:44:33] AdamMeeti: Mic: like to know things like where the pain points are, traffic, etc?
  • [13:44:54] AdamMeeti: Rob: as linden lab engineers are collaborating, would be best to have them do it as publically / open as possible.
  • [13:45:48] AdamMeeti: Zha: if you attend Zero's office hours the pain points seem to come out regularly
  • [13:46:03] AdamMeeti: Iridium: our new director of operations is focused on that
  • [13:48:12] WORKITEM: use case librarian
  • [13:48:19] Tree Kyomoon: be good to organise tasks among this group for testing that it remains flexible
  • [13:49:00] AdamMeeti: Zha: should have use cases for how each of these 500 packet functions work
  • [13:49:04] AdamMeeti: What problems do they solve, etc?
  • [13:50:37] AdamMeeti: Adam: look at deciding what is important -- eg ground, clouds, sun etc -- can we abstract those to a higher level?
  • [13:50:57] otakup0pe: pretty things are important.
  • [13:51:09] AdamMeeti: (not a fundemental packet level though)
  • [13:52:48] robla: Architecture discussion
  • [13:52:48] robla: * Major Architectural Form
  • [13:52:48] robla: * Identity & Agents
  • [13:52:48] robla: * Regions
  • [13:52:49] robla: * Assets & Permissions
  • [13:53:14] robla: first part: major architectural form
  • [13:54:02] Zha Ewry: is there a reason to split the domains that way?
  • [13:55:26] AdamMeeti: Exactly right: LL-only
  • [13:55:36] AdamMeeti: Mostly right: reasonable number of people
  • [13:56:06] AdamMeeti: Possibly wrong: three people in the middle between mostly right / posssibly wrong
  • [13:56:13] AdamMeeti: definetely wrong: no-one
  • [13:57:26] robla: adam: possibly needs to be decomposed into more domains
  • [13:58:25] AdamMeeti: adam: not defined how these central servers hook in
  • [13:59:20] Zero Linden: are there other things that need to be in other domains
  • [13:59:40] robla: adam: might need to be an inventory domain
  • [13:59:50] robla: adam: e.g. corporate inventory story
  • [13:59:53] robla: *store
  • [14:00:24] robla: mic: further decomposition may simplify the interactions
  • [14:02:20] liana: Lots of questions about how to handle inventory. Dr brings up permissions on objects issue.
  • [14:03:55] liana: Zero makes a distinction between administrative controls of services vs. where services might "live" technically.
  • [14:04:31] robla: (thanks liana...was trying to figure out how to summarize that)
  • [14:05:22] liana: many hands...
  • [14:05:26] Zha Ewry: region domain still feels heavy (intuitively...no use case yet)
  • [14:06:50] WORKITEM: list possible additional/replacement domains
  • [14:07:26] robla: mic: explore alternative models of decomposition
  • [14:07:36] WORKITEM: explore alternative models of decomposition
  • [14:09:14] WORKITEM: use cases which expose need for alternative decomposition
  • [14:09:38] liana: mic: decomposition is going to drive complexity on the trust issue
  • [14:11:51] robla: mic: need to capture the patterns assoicated with individual operations
  • [14:12:20] Zero Linden: need to specify the 500 or so messages into 100 or so meta-patterns
  • [14:13:09] robla: gigs: resident resistance to architectural changes with privacy implications
  • [14:13:25] robla: (e.g. exposing IP addresses to content creators)
  • [14:14:02] robla: simon: what about continuous flows not well suited to HTTP
  • [14:14:05] robla: ?
  • [14:14:45] Zero Linden: rather than worrying about viewer itself, worry about using web services that provide *access* to flows
  • [14:15:27] Zero Linden: is that type of thing part of fundamental architecture, or is this something that's implemenation specific?
  • [14:15:50] WORKITEM: design for extension
  • [14:16:07] WORKITEM: design for discovering capabilities
  • [14:16:41] Tree Kyomoon use case: understanding how to stream midi to control avatars
  • [14:17:05] WORKITEM: how to provide streaming input
  • [14:18:27] Zha Ewry: there's naming and coupling in the architecture
  • [14:19:19] Zha Ewry: is domain stuff hiding stuff that could just as easilly publish as URIs
  • [14:20:02] WORKITEM: URL addressing requirements
  • [14:20:40] robla: (or, "requirements on naming services")
  • [14:21:36] robla: jhurliman: what all is necessary to document inside the box?
  • [14:22:22] Zero Linden: the domain boundries should be well-specified. the internals should be hidden so that different architectures are possible
  • [14:24:08] WORKITEM: identify functions by at least by Linden Lab's use (at least), and cluster as stateful and stateless
  • [14:25:45] Zero Linden: central utilities....we understand that we still need to distribute those
  • [14:26:32] robla: mic: central services are not necessarily architectural in nature
  • [14:27:09] Zha Ewry: some are, some aren't. identity is pretty central
  • [14:27:24] liana: Mic: central utilities ar enot necessarily an "architecture" problem.
  • [14:28:08] liana: Defining how utlities work could/should be a separate process to arch definition. Leaving them up to the market sidesteps politics and allows creative solutions.
  • [14:28:28] liana: Zero, agrees somewhat. Not entirely.
  • [14:28:42] Zero Linden: linden dollars may be valuable to have architectural
  • [14:28:48] liana: Zha says "Identity" does not fully capture the issue of trust.
  • [14:32:08] Dr Scofield: dr: identity provides one piece of the trust story
  • [14:34:03] Tao_T: back :)
  • [14:35:56] Zha: DRM is importsant, because it defines when someone has broken it
  • [14:36:15] Dr Scofield: no
  • [14:36:31] Dr Scofield: with an open source viewer DRM is null and void
  • [14:36:44] Tao_T: DRM is just psychological
  • [14:37:00] Tao_T: the problem is only that we have it right now and if some restrictions are removed people will cry
  • [14:37:08] Dr Scofield: useless
  • [14:37:11] DaleGlass: DRM is null and void open viewer or not. It's based on things that aren't possible in reality
  • [14:37:18] Tao_T: sure
  • [14:37:25] Tao_T: but people nevertheless will be upset
  • [14:37:48] Tao_T: it's a tough issue for the community team or for the tech team to come up with something which looks like working DRM ;-)
  • [14:38:26] Tao_T: sounds like break time right now
  • [14:38:33] Tao_T: how long is it?
  • [14:43:01] Tao_T: otakup0pe: my notes from back then are at <a href="http://comlounge.net/slarchscribbles.txt">http://comlounge.net/slarchscribbles.txt</a>
  • [14:43:05] otakup0pe: what
  • [14:43:07] otakup0pe: oh thanks.
  • [14:43:17] otakup0pe: poke liana if she
  • [14:43:20] otakup0pe: 's in world <_<
  • [14:43:41] Tao_T: no idea where it belongs on the wiki and it's also more meant as a list of questions for the discussion
  • [14:43:48] Tao_T: but need to read what has been actually discussed by now
  • [14:44:14] otakup0pe: which wiki - the open one or slv ? i'm assuming former ?
  • [14:44:22] Tao_T: open
  • [14:44:43] otakup0pe: hooray !
  • [14:44:46] Tao_T: no idea if I am still allowed the latter ;-)
  • [14:44:51] otakup0pe: :o
  • [14:45:27] Tao_T: so will the server code of this project be in the public, too?
  • [14:45:39] Tao_T: or what will svn.secondlife.com contain? example clients?
  • [14:45:57] Dr Scofield: i think that's not clear yet
  • [14:46:42] Tao_T: If I may make a suggestion....
  • [14:46:44] Tao_T: ;-)
  • [14:47:23] Tao_T: and what does client mean then anyway
  • [14:47:51] Tao_T: it would be helpful to have example implementations of the protocol function and e.g. login logic etc.
  • [14:48:17] otakup0pe: i think the clients are the ones you don't trust...... libsl/viewer/yet to be seen apps
  • [14:48:26] otakup0pe: well some of the ones you don't trust.
  • [14:50:50] jhurliman: otakup0pe, in this new architecture the traditional "client" can take on more roles. a client can also become an agent domain or even a region domain
  • [14:51:12] Dr Scofield: watermarking instead of DRM
  • [14:51:17] Dr Scofield: that would make sense
  • [14:51:23] otakup0pe: jhurliman: so p2p 4tw ?
  • [14:51:27] Zha: My sense of DRM.. is that the purpose is to encourage people to build content, and that end of day, you mostly end up using it so you know whensomeone has stolen from you
  • [14:51:52] jhurliman: so what happens when you write an awesome proprietary script that is hosted in XYZ Co. agent server, but someone who bought a copy wants to rez a copy of a prim containing it in a region XYZ Co. has never heard of?
  • [14:51:53] Zha: (If you want it to be real, then you need to go to watermarking, and such)
  • [14:52:00] Dr Scofield: agree...it's just that DRM is not working and we should think about watermarking right awaz
  • [14:52:31] jhurliman: and of course the purpose of that unknown region domain is to reverse engineer the script in to code files and publish it on the internet
  • [14:52:46] Zha: Chuckle, probably
  • [14:53:29] jhurliman: and that region domain happens to be running on the malicious users home pc in fact ;)
  • [14:53:41] jhurliman: new attack vectors 4tw
  • [14:53:48] otakup0pe: hooray !
  • [14:54:10] robla: discussion topic: identity
  • [14:54:11] Tao_T: the whole thing is a griefer domain ;-)
  • [14:54:33] Tao_T: I did already, otakup0pe
  • [14:54:46] Tao_T: but in fact it should be pluggable
  • [14:54:51] otakup0pe: yeah for sure.
  • [14:54:58] Zero Linden: don't want to forever make separate accounts
  • [14:55:17] jhurliman: openid would easily be pluggable in to the proposed arch
  • [14:55:31] Tao_T: I don't see the arch for identity yet
  • [14:56:29] Tao_T: otakup0pe: they say you are fine to be here! ;-)
  • [14:56:56] otakup0pe: cool.
  • [14:56:59] otakup0pe: who can i bug for a tp
  • [14:57:01] otakup0pe: i hav ea meeting now though
  • [14:57:26] Tao_T: well, openid might be crude but that might be worked out.. problem is that few people use it yet
  • [14:58:22] robla: openid provides a component, but not full solution
  • [14:58:41] Tao_T: well, I would be happy to at least be able to use my openid user/pw
  • [14:58:52] Tao_T: so I don't need to remember my pw
  • [14:59:07] Tao_T: and maybe have n agents attached to this account (== alts)
  • [14:59:10] robla: strife: how does banning work in distributed identity?
  • [14:59:46] Zero Linden: is there one central notion of identity, or do you have linked identities?
  • [15:01:20] Zha Ewry: one model: certificates
  • [15:02:34] Zero Linden: is this centralized
  • [15:02:52] robla: mic: don't think so, because different organizations are going to have different requirements
  • [15:04:22] Tao_T: not really I get all of the discussion right now
  • [15:04:28] Tao_T: not sure I really..
  • [15:05:21] lysol: what is this chicanery?!
  • [15:05:32] robla: rob: region domains have pretty heavyweight requirements for identity
  • [15:05:46] robla: for agent domains
  • [15:06:54] Zero Linden: the idea of a "customer account" is not part of the architecture
  • [15:08:26] Zha: Architecturally, an agentt ties to assets and rights, does it need anything else?
  • [15:09:16] Tao_T: region rights maybe
  • [15:09:26] robla: adam: opensim has a "foreigner system", where different domains have different access
  • [15:09:56] AdamMeeti: I'll dig up the spec on that. Give me a second.
  • [15:10:08] Dr Scofield: we should provide a backing identity for each agent
  • [15:10:59] Dr Scofield: that identity could point to an authority that's either trusted by a domain or not
  • [15:11:09] Zero Linden: when you accept a web certificate, you're trusting that verisign has validated
  • [15:12:05] AdamMeeti: <a href="http://docs.google.com/Doc?id=ddk3j5t6_0hsdhvt">http://docs.google.com/Doc?id=ddk3j5t6_0hsdhvt</a> -- really old, and has some bad ideas in it.
  • [15:12:17] robla: jhurliman: if someone starts an agent domain, how would they log into the main grid?
  • [15:12:52] Tao_T: ok, so Agent Domains are trusted by Region Domains
  • [15:12:57] Zero Linden: it depends. it may be that they are allowed, but not given no copy objects, or it could reject them outright. it's policy of the region domain
  • [15:13:05] Tao_T: if not they might have limited rights or might not be able to get on these regions, right?
  • [15:13:34] Tao_T: we should not talk about "main grid" IMHO ;-)
  • [15:13:35] Dr Scofield: Tao_T: no, the identity authority certifying a set of identity attributes needs to be trusted
  • [15:13:56] Tao_T: what is the identity authority?
  • [15:14:23] jhurliman: Tao_T, i meant the grid run by linden labs as an example of the issue. could easily talk about trust between an agent domain owned by corp A and a region domain owned by corp B instead
  • [15:14:32] WORKITEM: list the requirements that the community needs of identity to have the policies it wants
  • [15:14:39] Tao_T: ok, but grid = region domain?
  • [15:14:45] Dr Scofield: yes
  • [15:14:49] Tao_T: that's what I meant earlier with defining names ;-)
  • [15:15:04] Tao_T: ok
  • [15:15:09] robla: mic: we've been talking about having verifiable reputation, so this ties to the workitem
  • [15:15:27] robla: wyn: griefers will find a way around this
  • [15:15:45] Tao_T: griefers will get around everything ;-)
  • [15:15:48] robla: mic: let's table this
  • [15:16:28] Tao_T: I hope somebody writes down a good explanation of this identity thing ;-)
  • [15:17:15] Zha Ewry: we're going to have agent domains and region domains that have very different notions of policy (very open, very closed)
  • [15:17:32] jhurliman: it's more of a general discussion about the trust relationships between independent things like the viewer (user), agent domain(s), and region domain(s)
  • [15:17:42] jhurliman: rather than a "here is how it will work".
  • [15:17:51] Tao_T: true but I don't get everything there right now
  • [15:17:58] Tao_T: due to noise and audio and late night ;-)
  • [15:18:14] Tao_T: so if it's a list of ideas/issues it might be also good to have that written down
  • [15:18:31] Zero Linden: one thing about Second LIfe architecture is the balance built into the architecture (which predates my involvement)
  • [15:19:59] Zero Linden: some decisions will have big range of possibilities, but will be influenced by community cohesiveness
  • [15:20:43] Zha Ewry: want to make it flexible as possible
  • [15:21:28] robla: mic: flexibiltiy in the architecture
  • [15:22:08] Dr Scofield: family friendly env: higher reqs on identity provider
  • [15:22:24] Tao_T: you mean Teen Grid ;-)
  • [15:22:34] Dr Scofield: not, only
  • [15:22:44] Tao_T: I know
  • [15:23:07] robla: robla: sometimes making a decision simplifies the architecture, an dmake it not flexible
  • [15:23:15] Tao_T: so what does that mean? the agent provider needs to prove it's trustedness
  • [15:24:11] Dr Scofield: yes
  • [15:24:23] Zero Linden: just to be clear, for example, we're not going to design gambling out of the architecture, but we also not going to boil the ocean
  • [15:26:08] Zha: Zha, putting on her private hat, points out that people will build sims which do things which may be anti social, and since this is going to be open
  • [15:27:06] WORKITEM: write the architectural design principles
  • [15:27:37] WORKITEM: we need to list a set of agent rights (e.g. being able to leave a region)
  • [15:28:52] Zha Ewry: given an open codebase, there are any number of things that viewers will be able to do regardless of set policies
  • [15:29:32] robla: tree: this will require a committee to decide, and broader than this group
  • [15:30:22] robla: adam: all requirements need to have fundamental technical underpinning
  • [15:31:02] Zha Ewry: regions can have rules, and we need to mark that in the protocol
  • [15:31:37] robla: tree: some of these things may not be so cut and dried technical decisions
  • [15:32:31] Zero Linden: it may be decided that it's the fundamental right of an end user to decide what their avatar looks like
  • [15:33:04] robla: adam: may need to consider client to client protocols as well
  • [15:34:40] Tao_T: 1-1 chat might in fact go between clients but might have FW issues
  • [15:35:01] robla: jhurliman: don't see where there would ever be a scenario where people would always have the right to choose appearance rather than just opt out of the region
  • [15:35:12] Dr Scofield: 1-1 chat should go between clients! solve the confidentiality issue with one stroke
  • [15:35:38] Tao_T: actually chat should be a separate system anyway
  • [15:35:39] AdamMeeti: How do you determine your actgually speaking to the client?
  • [15:35:44] DaleGlass: confidentially is easy: integrate PGP into it
  • [15:35:49] Zero Linden: this gets to the heart of the dicussion. we'll need to come up with architectural principles to refer back to
  • [15:35:50] DaleGlass: in fact I've got plans for that
  • [15:35:56] Tao_T: and a pluggable one
  • [15:36:18] Zero Linden: one principle I'll throw out now.....we can never forget we're talking about people (bring out dogs and tears and violins)
  • [15:37:24] Tao_T: right, it should be very basic but in the end there need to be all those services we need right now
  • [15:37:33] robla: wyn: if you open source the server, people are going to do what they want regardless of what we decide
  • [15:37:34] Tao_T: but it should all be components which are easy to exchange
  • [15:37:38] Dr Scofield: if regions require certain outfit/behaviour they can do so but need to declare it upfront (before teleporting concludes or so), agent doesn't have to go there, does she?
  • [15:37:57] Tao_T: nobody should be forced to go to regions ;_)
  • [15:38:25] Dr Scofield: so we should have a way of letting regions specify policies?
  • [15:38:47] Tao_T: should be in the region services
  • [15:39:01] DaleGlass: well, IMO people are going to do whatever they want in any case :-)
  • [15:39:10] Tao_T: of course ;-)
  • [15:40:23] robla: next topic: region architecture
  • [15:40:46] WORKITEM: what are the feature sets that regions may or may not implement
  • [15:41:33] Tao_T: topology might get very interesting with different sized regions
  • [15:41:41] WORKITEM: a list of possible parameters to consider (e.g. region size: should they be somethign other than 256x256m)
  • [15:42:07] Dr Scofield: also, basic avatar support: move, communicate, teleport away
  • [15:42:14] Tao_T: make them also stackable! :)
  • [15:42:57] Dr Scofield: (and SL<->IRC bridge would be nice, find myself saying this stuff in-world as well as in IRC)
  • [15:43:06] WORKITEM: figure out vhosting possibilities for regions
  • [15:43:17] Tao_T: they don't read it in-world usually anyway ;-)
  • [15:43:24] Tao_T: but anyway, where is that libsl-bot? :)
  • [15:43:47] DaleGlass: which bot?
  • [15:44:26] robla: ping AdamMeeti when you want to raise an item
  • [15:44:26] AdamMeeti: Ping me
  • [15:44:37] Tao_T: k
  • [15:44:45] jhurliman: DrScofiel, other then teleporting away, there's no real reason those should be fundamental rights everywhere. first, a sim could simply break the protocol here and "forget" to hear any chat or movement packets. second, there are valid use cases where a sim would want to mute avatars or freeze their positions, and both of those are built in to the current protocol already
  • [15:45:07] Dr Scofield: true
  • [15:45:11] Dr Scofield: got a point
  • [15:45:40] jhurliman: most of the last 15 minutes of discussion here has been over that topic exactly. what fundamental rights should avatars have?
  • [15:45:50] otakup0pe: DrScofiel: fine i'll stay up all night and fix my sl-irc bridge. gosh.
  • [15:46:00] Dr Scofield: lol
  • [15:46:01] jhurliman: i'm in favor of teleporting away as the only one, everything else can be up to the regions
  • [15:46:26] Dr Scofield: agreed, if the region has some way of signalling that to the agent
  • [15:47:53] Tao_T: question is if there needs to be a list of possible avatar actions
  • [15:48:00] Tao_T: but probably this also should be open
  • [15:48:01] WORKITEM: if there are good use cases for topology changes to the architecture, capture
  • [15:48:03] Tao_T: and extensible
  • [15:48:34] Tao_T: might a 3D grid make sense?
  • [15:48:47] Zero Linden: we don't want to architecture where we bust up the world into tiny individual worlds
  • [15:49:00] otakup0pe: Tao_T: who cares it woudl be awesome :D <_<
  • [15:49:07] otakup0pe: can i haz tp ?
  • [15:49:23] Dr Scofield: virutal galaxies!
  • [15:49:32] Dr Scofield: virtual galaxies evne
  • [15:49:34] Dr Scofield: even
  • [15:49:39] Dr Scofield: (getting late)
  • [15:49:39] Tao_T: Liana seems not to be here
  • [15:49:56] liana: I"m in the conference room in SF.
  • [15:50:10] Zero Linden: the best example is the evolution of the space archipelego
  • [15:50:39] Tao_T: otakup0pe: well, it would be nice to have somebody build "space" ;-)
  • [15:50:58] otakup0pe: you mean stacking them vertically ?
  • [15:51:00] Tao_T: or have something up in the air like skyboxes but more sky cities
  • [15:51:03] Tao_T: yep
  • [15:51:05] otakup0pe: yeah.
  • [15:51:53] Tao_T: more so to have more prims available right now would be cool ;-)
  • [15:52:01] Tao_T: I'd like to build another city on top of my existing one
  • [15:52:06] otakup0pe: that shoudl go without saying :o
  • [15:52:34] otakup0pe: it just depends on what you are using the platform for.
  • [15:52:52] robla: break until 4pm
  • [15:52:59] Tao_T: for some region domains that might be nice to have
  • [15:53:07] otakup0pe: time zones are fun.
  • [15:53:09] Tao_T: like in skyscraper region domains
  • [15:53:15] Tao_T: timezones are awful ;-)
  • [15:53:18] Tao_T: 1am here
  • [15:53:27] Dr Scofield: here too :-)
  • [15:53:37] otakup0pe: oh you aren't in sf ?
  • [15:53:45] Dr Scofield: me? nah
  • [15:53:48] Dr Scofield: zurich
  • [15:53:58] Tao_T: also enough travelling for me these days
  • [15:54:00] Dr Scofield: well, suburb of zurich
  • [15:54:05] otakup0pe: i'm GMT-4
  • [15:54:15] Tao_T: and Naples in October again
  • [15:54:18] Dr Scofield: GMT+2 currently
  • [15:54:35] otakup0pe: i'm usually -4 or -5. <3 montreal
  • [15:54:39] Tao_T: but I hope to have some toy servers up soon so I can play with some python code :)
  • [15:55:08] otakup0pe: wait what.
  • [15:55:11] Tao_T: is there actually code by LL already?
  • [15:55:19] otakup0pe: eventlib/mulib
  • [15:55:27] Tao_T: I mean for this project
  • [15:55:28] Tao_T: internally
  • [15:55:37] otakup0pe: i missed that text.
  • [15:55:37] Tao_T: eventlet
  • [15:55:55] Tao_T: wonder if they packaged it yet as an egg
  • [15:56:41] Tao_T: and I wish I once did my web based use case editor, it would be very handy :)
  • [15:56:53] otakup0pe: wait so what toy servers do you mean
  • [15:58:05] Tao_T: agent domain, region domain
  • [15:58:18] otakup0pe: O_o
  • [15:58:38] Tao_T: if they say we might meet again in 4 months with laptops and test our code they'd better have some code already ;-)
  • [15:59:58] robla: about to start again
  • [16:00:09] liana: we're back
  • [16:00:53] robla: last arch topic: Assets & Permissions
  • [16:01:49] Zero Linden: Second Life has permission system, with plus and minus
  • [16:01:58] Zero Linden: there's no DRM in our system
  • [16:02:19] Zero Linden: Cory has all sorts of quotes you can search for
  • [16:02:29] Zero Linden: we don't want to bake DRM into the system
  • [16:02:55] Zero Linden: however, there are needs for transfer permissions
  • [16:03:12] Zero Linden: we need to focus not on the final design, but work items
  • [16:06:33] Zha: (zero) Would a work item be to a list of uses of content
  • [16:08:31] WORKITEM: classify modes of content use (not exhaustive list of every possible use)
  • [16:09:34] robla: (battle nearly erupts about whether certain restrictions are possible with open source)
  • [16:09:43] AdamMeeti: hahah
  • [16:10:00] Zero Linden: we're not going to force DRM on anyone
  • [16:10:25] Zha Ewry: may need watermarking
  • [16:10:29] Tao_T: sounds as if you can opt out DRM ;-)
  • [16:11:01] robla: adam: even watermarking isn't possible
  • [16:13:19] robla: DrScofiel: DRM probably won't work. watermarking is the only way it might work
  • [16:14:24] robla: Tao: if you can copy objects from agent to region domain, where are the textures?
  • [16:14:38] robla: might alsways be needed to be copied as well
  • [16:15:09] Dr Scofield: DrScofield: DRM is NOT going to work, watermarking might be the only way to tackle this problem
  • [16:15:12] Dr Scofield: :-)
  • [16:15:16] Zero Linden: we're going to need to make it so that regions and agents can share asset stores, but that it's possible for them to not
  • [16:17:35] WORKITEM: we need a clear definition of what parts of an object are included vs which parts are referential
  • [16:18:33] AdamMeeti: Question for the above work item: the object itself as referential?
  • [16:21:11] robla: mic: trust relationships between region domains and agent domains will be important, both direct and transitiive
  • [16:21:20] Tao_T: question might be where to where it's referenced
  • [16:21:49] robla: tree: how does this impact scripting and how object is stored
  • [16:22:49] robla: jhurliman: trust relationships between domains, so can I take my assets over region domain boundaries
  • [16:24:50] robla: wyn: inventories need to be simpler, not more complicated?
  • [16:25:08] robla: strife: local inventories?
  • [16:25:31] robla: adam: once you talk multiple sims, all bets are off
  • [16:26:05] robla: simon: (personal opinion) drm in general is quagmire....leave it to domain
  • [16:26:52] robla: gigs: interregion trust has been big topic. what about end user trust? how will end user acceptance work?
  • [16:27:29] robla: gigs: I'm on the side of not caring about IP addresses, but a lot of people push back on IP address privacy
  • [16:27:41] otakup0pe: gigs that's a big point
  • [16:27:43] AdamMeeti: tor / proxies?
  • [16:28:20] otakup0pe: you ever over tor ?
  • [16:28:38] Zero Linden: for some people: when you visit Second Life, you are traveling with your identity, so it's the combination of IP and identity
  • [16:29:19] WORKITEM: need to define avatar rights on privacy
  • [16:30:07] robla: strife: so can we all agree not to put DRM in the architecture?
  • [16:30:20] robla: (general consensus not)
  • [16:30:37] jhurliman: side note: if my understanding of the voice protocol is correct, i think i should be able to release a proof of concept soon that exposes IP addresses of SL users
  • [16:30:45] otakup0pe: jhurliman: <3
  • [16:30:49] otakup0pe: you griefer.
  • [16:31:01] jhurliman: ruining people's days, one line of code at a time?
  • [16:31:31] robla: next section of agenda: working items/closing
  • [16:32:13] Zero Linden: we could walk away from this meeting happy that we've had this meeting
  • [16:32:24] otakup0pe: i award you this award.
  • [16:33:15] Zero Linden: but, for this to continue, we need to work
  • [16:33:15] jhurliman: today is the 256th day of the year, aka programmer's day
  • [16:33:37] jhurliman: LOOK IT UP SOMETIME
  • [16:33:39] Zero Linden: we had a sense we needed to keep this small in order to start
  • [16:36:45] bushing: did we say something about using kerberos? or not using it? or something
  • [16:36:52] jhurliman: yeah
  • [16:37:09] AdamMeeti: I said a few things about using the same principles for login / authentication
  • [16:39:04] AdamMeeti: note taking again-
  • [16:39:22] AdamMeeti: zero: Work items listed should have contributions made, but no-one owns them
  • [16:39:37] robla: (yet)
  • [16:39:45] bushing: axial is a kerberos expert.
  • [16:41:26] AdamMeeti: zero: Want as many people contributing to these items - and also for the community to be aware of what's happening
  • [16:42:25] bushing: otakup0pe, no
  • [16:43:01] otakup0pe: bushing: i'm assuming you are answering my sl question ok. mixed mediums are fun.
  • [16:43:06] liana: Action Item: Rob, Zero, Liana will get chat and irc logs posted to sldev; blog post; slides up on wiki... This will take a few days to complete.
  • [16:43:08] bushing: yup
  • [16:43:26] Tao_T: also the in-world chatlog?
  • [16:43:30] liana: yes
  • [16:43:33] Tao_T: k
  • [16:43:33] AdamMeeti: zero/rob: timeline --- a week of 'regrouping' ?
  • [16:43:37] Tao_T: what about my audio?
  • [16:43:38] Dr Scofield: good!
  • [16:43:38] WORKITEM: Rob, Zero, Liana will get chat and irc logs posted to sldev; blog post; slides up on wiki... This will take a few days to complete
  • [16:43:42] AdamMeeti: zero: meetings vs correspondance?
  • [16:44:24] liana: Tao, let's follow up off line once I hear how well Terra Tester did. Ok?
  • [16:44:41] Tao_T: ok
  • [16:44:47] AdamMeeti: Zero/consensus: meet within a month inworld?
  • [16:44:52] AdamMeeti: Zero's office hours as well
  • [16:44:54] Tao_T: but tomorrow then as I need to go to bed 2 hours ago ;_)
  • [16:45:03] Tao_T: and I am still in the office
  • [16:45:59] liana: k
  • [16:46:53] Tao_T: but you can send me an email, I will read it before you are in the office I guess ;-)
  • [16:48:35] WORKITEM: come back and two weeks with everyone deciding what they can commit to
  • [16:49:20] otakup0pe: i can have that within two weeks for sure.
  • [16:50:33] WORKITEM: Mark: post list of messages in the wiki
  • [16:50:51] otakup0pe: protocol-level messages ?
  • [16:50:57] AdamMeeti: Yes
  • [16:54:12] AdamMeeti: Meeting adjourned
  • [16:54:40] otakup0pe: safe travels everyone.
  • [16:54:56] Dr Scofield: good night everyone
  • [16:55:37] Tao_T: good night, off home and to bed now..