AW Groupies/Chat Logs/2007-11-06

From Second Life Wiki
Jump to: navigation, search
[9:35] Zha Ewry: okies
[9:35] Zha Ewry: As usual. Nobody added to the agenda, or at least hasn't last night.
[9:36] Goldie Katsu: Seemed like a good one to me.
[9:36] Zha Ewry: So.. I have three things, I think.. and.. then.. We can kick people.
[9:36] Code Tracer: well i got notice 1 hour ago
[9:36] Zha Ewry: Yeah, I sent one two days ago, but it appears SL ate it for 90% of people
[9:36] Zha Ewry: So....
[9:36] Tillie Ariantho: darn ,)
[9:36] Code Tracer: do we have google cal?
[9:36] Tillie Ariantho: Oop, we have an agenda now? .D
[9:37] Zha Ewry: Presence.../Group IM, which, tihs past week has proven to be inmpressive.. so we should try to understand from Linden how they borke dit, and talk about how that really works in a het grid, ebcause it's not obvious
[9:37] Goldie Katsu: It's a hidden agenda for you ;)
[9:37] Zha Ewry: Yeah, we do, and I need to remember to update it. (sigh) too many tools
[9:37] Zha Ewry: Scope, and where we want to push back a bit
[9:37] Code Tracer: IM me
[9:37] Zha Ewry: and.. then.. Making some progress.. since, we are *so* not, doing much of that
[9:38] Code Tracer: i update it
[9:38] Wright Juran: how does someone get added to AWGroupies group? Adam Zaius wanted to come here but he can't get access as hes not a memember
[9:38] Zha Ewry: Im me
[9:38] Code Tracer: NO! access list here
[9:38] Code Tracer: upps sorry for gesture
[9:39] Zha Ewry: And .he's invited to the group now...
[9:39] Wright Juran: ahh opps , he wants Adam Xinpeng (his alt ) to be added
[9:40] Zha Ewry: OK
[9:41] Zha Ewry: Welcome all.. including new folks
[9:41] Zha Ewry: In case you don't know (ie have been living on airplanes all month) Group IM is borked about 90% of the time, at the moment
[9:41] Adam Xinpeng is guilty of declaring a airplane his prinicple place of residence this year.
[9:42] Zha Ewry: Which. actually raises the questoin, of why it's so hard to do, and how we do presence, and group IMs, and such in a widely distribued grid
[9:42] Zha Ewry: We also, have some questions, about how to think about Linden's scoping.. efforts
[9:42] Code Tracer: neas buy it
[9:43] Zha Ewry: I personally think, they are about, 25% to restrictive
[9:43] Zha Ewry: *too
[9:43] Zha Ewry: But. it might be 23, or 27.,
[9:43] Code Tracer grins..
[9:43] Code Tracer: more than that
[9:43] Zha Ewry: Seriously, tho.
[9:43] Code Tracer: say 50
[9:43] Zha Ewry: We are going to need to keep in mind that, as a community, we probably want to do more thn Linjden does.
[9:43] Goldie Katsu: I agree on that.
[9:44] Zha Ewry: Which.. They sort of get.. But.. can't keep in thier heads, as they try and fix their world
[9:44] Goldie Katsu: Well we can just run our track of what our scope is and when we talk with lindens we focus on the piece they want and then gradually work to expand it.
[9:44] Zha Ewry: My take, is that, on the whole, we just keep gently reminding them, that, OpenSim people, and other people looking at the space will have
[9:44] Burhop Piccard: I'd be ok with the current level of restriction if they had a better way to organize what is outside of it.
[9:44] Zha Ewry: stuff they care about that Linden doesn't.
[9:44] Goldie Katsu: They may be worried that the bigger scope means that their concerns won't be addressed.
[9:45] Multi Gadget v1.52.0 by Timeless Prototype
[9:45] Zha Ewry: Right. and.. I get that,m especially, since. frankly, on the Wiki, there is a lot of ocean boiling, and random nonesense
[9:45] Goldie Katsu: If they see action in the area that they need they will be more open to the idea of broader scope.
[9:46] Zha Ewry: So.. for now, wehn we get pushback, I'd like to keep it in the SL wiki, in the AWGroupies area. If.. Linden pushes back hard enough. We can as a group, move the broader scope stuff to another wiki, and figure out how to keep things coordinated
[9:46] Zha Ewry: That's my thought, but I'm
[9:46] Burhop Piccard: I've personally not had much luck getting anyone with a last name of "linden" to discuss things.
[9:46] Zha Ewry: eager to hear other thoughts
[9:47] Adam Xinpeng: I think to a degree LL has been scared off a bit.
[9:47] Zha Ewry: When.. you can get Zero's attention.. he's pretty coherent
[9:47] otakup0pe Neumann: I notice there is nobody from LL here now.
[9:47] Zha Ewry: But.. it's REALLY hard to get his attention
[9:47] Adam Xinpeng: My best guess puts LL keeping thing internal again since a lot of the public chats have turned to rubbish.
[9:47] Zha Ewry: Right.
[9:47] Goldie Katsu: We do tend to get very multithreaded at the meetings.
[9:47] Zha Ewry: I agree Asam
[9:48] Zha Ewry: And.. The wiki, frankly, has caused me personal grief, both inside IBM, and with peers in big corporationgs
[9:48] Goldie Katsu: How so?
[9:48] Zha Ewry: I can't say "Come look here" because if I do, they get "Paravirtulization" and "allusions as a progrmaming model"
[9:48] Zha Ewry: Not to mention "Every prim has an agent"
[9:48] otakup0pe Neumann: heh.
[9:48] Burhop Piccard: Zero's input would be helpful. I've got this fear a lot of blood and sweat will be put into this only for somone to cme back and say "Thats not quite what we want".
[9:48] Adam Xinpeng: I say we use the opensim mailing lists *grin* :P
[9:49] Code Tracer: no.. mailing list pleeeasse
[9:49] Zha Ewry: me shrugs
[9:49] Goldie Katsu: I think we need to define what our scope is, which obviously is a superset of what linden wants.
[9:49] Zha Ewry: One thing, which would be useful, is if I want' the only person slowly pruning out the Wiki
[9:50] Goldie Katsu: And that we need to work on focus within that scope.
[9:50] Burhop Piccard: Code, Adam - I don't know. People seem to use one or the other. I get different inputs from SLDEV and WIKI.
[9:50] Zha Ewry: I will, keep it up, but.. I hate being the only person going "This is pretty bizzare"
[9:50] Burhop Piccard: No, this IS pretty bizzare.
[9:50] Code Tracer: Burhop, I like wiki better
[9:51] Goldie Katsu: I find the current dev mailing list to be a bit of a torrent of information.
[9:51] Zha Ewry: The wiki.. has a lot of stuff, which, frankly is one person's unique view of the universe
[9:51] Burhop Piccard: Code -yes - but I was able to get much more discussion on SLDEV.
[9:52] Zha Ewry: Scope:
[9:52] Morgaine Dinova: 'Afternoon
[9:52] Tillie Ariantho: Hello :)
[9:52] Emma Nowhere: that was my concern, reading the wiki, was how much of it represented the group
[9:52] Goldie Katsu: afternoon Morgaine.
[9:52] Zha Ewry: Linden's got it about 1/2 right, maybe more. Interop, scale, and all. Except they don't mean scale it seems sometime
[9:52] Zha Ewry: and.. they don't mean interop, always
[9:52] Adam Xinpeng: Yes - I have some serious concerns about parts of the client protocol as well
[9:52] Burhop Piccard: Do others think Lindens aren't giving enough information?
[9:52] otakup0pe Neumann: well i'm sure they want it to interoperate between different RetailSim instances....
[9:52] Zha Ewry: And. they are scared of varients
[9:52] Adam Xinpeng: EventQueueGet is just asking for trouble.
[9:53] otakup0pe Neumann: heh.
[9:53] Adam Xinpeng: Actually there are a few things like that which are really just bad-ideas-in-motion.
[9:53] Morgaine Dinova: The wiki doesn't represent the group. It represents the people who put in the effort to write on the wiki :-) All we can do is encourage you to assist though. :P
[9:53] Goldie Katsu: My impression is that they want input on a specific area which they believe will lead to interoperability and scalability.
[9:54] Zha Ewry: They do.. and that's fine
[9:54] Zha Ewry: But.. for example
[9:54] Turtle: Hello, Avatar!
[9:54] Zha Ewry: If I build and OpenSim with a 1024x1024 grid
[9:54] Zha Ewry: *I* need to be able to mark that
[9:54] Zha Ewry: They keep going "We don't plan to change the grid"
[9:54] otakup0pe Neumann: That
[9:54] Zha Ewry: which.. is just.. so limiting
[9:54] otakup0pe Neumann: 's a perfect example - LL just said "no way in hell" to non-standard grid sizes.
[9:54] otakup0pe Neumann: i use the word standard quite lightly.
[9:54] Zha Ewry: We need to allow them to have the grid
[9:55] Saijanai Kuhn: did you see the responses to my original "Second Life 2.0" jira? they don't dare admit to any such plans
[9:55] Zha Ewry: as they have it. Make it easy for them to have a proifile, which is Linden as is.
[9:55] Morgaine Dinova: otakupOpe: who said that?
[9:55] Zha Ewry: But.. We must open up evolutionary pressure
[9:55] otakup0pe Neumann: morgaine not in as many words, but that floated across sldev early last week i believe.
[9:55] Burhop Piccard: Zha - right. So I have no trouble with them wanting to smoothly migrate to being "open" the the end goal should really be "open".
[9:55] Zha Ewry: People will do it anyway. so denying it is silly
[9:55] Zha Ewry: I keep pounding on this
[9:55] Zha Ewry: If I have an openSim, I can, and will make ones which vary from SL
[9:55] Adam Xinpeng: Region sizes is an interesting topic actually -- does that really need to be in the standard at all?
[9:55] Zha Ewry: Which is GOOD
[9:56] otakup0pe Neumann: The catch seems to be encouraging LL to cooperate, even where it's not directly "in their interests"
[9:56] Zha Ewry: Yeah
[9:56] Goldie Katsu: From what Zero put on the AWG page it looks like he's talking open protocols, but not nesc. open source grid.
[9:56] Adam Xinpeng: Things like geography arnt nessecarily things we need to make a fixed standard.
[9:56] Adam Xinpeng: Goldie: one will follow the other.
[9:56] Zha Ewry: Adam, only to the extent that we can know what we need to know to run
[9:56] Zha Ewry: But o further
[9:56] Zha Ewry: *no
[9:56] Morgaine Dinova: Well, you have to interpret "No way in hell" as "No way in hell will LL add this today, until 3rd party subgrids do it and make us look silly."
[9:56] Saijanai Kuhn: Did anyone mention to you the reference client VAG idea, Zha?
[9:56] Adam Xinpeng: If LL doesnt release a reference grid implementation, I guaruntee you there will be an OpenSim one within a month or two.
[9:56] Adam Xinpeng: and there probably will be one anyway.
[9:57] Zha Ewry: Right
[9:57] Burhop Piccard: Adam, geography isn't the standard. How to comunicate the geography is.
[9:57] Goldie Katsu: Yes. But that also means that as long as the other grids talk the protocol correctly LL really has no say on what those grids look like.
[9:57] Zha Ewry: Exactly Burhop, and Goldie
[9:57] Morgaine Dinova: Sai: there hasn't been any discussion of thau kind today, afaik.
[9:57] Tillie Ariantho: I think what we are building in the wiki is the SL 2.0 100%, but we can't get that thus quick, and I think LL is a bit horrified by some ideas. Maybe we have to "highlight" the things that are required/doable right now. We can repackage the big things later and present when other things are added/fixed/finished. :)
[9:57] Zha Ewry: I should be able to say "My grid is a 2048x10 mobious strip" and within reason make it work
[9:57] Morgaine Dinova: Goldie: exactly
[9:58] Emma Nowhere: well, maybe you should abstract the notion of a grid?
[9:58] Goldie Katsu proffers a klein bottle
[9:58] Saijanai Kuhn: Its an idea we really need to do: we need a working, extensible, modularized client that can handle today's protocols AND whatever else the AWG comes up with
[9:58] Morgaine Dinova: Sai: yep
[9:58] Emma Nowhere: all sims within a grid probably need to share a common geomatry definition
[9:59] Emma Nowhere: so you could have a mobius grid with sims designed to fit in it
[9:59] Tillie Ariantho: I think LL want some grid stuff get done now, with only marginal changes to the client, yet...
[9:59] Goldie Katsu: Would it help if on the Wiki we indicated a division between "this is the next step" and "these are ideas for future steps".
[9:59] Emma Nowhere: but the LL grid, by necessity is going to be a flat 2d grid
[9:59] Burhop Piccard: Saijanai - yes.... And I think there has to be some Viewer discussion along side the AWG.
[9:59] Saijanai Kuhn: the LInden proposal, as it stands, will require any non-LL grid to have a single portal region/sim to the "main" grid.
[9:59] Burhop Piccard: (if its not part of the AWG)
[10:00] Adam Xinpeng: There's been a lot of whisperings about building a third client recently.
[10:00] Adam Xinpeng: third-party*
[10:00] Adam Xinpeng: But that's a side discussion probably best held later.
[10:00] Goldie Katsu: I can understand LL freaking out when we talk about mobius sims when they are having a hard time keeping the monolith grid together.
[10:00] otakup0pe Neumann: keep me posted plz.
[10:00] Zha Ewry: Yeah, because they view edge touch as trust, which is it
[10:00] Zha Ewry: But.. messsy to an extent, becuase the don't have a good trust model
[10:00] Morgaine Dinova: Burhop: The viewer is obviously a part of the system architecture, and no architectural design team can possibly ignore it and do a good job. So yes, it needs to be in there somewhere.
[10:01] Goldie Katsu: Yeah, I am coming up with a list of questions as I go through the proposed architecture.
[10:01] Burhop Piccard: Adam - I'd like to see the AWG set the standards nessary to allow 3rd party viewers.
[10:01] Code Tracer: yes i agree Burhop
[10:01] Morgaine Dinova: Burhop: just the protocols
[10:01] Zha Ewry: It has to Burhop
[10:01] Burhop Piccard: Morgaine - yes, but is it with how Zero is scaling back.
[10:01] Zha Ewry: The protocls and the conteacts
[10:01] Zha Ewry: contracts
[10:01] Adam Xinpeng: That's a given, but I think it's more along the lines of "a new client now"
[10:01] Morgaine Dinova: And we're free to make reference clients that implement them, of course.
[10:01] Adam Xinpeng: rather than in 18 months
[10:02] Zha Ewry: If I send you this, you must do that
[10:02] Morgaine Dinova: "You're expected to do that"
[10:02] Saijanai Kuhn is learning more about python. As long as you have a pretty printer IDE, its actually quite nice for prototyping
[10:02] Tillie Ariantho: IMHO LL want to do the server splitting stuff first, like having more than one asset server and such.
[10:03] Zha Ewry: yes, they want.. correctly, I tihnk, to move 90% of tihs to Web Services, which allow that
[10:03] Tillie Ariantho: That would help the current grid's stat, too.
[10:03] Zha Ewry: Which is.. about 50% of the battle
[10:03] Tillie Ariantho: state
[10:03] Morgaine Dinova: Tillie: they're parallel tasks
[10:03] Zha Ewry: The easy 50%, tho
[10:03] otakup0pe Neumann: easy of course being relative.
[10:03] Tillie Ariantho: Yes, but there aren't that many lindens...
[10:03] Tillie Ariantho: and I think they need 1 or 2 key topics to implement.
[10:04] Tillie Ariantho: So maybe we should finish the design for that stuff so they can start coding ...
[10:04] Goldie Katsu mumbles something about coding and testing practices and build trees.
[10:04] Tillie Ariantho: And meanwhile we discuss the 'scary' things. .)
[10:04] Morgaine Dinova: Yeah, I don't think we can expect Lindens to refactor the client. But they can sure assist with ideas
[10:04] Zha Ewry: I am trying to lay out some o fthat
[10:04] Burhop Piccard: Tillie - i'm not sure they want to code.
[10:05] Tillie Ariantho: Sure they want!
[10:05] Code Tracer: they want
[10:05] Burhop Piccard: I think they want to harness open source development to do more.
[10:05] Zha Ewry: The core web bits, so we can say "So, Asset, Sim, Agent, avatar-delgate.. Client.." really talk about coding that
[10:05] Tillie Ariantho: Didnt they say in some in the last meetings they have some people sitting there right now, wanting to do some programming till december?
[10:05] Tillie Ariantho: in one of the last meetings ...
[10:05] Code Tracer: and OS communitity too
[10:06] Tillie Ariantho: They need that asset server stuff etc now maybe with some example we build.
[10:06] Tillie Ariantho: Then they are busy for a while and dont interfere with our other ideas. :p
[10:06] otakup0pe Neumann: Zha, that seems like the best part to focus on if we are going to maintain a relationship with LL.... those are their pain points.
[10:06] Adam Xinpeng: Zha: this is a little bit of a diversion - but there's been a bit of discussion recently (and my apologies to Neas for bringing this up again) about defining this whole standard as a subset of XMPP -- much like Jabber is a subset of it.
[10:06] otakup0pe Neumann: O_o
[10:06] Zha Ewry nods
[10:06] Zha Ewry: I think that's a little off course, but interesting
[10:06] Adam Xinpeng: It makes a lot of sense after you look at the standard itself - it also cuts down on the amount of "new standards" and work that needs to be done.
[10:07] Zha Ewry: XMPP, isn't REST...
[10:07] Adam Xinpeng: No it's not.
[10:07] Zha Ewry: and.. thare are some good reasons to be scared of that, in things like assets
[10:07] Adam Xinpeng: and it's not suitable for high volume data either (eg agent connections)
[10:07] Neas Bade: though the Intel folks had some good comments of xmpp not scaling well
[10:07] Adam Xinpeng: Yeah they did.
[10:07] Morgaine Dinova: Adam: we were discussing solving some of LL's IM problems by replacing it lock stock and barrel with commercial XMMP backend. It didn't get much support though.
[10:08] Adam Xinpeng: That's probably because you'd need to shift IM to the client completely to do that
[10:08] Zha Ewry: One t thing, Linden would love to see, is duscussion like that, and about the c-hhtp isdeas, and escrow and capabilities
[10:08] Emma Nowhere: XMPP could certainly address a lot of the pieces
[10:08] Unmasked Shepherd: sorry to be so quiet everyone, but i am trying to get myself started up here and that requires a lot of effort, pills and a little bit of food hheheh
[10:08] Adam Xinpeng: XMPP-IM isnt suitable for how the grid works right now due to IM's being handled by the region your in.
[10:08] Neas Bade: honestly, the REST stuff in opensim is actually starting to look pretty promissing
[10:09] Goldie Katsu: We need to give them something they can move to now without a complete redesign, and but with an eye towards where we think they should be going to.
[10:09] Tillie Ariantho: Zha: yes, think so... they dont mention the protcol stuff, REST and all that for fun. They want to do some work there now and are waiting for some pre-work with examples. .)
[10:09] Goldie Katsu: Replacing the whole back end isn't likely.
[10:09] Neas Bade: at some point we'll need to seed in the chttp stuff so we can have the reliability characteristics we need, but even without that it's doing quite well
[10:10] Zha Ewry: I think, Zero.. would love to see some focus on this, on what an asset service really means
[10:10] Zha Ewry: And.. to an extent. how to cache and handle assets in general
[10:10] Zha Ewry: And. whether they know it or not
[10:11] Neas Bade: honestly, I'd love to see an asset definition as part of that :)
[10:11] Zha Ewry: They need to think about reliable presence and such
[10:11] otakup0pe Neumann: zha, you have a good point.
[10:11] Zha Ewry: Because I think that's one reason they are having so much trouble at the moment
[10:11] Adam Xinpeng: "do we need a new standard?" - HTTP?
[10:11] Goldie Katsu: So, what is our scope. LL wants the protocols for a more component like grid that would lead towards interoperability and distribution across multiple providers. What do we see as our scope that is beyond that.
[10:11] Adam Xinpeng: http://some.server/some.file.j2c ?
[10:11] Morgaine Dinova: Here's a link for XMMP scaling to a million: (which is a good start) --- is there a similar link to Intel saying why it doesn't scale?
[10:11] Zha Ewry: Allowing varients is the big bit
[10:11] Zha Ewry: That they resist
[10:11] Adam Xinpeng: Morgaine: yes - check the OpenSim-dev lists from this week, look for "Mic Bowman"
[10:12] otakup0pe Neumann: a new viewer and mobius strip grids are fun but..... it's certainly not going to get ll excited.
[10:12] Neas Bade: Adam, you still need waht your data packet looks like
[10:12] Emma Nowhere: scalable IM is not something new
[10:12] Zha Ewry: That's what i thinhk Linden is missing, scared of, that we care about a lot
[10:12] Neas Bade: we're adhocing that right now, which works fine for just OpenSim, but it would be nice to have some true interchange
[10:12] Tillie Ariantho: otakup0pe: true, they need some 'real' stuff to work on, fun can come later. .P
[10:12] otakup0pe Neumann: variants ?
[10:12] Emma Nowhere: and the level of chat and IM within SL today is not some unheard of level
[10:12] Adam Xinpeng: XMPP != Jabber/XMPP-IM
[10:12] Burhop Piccard: I'd like so see more more on assets too. In talking about geomtry, we sometimes slip into assets (since geometry can be a sub-piece of an asset)
[10:13] Goldie Katsu: What do you mean by variants, Zha?
[10:13] Unmasked Shepherd: p2p group IM sessions
[10:13] Zha Ewry: Things like sims with different sizes, different prim types, different physics.
[10:13] Morgaine Dinova: There's nothing at all hard about scaling IM arbitrarily, so don't think it's an issue. Using an existing system is just a time saving, that's all.
[10:13] Zha Ewry: Making sure that we can do the right things in the design to make that easy, not hard
[10:14] Burhop Piccard: Zha -right!
[10:14] Goldie Katsu: So designing the protocols to allow more variation?
[10:14] Burhop Piccard: Goldie- that is what I think.
[10:14] Zha Ewry: Protocols, and the core contracts.
[10:15] Zha Ewry: The thing is, if you don't people iwll anyway. Just badly
[10:15] Goldie Katsu: Core contracts being?
[10:15] otakup0pe Neumann: i'm assuming social/legal contracts ?)
[10:15] Morgaine Dinova: Zha, are IM services already defined within your REST work, or is that still left to do?
[10:16] Wyn Galbraith: Morning.
[10:16] Morgaine Dinova: Hi Wyn
[10:16] Code Tracer: hello Wyn
[10:16] Goldie Katsu: Heya Wyn
[10:16] Zha Ewry: I am not sure, I hadn't thought about it for long, until recently. It's actually a litlte messy.
[10:16] Zha Ewry: The protocols, are easy, in some sense
[10:16] Morgaine Dinova: kk
[10:17] Zha Ewry: it's the modelling, of how to make it realy scale, with, large groups and lots of state change that's icky
[10:17] Zha Ewry: I mean, SL today, has groups with order 5-10,000 members
[10:17] Zha Ewry: Which means. you add and delete continually
[10:17] Morgaine Dinova: Yeah, it's noddy, and still it doesn't work, lol
[10:17] Zha Ewry: So. you want that pub/sub style
[10:18] Goldie Katsu: So LL would like the scope to be protocols that allow for partitioning the grid into a new architecture (theoretically the one Zero has proposed.) We want to expand that to include making the protocols extensible/robust enough to handle grid variation & address core contracts. Is that right?
[10:18] Goldie Katsu: (And I'm still working on understaing core contracts)
[10:18] Tillie Ariantho: I think so.
[10:19] Burhop Piccard: Goldie - I'd agree with that.
[10:19] Zha Ewry: Roughlyu
[10:19] Zha Ewry: LL wants the full thinjg
[10:19] Zha Ewry: They really do
[10:19] Zha Ewry: Just.. they are scared
[10:19] Goldie Katsu: They just don't know it.
[10:19] Zha Ewry: They need to be able to do
[10:19] Morgaine Dinova: Goldie, in part, yes. But more generally, it's to address everything that could limit scalability and interop. Zero's interests are more narrow than that, so he'll just pick up on a few of the topics.
[10:19] Tillie Ariantho: yes, so we need to give them the chocolate in small bits, starting with the protocol stuff. .)
[10:19] Zha Ewry: 20 releases, in the next 2 years, and get a lot closer to the full thing
[10:19] Burhop Piccard: Goldie- my own personal concern would be variations in Geometry)
[10:19] otakup0pe Neumann: zha, also timing. they want some things right away, some six months ago, and some things can wait a little while
[10:19] Zha Ewry: Aye
[10:20] Burhop Piccard: Neumann - laying ot the design doens't mean it all has to be implmented.
[10:20] otakup0pe Neumann: i know
[10:20] otakup0pe Neumann: i never said it did.
[10:20] Goldie Katsu: So how do we address our focus so that we give them what they ready for now yet plan and move towards the future (work towards what we know is needed.)
[10:20] Zha Ewry: I'd like to see two things,
[10:20] Morgaine Dinova: Aye. Many parts will be implemented only by ourselves as reference designs, or by competitors.
[10:21] Burhop Piccard: Sorry - just didn't want anyone to assume it would be too big to do.
[10:21] Zha Ewry: I'd like to actually have a focused effort on the core defintions and basic building blcosk Becuase. if we do that.. It will help. A lot
[10:21] Zha Ewry: I'd also like us to lay out a way to work on both things at once
[10:21] Zha Ewry: Core near ter stuff, like c-http, and escrow
[10:21] Morgaine Dinova: No doubt OpenSim will be in there too. Even though OpenSim is not working on scalability at this time, it will in general, or it will become irrelevant.
[10:21] Zha Ewry: and.. the whole picture, so we can say 'This is short term, this is long term"
[10:22] Zha Ewry: And. we need a good story abnout how OpenSim walks along with us
[10:22] Morgaine Dinova: Zha: a combination of "kicking and screaming", and "way ahead of us" :-))))
[10:22] Tillie Ariantho: haha
[10:22] Zha Ewry: LOL
[10:22] Zha Ewry: Yeah
[10:23] Burhop Piccard: Zha - yes. a project managers view of the architecture :-)
[10:23] Goldie Katsu: Is it possible to have short term and long term sessions? Focus on one at one meeting and the other at another? (just throwing out an idea.)
[10:23] Zha Ewry: I think so, yes, Goldie
[10:23] Tillie Ariantho: good idea. :)
[10:23] Saijanai Kuhn: Last night we experimented. If all VAG members exchange calling cards, they can create ad-hoc group IM whenever they want
[10:23] Saijanai Kuhn: so the VAGs can have their own private IM whenever
[10:23] otakup0pe Neumann: yeah, calling card or friends
[10:23] Morgaine Dinova: Goldie: It's a great way to focus. Just as long as the near-term doesn't become "everything".
[10:24] Goldie Katsu: Yes, we need to have both. And the long term does inform the near term.
[10:24] Zha Ewry: absolutely
[10:24] Zha Ewry: which. is where, I worry Linden's urge to keep focused, hurts
[10:25] Zha Ewry: And.. I want to, if posisnble
[10:25] Zha Ewry: make sure we cross link the efforts, so we say 'Here, we're pushing this off to the longe term effort" explicitly
[10:26] Burhop Piccard: Yes, so a roadmap.
[10:26] Morgaine Dinova: It's OK, I can't blame Zero for having specific interests. But we need to make the point that other things have to be considered, eg. client-side architecture is part of the picture. I think he'll accept that, despite not working on it himself.
[10:26] Zha Ewry: Yep
[10:26] Zha Ewry: I think so
[10:26] Zha Ewry: If we're clear and expcliti on things
[10:26] otakup0pe Neumann: the last part is key morgaine
[10:26] Zha Ewry: I think, that the thing that scares me
[10:26] Zha Ewry: (in his shoes)
[10:26] Goldie Katsu: I think it may be a matetr of building a trust. As LL sees we are contributing to wha tthey think they need they will be more open to a broader focus.
[10:27] Zha Ewry: is that they need to do it stepwise
[10:27] Zha Ewry: They are fearful, of being pushed to fast, and putting effort in the "wrong" place
[10:27] otakup0pe Neumann: like mobius-strip grids.
[10:27] Saijanai Kuhn: ringworlds at least
[10:28] Zha Ewry: Frankly, at some level, any grid other than 512sqare
[10:28] Zha Ewry: They have good reasons
[10:28] Wyn Galbraith: LOL
[10:28] Zha Ewry: When you look at it, it does simplify a bunhc of stuff
[10:28] Burhop Piccard: So our protocol suports topology and the only one we implement is a plane :-)
[10:28] Zha Ewry: its deeper than that
[10:28] Wyn Galbraith: Mobius-strip grids would make a good space.
[10:28] Zha Ewry: Here's why.
[10:28] Chase Marellan: If a flat grid is the biggest of our problems, I think we're in good shape.
[10:28] Burhop Piccard: And the mobiusistas do their own open source grid.
[10:29] Zha Ewry: The current scheme, lets the client always know who owns every bit of the space it is rendering
[10:29] Morgaine Dinova: (Cueing Tillie ...) Well they'd better not be fearful of different topologies, because Aperture Science worlds are probably already being designed.) :P
[10:29] Zha Ewry: and.. all the other scheme. require sharing and marking that.. which.. means a bunch o fnew issues
[10:30] Alexander Regent: yay...I am an official groupy :)
[10:30] Goldie Katsu smiles
[10:31] Alexander Regent: far this mainland...or just a really big private estate? :)
[10:31] Zha Ewry: Here?
[10:31] Zha Ewry: One of IBM Research's private islands
[10:32] Zha Ewry: Its fun to be a goddess!
[10:32] Goldie Katsu laughs
[10:32] Code Tracer: ...
[10:32] Code Tracer: lol.
[10:32] otakup0pe Neumann: O_o
[10:32] Alexander Regent: we know who NOT to pissoff :P
[10:33] Zha Ewry grins
[10:33] Morgaine Dinova: So, how do we get more participation by SLdev techs in AWG? We seem to be split into two worlds atm.
[10:33] Zha Ewry: I do pay the bills
[10:33] Code Tracer: let's carry on folks.
[10:33] Zha Ewry: anyway
[10:33] Burhop Piccard: Morgaine - yes.
[10:33] Alexander Regent: what are the two worlds? SL and AW?
[10:34] Chase Marellan: My sense, and I'm not claiming to be a died in the wool tech, is that short term practical application that can actually be accomplished might attract them.
[10:34] Burhop Piccard: Alex - short term thinkers and long term thinkers :-)
[10:34] Chase Marellan: :)
[10:34] Morgaine Dinova: Hehe, no. AWG and SLdev, where SLdev is a fuzzy melting pot (which is great), but hard to actually interop with AWG.
[10:34] Zha Ewry: We need an openSim story. one about how we get things fuly in sync
[10:35] Alexander Regent: lol..and what do you mean by "Things"
[10:36] Alexander Regent: as an SLDev person.....I like the way they define things :)
[10:36] Burhop Piccard: ENTJ versus ESTJ
[10:36] Zha Ewry: We need, over time.. to get this towards a place, where it looks like a full up
[10:36] Wyn Galbraith: Things are like objects.
[10:36] Zha Ewry: set of standards
[10:37] Zha Ewry: We need a wiki, or such, to do that
[10:37] Burhop Piccard: So I'm not plugged into Opensim. Where's a good place to start?
[10:38] Zha Ewry: Where when we say "Calling the asset server, with a request that looks like X"
[10:38] Zha Ewry: we can say it in enough detail, that one can build it
[10:38] Alexander Regent: at the website I would presume :)
[10:38] Alexander Regent: opensim still has features missing though yes?
[10:39] Morgaine Dinova: Well that's one example where there's a small disconnect, just terminology. We tried to help by creating a Glossary of the terms as used in AWG, but since so few people popped over from SLdev to help that there probably still isn't a joint view.
[10:39] Zha Ewry: OpenSim.. yeah
[10:40] Zha Ewry: Lke physics, tho, ODE is getting really close. Was in an OpenSim today, puswhing cubes around, and flying
[10:40] Zha Ewry: And scipritng, and deep grid mode
[10:40] Alexander Regent: last time I check..collision det4ect was still an issue
[10:40] Alexander Regent: is this gloassary posted somewhere?
[10:40] Code Tracer: ODE handles it
[10:40] Zha Ewry: Getting really close, Alexander, at least for basic cubes
[10:41] Zha Ewry: It son the SL wiki
[10:41] Morgaine Dinova:
[10:41] Zha Ewry: And. it's about 1/3 wtrong, at least
[10:42] Goldie Katsu: So is the architecture of opensim segmented or a different monolith?
[10:42] Zha Ewry: And.. watch out for the oddly written one liners.
[10:43] Zha Ewry: Segmented, in somewhat nicer ways
[10:43] Zha Ewry: But.. it's also a 1.0 effort
[10:43] Morgaine Dinova: It's evolved from a list of one-liners to a page of explanations, so work in progress.
[10:43] Morgaine Dinova: But for the most part it's at least self-consistent.
[10:43] Zha Ewry: The problem, is that you can't tell, easily, whch are the one lines, from random people, and which are more considered. But.. help there is appreciated, would be really good
[10:44] Zha Ewry: (No, morrg, it's still got about 1/3 nonsense, last I looked, the nonsense, is not even self consistent always, and we need to drive out the nonnesen)
[10:44] Goldie Katsu: Dumb question here. Is the architecture of opensim documented beyond the code?
[10:44] Morgaine Dinova: Zha: oh dear. I gathered someone's been "helpful" again ....
[10:45] Zha Ewry examines her nicely french tipped fingernails
[10:45] Morgaine Dinova: Goldie: great question :-)
[10:45] Saijanai Kuhn: libsl isn't well documented either that I can tell
[10:45] Tillie Ariantho: Coder are seldom documenters, too. .)
[10:45] Alexander Regent: is not :)
[10:45] Alexander Regent: lol
[10:46] Zha Ewry: It is tho, OpenBS, so anyone can read it safely
[10:46] Zha Ewry: *BSD
[10:46] Saijanai Kuhn: as opposed to GPL?
[10:46] Goldie Katsu: Well I don't mean the code is documented, I mean that if you read the code you get a picture (assuming you have code reading skills.)
[10:46] Saijanai Kuhn sighs heavily
[10:46] Morgaine Dinova: Funnily enough, this is one area where the hackers will be streets ahead of LL. The official answer we've received from Lindens is "read the code".
[10:46] Wright Juran: most of opensim is fully documented, the problem is getting most of that out of my head and onto the wiki
[10:47] Zha Ewry: Wright, I have machine here.
[10:47] Zha Ewry: we put your head in
[10:47] Goldie Katsu: So it is fully documented in your head?
[10:47] Zha Ewry: Grind it up real fine
[10:47] Morgaine Dinova: Hahaha
[10:47] Goldie Katsu: ewww
[10:47] Saijanai Kuhn: ewwwww
[10:47] Zha Ewry: And inject the mRNA into everyone's head
[10:47] Zha Ewry: You ok with that?
[10:47] Chase Marellan: just don't get any in your keyboard
[10:47] Saijanai Kuhn: isn't there a guy on Heroes with that power?
[10:47] Chase Marellan: liquified brains general void your warrantee
[10:47] Chase Marellan: generally*
[10:48] Zha Ewry: Well, tech support, is pretty limited anyway
[10:48] Zha Ewry: Seriously, tho. It is somewhat more accasile
[10:48] Zha Ewry: accesisblek in temrs of license
[10:48] Zha Ewry: And.. the SIM code, and the servcies, at lindne
[10:48] Zha Ewry: isn't open at all
[10:48] Goldie Katsu: So the lessons learned from Opensim, can they be applied to what LL is ready to listen to now?
[10:49] Zha Ewry: Some of them, for sure
[10:49] Alexander Regent: like what?
[10:50] Zha Ewry: OpenSim has done some work towards fully REST asset services, for example
[10:52] Goldie Katsu: That could be useful to see.
[10:52] Morgaine Dinova: We were discussing yesterday whether it would be easier to use libsl to create a client comms mux/demux front-end with a local socket-based API onto which to attach everything else, or whether to recode the front-end process again. Any views?
[10:52] Morgaine Dinova: Client-side.
[10:53] Lalinda Lovell: recode
[10:53] Saijanai Kuhn: do we replace on black box with another? I can't find the documentation for what is going on
[10:53] Morgaine Dinova: Ew
[10:53] Lalinda Lovell: libsl is tied to all the current CAPs
[10:54] Goldie Katsu: CAPs?
[10:54] Lalinda Lovell: look on the wiki
[10:54] Alexander Regent: rtfm? lol
[10:54] Morgaine Dinova: Lalinda: so you're saying to basically write the front-end process as the REST API appears, I take it, in step.
[10:55] Lalinda Lovell: yes
[10:55] Morgaine Dinova: kk
[10:55] Zha Ewry: Something like that, to me, seems plausible
[10:55] Zha Ewry: Linden, tho
[10:55] Zha Ewry: Is missing at least two bits inthier commo story, thus far
[10:56] Morgaine Dinova: Zha, your typing today sucks :-)))))
[10:56] Zha Ewry: (linden is missing at least two key conceost thee)
[10:56] Zha Ewry: Heh
[10:56] Zha Ewry: I am at home.
[10:56] Zha Ewry: Laptop, not the normal keyboard in my lap
[10:56] Zha Ewry: It hurts
[10:56] Morgaine Dinova: Heh
[10:56] Zha Ewry: Anyway
[10:56] Zha Ewry: They are missing a model for how to susbcribe to ongoing events
[10:57] Tillie Ariantho: yes
[10:57] Zha Ewry: ie. "Tell me the current location of my Ave, as it moves"
[10:57] Zha Ewry: And.. the model of the UDP like stuff
[10:57] Zha Ewry: Related
[10:57] Zha Ewry: and both un specced for the future
[10:58] Zha Ewry: We really need to talk to them about the former
[10:58] Zha Ewry: Because the "tell me about X" pattern is really important
[10:58] Lalinda Lovell: can i ask something please?
[10:59] Zha Ewry: Sure
[10:59] Chase Marellan: pardon me for being new to the conversation, but what about existing grid-related specs (ie, WSRF)
[10:59] Zha Ewry: WSRF... Heh
[10:59] Chase Marellan: well, just as an example. :)
[10:59] Lalinda Lovell: just a very basic question on time scale here, is there a set date that something definitive is needed by?
[11:00] Zha Ewry: Linden wants ot beta-grad the agent doman split in the next 7 weeks
[11:00] Lalinda Lovell: ok thanks
[11:00] Zha Ewry: Well, by YE07, but close enough
[11:01] Zha Ewry: WSRF...
[11:01] Zha Ewry: Not quite the opposite of REST, but... man, heavyweight
[11:02] Chase Marellan: true, true.
[11:02] Chase Marellan: I was just asking because you mentioned event subscriptions.
[11:02] Chase Marellan: :)
[11:02] Zha Ewry: And.. it puts you into the whole "just what tooling supports this?" wards
[11:02] Zha Ewry: *wars
[11:02] Zha Ewry sat in on MOWS and MUWS nd WS-RF for a while
[11:02] Zha Ewry ran away screaming
[11:02] Morgaine Dinova: The closest we've seen discussed about stateful services is with chttp for escrow services, at Which Linden's.
[11:03] Zha Ewry: Right. Which is very carefully crafted, to keep the state in one, logged place
[11:03] Chase Marellan: hahahaha.... I did some WSRF tutorials for developerworks and had the same reaction. :)
[11:03] Chase Marellan: gotcha.
[11:04] Zha Ewry: But..
[11:04] Chase Marellan: ack, gotta pick up my daugter, sorry, back shortly
[11:04] Zha Ewry: at the same time
[11:04] Zha Ewry: One can look at some of the discusisons on meta-data, identity, addressing, and enevlope.. and see
[11:04] Zha Ewry: some very familair issues
[11:04] Tillie Ariantho: yeh
[11:05] Tillie Ariantho: So how do we go from here?
[11:05] Tillie Ariantho: We need to hand over some work to the Lindens soon. .)
[11:05] Morgaine Dinova: Simple: Zha figures out what has to be stateful, then we tie Zero down and tickle him until he talks about them ;-)
[11:05] Goldie Katsu: I would ask is there any work already done that we can use as a starting point - beyond what is on the wiki.
[11:06] Zha Ewry: LOL
[11:06] Zha Ewry: Anyone willing to go to mountainview with some stout rope?
[11:06] Goldie Katsu: I have soem bungie cords you can use.
[11:06] Morgaine Dinova: Hehe
[11:06] Zha Ewry: People lookking at the wiki, pounding on it
[11:07] Zha Ewry: Would be good
[11:07] Morgaine Dinova: Zha: did your message on SLdev attract anyone to your REST services VAG?
[11:07] Zha Ewry: So, we can actually say "The avatar-delgate" offers a service which allows anothter endpoint to susbcribe to the followin updates
[11:07] Zha Ewry: I think so, yes
[11:07] Zha Ewry: (morg)
[11:08] Zha Ewry: The reason to have temrs like 'avatar-delate" and "asset-server" and if we need it "Local asset cache" formally defined, is so we can talk about which bits need to be done with
[11:09] Zha Ewry: simple HTTP and REST. Which with c-http, and which with escrow
[11:09] Morgaine Dinova: I'll post a couple of messages in SLdev too to try to get people involved in Scalability or QA VAGs.
[11:09] Zha Ewry: where event streams happen
[11:09] Goldie Katsu: and would it be avatar delegate or agent?
[11:09] Zha Ewry: Heh
[11:10] Zha Ewry: the agent, in SL 1.18 breaks into three bits
[11:10] Zha Ewry: (Per Zero)
[11:10] Zha Ewry: The inv and IM/presence bits, which go to the agent domain
[11:10] Morgaine Dinova: Zha: oh, reminds me that I posted some comments on one of your "melding" Talk pages.
[11:10] Zha Ewry: Then, the rements, (what's left of agent, and child-agent) become the avatar-delegate, and the avatar-delagte-children
[11:11] Zha Ewry: Oh. I will go look
[11:11] Goldie Katsu: ahh Ok.
[11:11] Zha Ewry: And. Thanks. I really want pepole to look at that
[11:11] Zha Ewry: The thought, being, that with the agent bits one, the bit on the region simulator, is really the Avatar's presence in the software world
[11:11] Morgaine Dinova: And also on Burhop's, about his Geometry-oriented view of archiecture.
[11:12] Morgaine Dinova: Hmmm, wasn't Burhop here?
[11:12] Goldie Katsu: he was
[11:12] Morgaine Dinova: Ah, gone
[11:12] Saijanai Kuhn: people logging out (or crashign)
[11:12] Saijanai Kuhn: rolling restart today, right?
[11:12] Goldie Katsu: Yes but this region hasn't. (obviously)
[11:13] Zha Ewry: yeah.. and I can postpone it an hour ;-)
[11:13] Wyn Galbraith has to run. "I'm at my son's. So busy."
[11:13] Zha Ewry: Well, we've spent a tone of time
[11:13] Morgaine Dinova: That itself needs to be addressed for a new arch ... you don't restart worlds as a standard MO.
[11:14] Wyn Galbraith notices that she's behind in her glossary reading, "I'll work on that."
[11:14] Goldie Katsu: So do we have a clearer definition and an idea of hwo we are going forward?
[11:14] Goldie Katsu: *how
[11:14] Zha Ewry: A bit
[11:14] Zha Ewwry: Someone want to post all of this to the wiki? I'll try and distill it
[11:15] Morgaine Dinova: Well Sai and I are going to prime the pump on client futures, because we couldn't get anyone else to pick up the ball and we're suckers for punishment :-)))
[11:15] Saijanai Kuhn has new business...
[11:15] Saijanai Kuhn: thats it
[11:15] Zha Ewry: Excellent
[11:16] Morgaine Dinova: But unless we can then draw in the tons of client-side experience in SLdev and libsl and elsewhere, it's all for naught. We'll try though.
[11:16] Saijanai Kuhn: we figure we need something that can handle today's design, and morph into the future.
[11:16] Saijanai Kuhn: otherwise we can't test any of this stuff effectively
[11:17] Zha Ewry: Yes. We need them
[11:17] Morgaine Dinova: This is why I have one foot in the QA VAG, Sai :-)))
[11:17] Saijanai Kuhn: and the current client design... is the current client design. 'nuff said.
[11:17] Saijanai Kuhn: somenoe said that we couldn't (or the Lindens couldn't) factor the client from scratch. It MUST be done
[11:18] Zha Ewry: Broken into four or so modular chunks, if you ask me
[11:18] Zha Ewry: But.. I am not going to go there
[11:18] Saijanai Kuhn: with sub modules
[11:18] Zha Ewry: (input/output de-mux, model, rendering, gui) with plug points
[11:18] Saijanai Kuhn: and scripting/building
[11:19] Morgaine Dinova: We'll see. I'm hoping that by focussing on test-driven development of a front-end that keeps in step with Zha's REST work, we can kill 2 birds with one stone ... ie. test, reference/validate, and end up with a new client front-end. OK, so that's 3 :-)
[11:19] Saijanai Kuhn: plus "other". Someone might want an address book embeded in the SL clinet for example. CLose friend's external IM/email/etc
[11:21] Morgaine Dinova: If we can achieve this ... my guess is that people will attach things to it that we've never imagined.
[11:21] Zha Ewry: Aye, it plugs in on the
[11:21] Tillie Ariantho: Sounds good. :)
[11:21] Zha Ewry: mux./demux for events, and the gui/render to get out at poerople
[11:21] Morgaine Dinova: This is no small job though.
[11:22] Zha Ewry: Yep
[11:22] Zha Ewry: But would be good
[11:22] Zha Ewry: OK.
[11:22] Zha Ewry: I need to runi