Second Life Grid Architecture Meetup Transcript.

So here is the transcript of the session yesterday, I hope everybody is ok with that, if not, please tell me and I will remove your parts. You can find the slides in this flickr set.

You: ok, I want to talk about how the new Second Life Grid architecture might look like
You: as you might have read on the blog or on slashdot there is some project underway (actually just started) to change how the grid is working and this is supposed to be a little introduction
You: if you want to know more about me or how to connect with me, feel free to check out
http://mrtopf.de/connect,
everything should be on there
You: so first of all about why we need a new grid architecture
You: in some numbers
You: imagine instead of having 10.000 region, 60 million of them
You: or instead of having just 10 million users we might have 2 billion users
You: this is about the amount if email addresses worldwide, not everybody will be active of course
You: the 60 million is half the amount of webservers out there
You: and instead of 50.000 concurrency we might have 50 million people online at once in the grid
You: of course in the future “being online” might mean something more lightweight
You: be it on a blackberry, your microwave, TV or hair dryer
You: so this is about scaling
You: but it is also about further requirements people already have
You: this is how Second Life looks today, basically it’s a big black box and viewers can connect to it

You: but you cannot host your own sim for instance
You: (except opensim but this is not connected to the main grid then)
You: and there is also the problem many people I’ve spoken to have with SL
You: many created an account but did not really get it diretcly and left again
You: and I meet them at Barcamps etc. so they are mostly bloggers. and many of them dream about a web3d what they call it
You: which is a term I cannot really do much with 😉
You: but I imagine 2 things with it: 1) a grid architecture similar to the web where everybody can connect their own server
Expert Market: SOL is not web, not web3, 3.0 or 3D 🙂
You: and 2) some mix of 2d and 3d where you do your research in 2d but you can interact with other people about your results in 3d
You: imagine Amazon
You: you search for a book in a 2d page and can then jump into a conversation like here about the author etc.
You: this might be directly connected
You: so if we look at the definition 1 it might look like this

You: we have many regions
You: some might be in subgrids (the dotted box)
You: you might also be able to host your own sim at home
You: which is only online when your computer is
You: or completely disconnected
Lem Skall: Tao, will the definition of “region” stay the same?
You: on your laptop when you are travelling (although in 5-10 years maybe there is no place without internet anymore anyway).
You: Lem: probably as LL wants to port the existing infrastructure to the new architecture
You: I will talk more on that later
Lem Skall: lol, like “your flies are open”
You: so this basically is the grid architecture today

You: we have the grid with it’s many regions
You: and the viewer mainly connects to a region which does all the stuff for it
You: apparently it simulates all the other avatars around you, the objects etc.
You: but it is also responsible for getting inventory listings to you if you say so
You: so it also acts as proxy to the central databases
You: now with the mentioned numbers you can imagine that these databases might soon have a problem
You: they seem to have it already 😉
Expert Market: tao : soon ? or since 2 years already ? 🙂
Tillie Ariantho: haha
You: heh 🙂
Tara5 Oh: i never crashed much up until this last week?
You: ok, so we met on Sept 13th for some meeting where Zero Linden was explaining us Linden Lab’s proposal
You: this looks like this

Expert Market: blurry 🙂
You: well, the future is not completely clear yet 😉
Tillie Ariantho: hihi
Expert Market: 🙂
You: actually it really is sort of blurry 😉
Jeffronius Batra: 🙂
You: these are just basic ideas, the details are still mostly missing
You: so we have now 2 domains, an agent domain and a region domain
You: the agent domain knows everything about an agent
You: like name, profile, inventory etc.
You: the region domain consists of a bunch of simulators and knows everything about regions, their name, probably location, what’s on them and so on
You: please don’t click! 😉
Fleep Tuque: Agent = avatar, I assume, but would also include bots?
You: An avatar is the 3d representation of an agent
You: so agent == user
Fleep Tuque nods.
Saijanai Kuhn: with an open source client, a bot is a given thing , whether “allowed” or not
You: so the viewer needs to connect to both domains to first login the agent and then connect to the region
Oliver Gassner: does the agent also run on a separateand might it be we have free agents who run slowly and #special’ dedicated agents that cost?
Saijanai Kuhn: I’ve suggested that the LIndens use scripted bots for testing anyway
Kiwi Alfa: Saijanai: Bots existed even before the viewer was open sourced, you know.
You: well, agent just means your profile information basically
Saijanai Kuhn: much easier now, though
Fleep Tuque: And both the agent domains and region domains are hosted on your own or a remote server? ie not LL?
You: so the agent domain consists of some web services which allows to login, to retrieve inventory etc.
You: Fleep, they can be
You: look here
Expert Market: so we can expect less server lag in a crowded region ? it’s kinda “multithreading” the region and split the load in half… with twice more core… so we could get 100.000 user online, still not 60 millions ^^
You: here we have several region and agent domains

You: and you can imagine one agent domain to be allowed to connect to differenr region domains
Jeffronius Batra: Expert, I think that’s separate.
You: an new region domain might be some social network
Susi Spicoli: why can’t we have all the idle servers help me out when I have an event?
You: actually it’s not that much about how a region is implemented
You: the idea is to define interfaces for the viewer to interact with a region
You: the region can then be implemented as you wish
You: it can be a cluster
You: or like now one CPU core
You: or your microwave in standby mode
Oliver Gassner: so the regions become scalable?
Jeffronius Batra: And then separated both logically and physically if desired.
Lem Skall: could we connect to the agent without connecting to a region? or connect only to a region but not the agent?
Tillie Ariantho: I like microwaving newbies. 😀
You: I guess you might be able to connect an agent without a region, e.g. to sort your inventory
You: but a region without an agent does not make sense
Frans Charming: or im with friends tao
You: or that
Ian Betteridge: What are the implications for the money system? That’s not going to be part of the agent domain, is it?
You: although IM maybe should also be pluzggable
Lem Skall: actually view a region while invisible
You: Ian, that’s still open
You: there are 4 areas quite blurry
Ian Betteridge nods
Expert Market: Here, i see more server to suck to ressource of asset server
You: identity, currency, location and search
Bell Boyd: Script run-time error
Bell Boyd: Stack-Heap Collision
You: although I don#t see search as that blurry
Oliver Gassner: Did I misudnerstand or can the agent run n ym own server? So I have an avatar that I could use in ‘different’ worlds?
You: I think there should be APIs for collecting data and then search will be done by Google 😉
Tillie Ariantho: I know 4 areas, that are open, too. Reliability, stability, speed and security. .)
You: Oliver: actually there is the concept of identity
You: you might have on identity and different agents attached to it
Miki Gymnast: Tao, whats their mid- or long-term strategy? Island servers, hosted outside LL, agent, search and money servers hosted @ LL?
You: like you can have a corporate agent from your company agent domain and you can have a more public LL agent for the SL grid
You: but you may be able to login to both via e.g. OpenID or another system
Oliver Gassner: I am not sure this answers my qiuestion and that I understood the answer 😉
You: maybe I did not understand your question 😉
Tillie Ariantho: haha
Fleep Tuque: Hm, that’s an interesting concept. one identity with multiple agents.
You: but the agent domain can run on your own servers
Jeffronius Batra: I think he’s asking if he could run his own agent domain.
You: as can the region domain
Oliver Gassner: so if my comany has a open source SL running, some day
You: the question is mostly here one of trust
Troy McLuhan: Homework question: Compare this architecture to the peer-to-peer architecture proposed by Croquet
You: like you don’t want to allow any agent domain to connect to your regions if they don’t guarantee a certain level of e.g. identity verification
Oliver Gassner: and I have an avatar on a local machine (agent domain) Can I move ith this avatar (same dataset) in both the company SL and the ‘pulic’ sl?
Expert Market: if i host my agent server. what will it manage exactly ? my inventory ?
Tara5 Oh: nteresting comparison Try!
Oliver Gassner: @Tao: Sure, Identity will matter.
Tara5 Oh: iTroy I was doing my hoewrk last night!
You: Oliver: it depends on what the LL grid needs from you, e.g. you might not be able to rez something but can connect if you don’t have a certain trust level
You: whatever trust level means here
Tara5 Oh: what do you think will happen to DRM in p2p?
Expert Market: duh ?
Oliver Gassner: so we have what we in German call ‘avatarübergang’? not sutre f the english term.
Tara5 Oh: sorry homework
Kiwi Alfa: So, let me get this straight… if the viewer connects to one agent domain, does that mean they would be able to use the same agent throughout any region, regardless of who hosts it?
Kiwi Alfa: (Trying to work things out here)
You: Kiwi. if all regions allow agents from that domain to connect, yes
You: there probably will be more trusted sources than others
Kiwi Alfa: Okay.
You: esp. if people want to copy objects around there needs to be a lot of trust
Soft Linden: You may be getting ahead of Tao here. Tao, are you going to show slides with more detail on the agent domain, or may I link the wiki page for it?
Lem Skall: so Verisign could become a vendor of trusted agents?
Ian Betteridge: So, for example, you could define a trust level for a region which you run which says (ie) “If the agent domain doesn’t support ID verifcation, then agents from that domain cannot rez objects”
You: please link the wiki page, these are basically my slides
You: I wanted to get too technical but I guess I should 😉
Soft Linden: There is more information about the agent domain here. It will answer many of these questions:
https://wiki.secondlife.com/wiki/Agent_Domain
Kiwi Alfa: Soft: Sorry, it’s just that I didn’t quite understand what the whole agent/region domain thing was about. I get it now though.
You: some certificates will be most likely part of the trust system
Ian Betteridge grins… it’s all a tekkiewiki conspiracy I tell you! 😉
Expert Market: most of us are on sldev, go on techno-nerdo-geeky stuff 🙂
You: tekkiewikis you! 😉
Saijanai Kuhn: you need to have an index for all things AcrchiWG related
You: I give you another example of a possible scenario
Soft Linden: Saijanai: At the bottom, click the category.
You: look at the Company X box
You: you see an agent and a region domain there
Kiwi Alfa: Saijanai:
http://wiki.secondlife.com/wiki/Category:Architecture_Working_Group
You: this can be a closed grid of the company
Fleep Tuque: Akin to an intranet
You: the region domain only allows agents from their agent domain to connect
Lem Skall: Tao, you sure the slide changed?
You: and they can be sure that all those people in their agent domain are actually employees
You: it’s this slide
Fleep Tuque: Same slide, I believe
Kiwi Alfa: Lem: Tao didn’t change it. 🙂
You: company X in the right uppper area
Tillie Ariantho: But with open region, couldn’t someone just ‘copy’ an identity of someone else and then enter the corporate domain?
Tillie Ariantho: open regions
You: this is something we need to discuss, how to prevent it
Fleep Tuque: Not if the agent information is separate from the region information?
Jeffronius Batra: Tillie, the agent domain would have to trust the caller.
You: I think the whole identity area is sort of open right now
Expert Market: write a prokobot. it will prevent anything to exist.
Saijanai Kuhn: OpenID, sercurity protocols, etc
You: but the agent domain of course needs to make sure that only the right people can login there
Dale Glass: kerberos?
You: yes, it will be something with certificates etc.
Soft Linden: It is likely agent domains would implement their own security, and that region domains would verify with agent domains that an agent is properly authenticated and in a specific session.
You: and it should probably be pluggable
Saijanai Kuhn: maybe multi-level certificates
Kiwi Alfa: Probably the best method would be to use some sort of private/public key mechanism.
You: as said there might be different implementations
You: some domains might not want any security maybe
Oliver Gassner: a pki for virtual worlds
You: on their own risk then though 😉
You: and if we go on in this example
Expert Market: and host your ID in a USB key to be able to connect from a cyber
Jeffronius Batra: Great, then no one will ever be able to configure it or to log in. Instant perfect security.
Tillie Ariantho: Yes, sounds like encrypted tokens are required, that are only readable by the client and the main LL controler. and checks against that latter one each time with added information in the token again, that might be readable by certain region domains only…
You: it might be possible then that Linden Lab allows agents from the corporate agent domain to connect to their region domain
You: but not the other way round
You: the same might be true for objects
You: you might be able to rez “company objects” on company regions but not on Linden Lab regions
Oliver Gassner: what would be a time frame for this?
You: or any other than the company regions actually
You: well, I guess it will take at least a year
Tillie Ariantho: Sounds like 2+ years for me to implement this… WELL.
You: but Soft might know more 🙂
Lem Skall: Tao, doesn’t that sound like another domain for objects?
Soft Linden: I don’t know a time table on this.
Expert Market: the asset server problem still exist. Separating Agent and Region isn’t really helpfull, or i missed something
You: Lem: might be, one workitem from the meeting was to check out other ways of decomposition it
Scope Cleaver: More latency in the handshake processes?
Expert Market: or agent server are hpsting asset ?
You: Well, of course organizations with big grids will still have big databases
Free Radar HUD v1.1 by Crystal Gadgets
Kiwi Alfa: I imagine Tao will probably c over the asset server later. 🙂
You: but I am not sure how to get rid of that
Expert Market: Kiwi: it’s 9/10 slide, so i don’t think so
You: well, it’s at least separated into a region inventory and an agent inventory
You: and these can be separate databases
You: but does somebody have an idea of how to get rid of the scalability issues on assets?
Boroondas Gupte: And those can be split by agent and by region
Soft Linden: (The sldev mailing list and Zero Linden’s office hours are the two best places for asking questions of LL on this work.)
You: one way might be to store them on the client but I don’t see this as very practical
You: at least not in my setup
You: with X computers
Expert Market: Unless you want to go 10 years back in the past (Active World), no. i don’t know
You: so here are the ways how you can participate
Tillie Ariantho: Assets are just information in a database, who does Google does that? They have thousands of servers and it works well, all the time.
Tillie Ariantho: maybe ask the Google people. .)
You: I don’t think the DB size is the actual problem
Tillie Ariantho: And I am quite sure the Search feature will be sponsored, soon. .P
Lem Skall: how will be meetings announced?
Saijanai Kuhn: assets aren’t even on a database. I think the index is, but assets are files on a file server, or such is my understanding
Expert Market: But… i’d LOVE to have some kind of “Second Life proxy”. So i can host on a dedicated server, most of the texture and object i usually see ingame.
You: actually with the agent domain you might also be able to create several agent domains for one organization and thus have more databases
Expert Market: i setup a 1TB quadcore server and *hop* less asset lag
You: and you could have region domains or agent domains for different countries etc.
Expert Market: like a dns cache, but for asset
Saijanai Kuhn: my understanding is that most lag is caused by a bottleneck with the sim, not the asset server itself
Expert Market: or like a http proxy
You: Sai: might be, but LL should know more
You: but Zero once explained (or tried to) why not more people can be on one sim
Saijanai Kuhn: just a vague recollection of what zero has said
You: and it somehow had to do how they do DB access IIRC
Saijanai Kuhn: the sim does it all, basically
You: not network load
Tillie Ariantho: Yeah, and filters and money transaction checks by country too. I see it coming that you have to connect to your countries login server soon.
You: well, LL seems to plan locally hosted regions anyway
You: although I am not sure how that helps me if I don#t know where which region is and I am jumping around anyway 😉
Expert Market: Tao: not network load, cpu load. when you have everything grey it mean that the DB is feeling gloomy.
You: at least you might be lucky and it might be faster sometimes
You: Expert: I think we can only guess here
Lem Skall: how soon object backup? I thought that was going to be soon and isn’t that connected to the new arch?
You: but actually Linden Lab is part of the project and I guess they will put some knowlegde into it
You: no idea about backup
You: it’s mayby also not really part of the architecture
You: more a client side function I’d say
Expert Market: it’s hoster responsability. backup is a detail
You: if you mean backup for yourself
Lem Skall: I can guess it’s not part of new arch but then it looks like it’s going to be implemented twice
Tara5 Oh: Tao what are the consequencess for DRM in this architecture?
Expert Market: DRM is bad
Peter Newell: as I understand it, they’re avoiding DRM
You: well, I think there was some concensus that DRM should not be central part of the architecture
You: but of course you should still be able to assign permissions etc.
You: and one needs to make sure that these are not broken, thus the trust system we need to have in place
Fleep Torok: DRM is a disaster in every other implementation, I hope they continue to avoid it here too.
Tillie Ariantho: heh… there will be a DRM popup “Pay 20L to hear this avatar say YeahYeah!”. .P
Dale Glass: Tara5: What exactly do you mean by DRM?
Boroondas Gupte: You shouldn’t rez things in regions who’s domain you don’t trust to keep to the permission system, easy as that.
Expert Market: DRM is an unefficient blackbox solution for legal problems
You: I don’t think any copy protection is going to work anyway
Troy McLuhan: DRM and permissions are just crude ways to enforce licenses. The key is to have attached and discoverable licenses
Tara5 Oh: digital rights management
You: DRM just means bad user experience for end users
You: for those who actually buy stuff
You: not for them who illegally copy stuff
Dale Glass: I know what it means, I’m asking what are you referring to in SL. Permissions?
You: they don’t have to deal with it
Ian Betteridge: But permissions = a DRM system.
Expert Market: no
Tara5 Oh: and how would you see this licensing being agreed apon in time?
Expert Market: permission exist since filesystem exists
Expert Market: and we weren’t talking about “DRM”
Tara5 Oh: well SL permissions and more IP in general in an open metavers
Troy McLuhan: Permissions can’t anticipate all possible licensing scenarios
Tara5 Oh: and how people envisage that evloving
You: as said, the protocol should of course support permissions
Ian Betteridge: OK, so let’s be really clear on this: will this architecture preserve the existing system of permissions, which content creators rely on?
Kiwi Alfa: The difference between filesystem permissions and SL permissions is that filesystem permissions don’t let you change the owner of the file yet give them restricted permissions. ;p
Tara5 Oh: the system of microsystem and permission in SL has enable many things to happen
You: and IMHO it should also be possile to attach a license to an object, e.g. a CC license
Tillie Ariantho: Ian: I dont think that will work with open source servers.
Expert Market: why not ?
Ian Betteridge: Tao: I’ve heard that’s an issue, as you’re not allowed under CC licensing to enforce the license using software.
You: so what can happen is actually that object permission get ignored on either the agent domain or region domain
You: by “bad people”
Lem Skall: hmm, this sounds like a topic for sldev, but difficult to discuss here
You: but this is why not everybody can simply connect to everybody else
Tara5 Oh: as far has i can see the biggest issue is not creating open architecture but rethinking IP
Tara5 Oh: not just creating open srchitecture
Expert Market: open architecture is not a solution for IP but for scalibity
You: there needs to be a trust system in place and if you choose an agent provider that provider needs to make sure to take care of your IP in e.g. checking who is allowed to connect
Lem Skall: a good architecture should enable better IP later, but this should be kept in mind
You: LL has this in mind
You: as they have a world to migrate to this new architecture
Tillie Ariantho: Yes, trust system is not enough for that.
Dale Glass: I talked to people a bit, content creators find the idea of losing permissions scary. I heard a proposal to make it possible to limit assets to a trusted domain. Say, avatars that will not be transferred to non-LL sims.
Expert Market: Tillie: it work for the good old web
Dale Glass: Excepting for maybe scripts I don’t really see how that’d work though
You: something like this might be possible, Dale
Tillie Ariantho: What works there? Everything is copied and stolen on the good old web.
Boroondas Gupte: I wrote down some ideas about assets here:

Click to access persistent_inventory.pdf

(this was written before the ArchWG, or I would have used the term ‘Agent Domain’)
Expert Market: when you host a website, your hoster isn’t stealing your content. when you subscribe to a forum they are not publishing your password, etc
Fleep Torok: Thanks Boroondas
Saijanai Kuhn: or content that is left behind. ONe issue that I mentioned is multiple avatar settings for multiple levels of turst, so you don’t have to stop and put clothes on if you leave the most trusted sims
Kiwi Alfa: My feeling is that people running businesses on SL would continue to do so on LL servers, unless they hosted their own.
Tara5 Oh: and what doea it mean for a community lik SL that has evolved a certain kind of virtual economy to go to open architecture
Tillie Ariantho: Yes, but I dont switch the hoster each hour.
Lem Skall: Tao, what kind of limitations are there on the new architecture in order to allow porting of the old one? for instance, the meaning of region/sim
You: actually the region domain will consists of 3 components
You: one of them will be sort of the simulator as we know it
You: but with different interfaces of course
Ciemaar Flintoff: The SL economy is going to get a major change in this, that we can figure, but the upside is actually quite nice
Tillie Ariantho: What if I create an item and sell it. Someone buys it and is on a server in a ‘not trusted’ domain. The person running the server can grab all the stuff and use it himself. No? As the script are transferred to his server to be run there etc….?
You: the other ones are the DB and one is about stateless information such as region name etc.
Ciemaar Flintoff: content creators will be able to charge a premium for “portable assets”
You: Ciemaar: that was my idea, too
You: content creators could set a flag “allow to take with you from this domain” and might set a higher price
Lem Skall: Tao: so what else besides sims have to be “backwards compatible”?
Expert Market: or include it as a one time fee. like copy vs nocopy item
Ciemaar Flintoff: “portable assets” will have the ability to move more freely and thus cost more, hopefully enough to afford standard legal processes if they are copied
You: Lem: this depends on how the new architecture will look like in detail I think
Saijanai Kuhn: just as avatar attachments/clothing might need multiple definitions, same with your huds and assets you try to rez
You: we don’t know the existing server details anyway
You: and it’s all in rought state right now
Saijanai Kuhn: your inventory might not work as expected if you go into the wild west
Dale Glass: I sort of like that idea from the POV that an economy is a good thing to have — people who wouldn’t have made stuff otherwise are motivated to make something.
Ciemaar Flintoff: remember the ultimate legal basis for content distribution is copyright, a legal, not a technical limitation
Expert Market: any rough ETA for that stuff ? 2y ? 6m ? 5y ?
Dale Glass: The problem is, how do you enforce it without going into real DRM of the user’s computer?
Ian Betteridge: Would region/agent domain owners be able to turn off specific aspects of LSL?
Tillie Ariantho: So if you wear ‘secure domain’ prim clothing only and zone onto an unsecure server, you are stark naked then? .)
You: Expert: no ida, probable >1y
Lem Skall: Tao: then we DO need to know the present architecture
Dale Glass: SL provides about everything needed to copy about anything in-world, excepting scripts
You: ok, so a little bit on how this is supposed to work
Expert Market: tao: no word about licencing and cost yet ? i suppose ?
Saijanai Kuhn: its enforced by the asset server refusing to acknowledge that you own the right to wear/use some item in the new sim
Fleep Torok: I find it disturbing all this focus on DRM, IP, content sellers.. The point is a flexible, cross-platform architecture that scales, yes? And the issues of economy and IP will vary by region/country/RL geographical law.. separate from the grid architecture.
You: apparently participation is needed by as many people as possible
Boroondas Gupte: brb
Ciemaar Flintoff: Tillie, I tihnk we can assume that clothing will be handled, either with alternatives or a process like ruthing
You: so we might need different use case scnearios which should be added to the wiki
You: also those issues about IP, DRM etc.
You: and there is the SLDev-List for discussion
Saijanai Kuhn: those issues are what allows SL to exist in the first place though
You: and Zero’s office hours
Dale Glass: Well, I don’t have a personal interest in that part, but the thing is there are people who do, and they want to hear something 🙂 Otherwise we’ll get lots of people freaking out
Soft Linden: Some things like content protection are unsolved problems. sldev is a very good place to float proposals on things that are still abstract.
Ciemaar Flintoff: agreed Fleep, but I thought part of the purpose of this meeting was to engage those concerned on that level
Peter Newell: no matter what we’ll get lots of people freaking out
Peter Newell: it’s just how it goes
You: and for now I guess it’s brainstorming phase, at some point though we need to define use cases more precisely and esp. components and interfaces
Expert Market: the problem is when you talk about open architecture, people think “copybot” and “it will hurt my very own 10US$/month business. Let’s whine on the forum”
You: and in the end everybody is invited to code stuff around this protocol
You: so we end up with a lot of (evtl. open source) components which hopefully interoperate
Lem Skall: Tao: how will be meetings announced?
You: the idea is to have as many people code components as possible to make sure, interoperability is given
You: Lem: no idea actually 😉
Dale Glass: Well, not all of them are $10/month, Luskwood pays for about 3 sims by selling avatars. It’s a quite significant bit of cash
Ian Betteridge looks for his coding ha, realises he sold it in 1988 🙂
Tara5 Oh: I think the issue is that one of the factors in SL’s specatular growth is the way it has rewarded creativity and user generated content
Expert Market still want his own asset proxy cache. *grrr*
You: and one component might actually be OpenSim
Lem Skall: good attendance today but announcement was not very public
Peter Newell must run, thanks for the chat.
Lem Skall: had trouble finding it again today
Tillie Ariantho: Sure people will whine, some live on that money from in here. .) And land barons will whine too, I think, as prices will drop…
You: Lem: Well, it was my little event here so I don#t have the power to post to the LL blog 😉
Savannah Glimmer: Aside from the Dev List, is there a group to join specifically for this?
Tara5 Oh: and that this is obviously the only way to do that but it is something worth thinking about
You: we might need to form a group maybe for announcements
Expert Market: Tillie : i’m a landbaron and make a living of it. and not whining 🙂
Tillie Ariantho: Expert: .. yet. .D
Fleep Torok: Egads not another group. 🙂
Soft Linden: You can subscribe to sldev and filter it all to the trash, saving only messages with [ann] in the subject to only get announcements. There are other tags as well.
You: actually as the timeframe is >1y I think there is also lots of time to adjust your business if it really needs adjustment
Savannah Glimmer: Thanks Soft.
Lem Skall: so it will be sldev [ANN}
Goldie Katsu: Where do I subscribe to sldev?
You: actually Adam Zaius is apparently also a land baron and seems to be very active in making SL alternatives 😉
Expert Market: my business is clear : i’m a unix system engineer used to work on cluster and i REALLY WANT to host my servers 😀
Tillie Ariantho: =D
You: the good thing with the regions and the open protocol is that you really can put behind it whatever you want
Lem Skall: ok, let’s clarify, ONE landbarron with land on the mainland
You: as long as you implement the interface right
Soft Linden: Goldie: sldev and other resources are linked here:
https://wiki.secondlife.com/wiki/Developer_communication_tools
Kiwi Alfa: Goldie:
https://lists.secondlife.com/cgi-bin/mailman/listinfo/sldev
Kiwi Alfa: Well, that too.
Ciemaar Flintoff: everyone will adapt or leave, we can figure that now, content creators will adapt to higher margins on portable assets, land barons will adapt to higher margins when they host their own servers, it’s just the way of it
Expert Market: bleh. mainland is braindead, who care ?
Goldie Katsu: Thank you Soft & Kiwi
Tillie Ariantho: heh
You: as for land prices I don’t think they will drop that soon anyway, hosting a server with good connection still costs somewhat
Dale Glass: oh yeah, did LL mention how will people connect to the grid when this is done? I mean, I setup a box and then what, pay $X/month for a connection to the grid?
You:
http://wiki.secondlife.com/wiki/SLDev
Expert Market: and LL will probably charge licencing fee to connect to main grid
You: Dale: nope, no word about that
Goldie Katsu: My interest is I’ma security architect with OS devel background and I’d like to see SL continue to succeed (even if I can’t spell this morning/afternoon)
You: maybe they don’t know yet themselves how their business model will look then but I guess they have thought about that before they created that group
Expert Market: hopefuly
Ciemaar Flintoff: and if the architecture improves properly the LL grid should be the best part of it for 6months atleast, hopefully they can continue to be the leading host provider
Tillie Ariantho: Just to make it clear: I am excitied about the ‘new grid’. Just want to point out … ‘things’. .)
Expert Market: but with LL, we never know
You: well, if not, we at least can implement an alternative ourselves
Kiwi Alfa: Dale: I imagine that different agennt/region subgrids will ask you to pay for them separately. The whole thing overall probably should be free, but the LL grid, for example, would still cost.
Kiwi Alfa: I suspect.
Expert Market: Your wold, your imagination, our asset server
Saijanai Kuhn: Zero indicated that they are looking forresumes from ‘ good secruity experts
You: Tillie: there are many issues of course and it’s important to think about them
Lem Skall: what interest is there in this new architecture from corps like IBM?
You: there is a brainstorming section in the wiki, feel free to add to that
You:
http://wiki.secondlife.com/wiki/Brainstorming
Fleep Torok: I assume the same as universities and researchers – want to maintain their own assets, backups, etc.
Tillie Ariantho: I’ll make up a list if things are not on there yet. 🙂
Saijanai Kuhn: Zha Ewry was invited and she appeared to like the concepts
You: well, at least one person from IBM was present at the first meeting
You: yes, Zha
Fleep Torok: Our scientists won’t work seriously on the main grid because they can’t control the security and continuity of their research.
Saijanai Kuhn: “really neat” or words to that effect
You: and for folks like IBM and Intel it might be interesting to build their own region servers or clustrers maybe
You: and for companies it’s important to be able to host their own servers
Saijanai Kuhn: but there will be no more “main grid” in any current sense of the word
You: as for confident information etc.
You: main grid might be the LL grid for a while
Lem Skall: yes, my question is whether IBM will be involved in the Working Group
You: Zha is involved it seems
Lem Skall: and how involved
Expert Market: could be nice
Lem Skall: one person is good
Tara5 Oh: not fleep but some of the most inyeresting science will evolve with the wisdom of crowds
You: it’s only 1 week old now so we need to see how involved who is 😉
Expert Market: IBM staff aren’t cheap startup
Saijanai Kuhn: Zha indicated that the invitation to future meetings was based on contributions to the previous meetings, and that she, naturally, was poking her nose into everything…
Fleep Torok nods at Tara5.
You: well, Zero mentioned that it might depend on what you contributed to the project to be invited
You: and it should not only mean technical stuff but also ideas, discussion etc. I’d say
Saijanai Kuhn: and Zha pointed out that she was speaking up as much as possible in the first meeting (hint hint)
You: it was not about the first meeting I think
Fleep Torok: That’s true indeed, but certain controlled environments and private regions for research are the #1 issue for some of the constituent groups on my campus.
You: I did not speak up that much in it mainly because I was inworld 😉
Expert Market: Tao: actually it’s too much focused on “how many patch y our wrote, imho”
You: and it wasn’t that easy to see what was going on in SF and when#s a good spot to interrupt 😉
Saijanai Kuhn: sure, but I think her point was that she really wants to be involved
Pham Neutra: sorry, have to leave too quick 😦 thanks, Tao
You: thanks for coming, Pham 🙂
You: well, I hope it’s not just about how many patches you did
You: actually right now there is not much to patch 😉
Expert Market: i still expect to be usefull as sysadmin… one day, who know ?
Soft Linden: There is discussion about getting better streaming from the SF office somewhere near a large screen so RL LL to in-world bridging for meetings shoudl be possible in the future.
Fleep Torok: That would be nice, Soft
Saijanai Kuhn: do you contribute to the jira? Do you tweak the wiki? Do you attend office hours? Do you heard cats on sldev?
Saijanai Kuhn: lots of ways of being involved besides patches
You: Soft: it would actually been better to a) see folks in SF and if everybody would have a headset and there wouldn’t be any feedback problem
You: there is also nothing set into stone yet
Expert Market: sai: i do. but not with that name 🙂
You: so the first meeting was really just a random group, no idea how selected 😉
Saijanai Kuhn: I think prominent people in the community whom they believed would be valueable contributors
Saijanai Kuhn: a subset thereof
Tillie Ariantho: I’d like to participate on a regular basis. 🙂
Soft Linden: Not random. Most of these people were engaged in significant projects, or contributed to related discussions in Zero’s office hours, as I look at the list.
You: Soft: ok, I also wouldn’t say it was cimpletely random 😉
Goldie Katsu: I have a question on the current open source project – which relates to the grid project.
You: you mean openSim?
Goldie Katsu: No the client.
You: or the viewer?
You: ah
Saijanai Kuhn lol
Dale Glass: oh yeah, the feedback thing was annoying
You: that#s why I never said something, because of the feedback 😉
You: so I need toi make these meetings now 😉
Goldie Katsu: How are patches being integrated and are they beign folded into the testing harness that tests the client pre-rollout.
You: I guess that’s more a question for Soft
Saijanai Kuhn: that’s a bug triage/user interface triage/feature thing
Kiwi Alfa: For what it’s worth, I’ve been meaning for a long time to attend office hours and Bug Triages, etc, but haven’t been able to yet.
Saijanai Kuhn: 2-3 meetings a week on that
You: oh, I also would like to invite everybody to spread information about this project in your community
Soft Linden: That’s kind of a big unrelated discussion. sldev is a good place for that kind of question, too.
You: I think we also need people outside SL to take part in it
Goldie Katsu: no the testing part is what is important
You: and LL maybe should attend some open source conferences to present their stuff
Expert Market: i think i should start to write report in french for the french community… i wish i had more time
Saijanai Kuhn: ask the SL network people to cover the open office hours, or host meetings like this one with auido webbroadcast
Goldie Katsu: The open source integration – and how it is handled has significant impact – and more so when you are talking about parts of the grid.
Saijanai Kuhn: I think they’re holding f on really heavy presentations ountil the various houses are in order.
Fleep Torok: SLCN.tv would likely do it.
You: like it might be nice to carry this over to the Python community esp. as LL seems to do more and more with Python
You: people might be interested
Saijanai Kuhn: The GUI is a total mess right now, for example
Tillie Ariantho: ok
Saijanai Kuhn: SLNC.TV wasn’t impressed with the idea. They like virtual reality hocky more
You: well, this projects starts sort of from 0, so there is much people can do
Boroondas Gupte: Apropos “outside SL”, will the new architecture help different online VR systems to converge to something compatible? Like interfaces between SL and croquet?
Saijanai Kuhn: the propposal is one-size fits all, which I thought was wrong at first
You: Boroondas; I don’t know much about croquet but I hope there can be some sort of compatibility layer
Saijanai Kuhn: every sim is like every other is like the rectanglar Linden Sims
You: I think e.g. we should make sure that region formats are not fixed as well as object formats
You: like not 256×256
Expert Market must be afk 3mn, sorry
Saijanai Kuhn: But I realized that this is the best way to go
You: and maybe a sphere
You: not that we should implement this right now but provide some field for it or some extension
Saijanai Kuhn: if you want to talk to the main grid directly, you expose a main-grid compatible gateway to your specialized grid
You: I wrote down some different scenarios in the wiki
Saijanai Kuhn: the portals or whatever are on YOUR sims, not the LInden mainland
You: like an EVE setup, a Google Earth setup etc.
You: of course you might need a different viewer eventually
Boroondas Gupte: sure
You: but it might also be nice to be able to implement online games on top of it so that you at least can reuse your identity
Saijanai Kuhn: but they shold all be approachable to a plain-vanilla sim via the plain vanilla interface
Soft Linden: I should mention – in addition to sldev, many Lindens hold office hours, for all these mentions of UI, networking, etc:
https://wiki.secondlife.com/wiki/Office_Hours
You: and your friendslist
Tillie Ariantho: I am missing some details about which linden is doing what in that list.
Saijanai Kuhn: that way, the inerface says as clean as possible and you don’t need specialized viewers to just wander by a random maingrid sim
You: and beside the content creators I would also like some people to think of opportunities for simple end-users
You: all those people are important
Frans Charming: there is also a google calender of office hours.
Ian Betteridge: Soft, can you get someone to update the Google Cal with the office hours on it? 🙂
Boroondas Gupte: like your browser has to support ftp (or start an ftp client), if you want ftp links to work from a html webpage, you’ll need a viewer that either supports different types of VR or can start other viewers
You: the Google calendar is unfortunately not uptodate
Soft Linden: Tillie: If you have questions about specific ones, asking on their user talk: page would be one way to get details. Another is to ask on sldev, as chances are the person you ask about will read it.
Tillie Ariantho: okay, thanks. 🙂
Frans Charming: ack. 😦
You: but I agree, they should add their topic on the wiki page with the office hours 😉
Tillie Ariantho: or in the calendar. .)
Saijanai Kuhn: right. Plugins or other viewers are fine, but the main grid setup shold be as simple and clean as possible (256×256 or some other figure like that)
You: as said, I’d like to have this open.. make it some field like “regionformat” with a label and a version number or so
Saijanai Kuhn: I’d rather see that as a plug-in deal.
You: btw, some things are also fixed already, like using LLSD, implementing most of it via REST and using chttp
You: but if it’s a plugin then you still need to be able to define it in the protocol
Tillie Ariantho: region_type = sculptie; .)
You: so that you know to which region type you can connect
Saijanai Kuhn: sure, but you don’t have to support anything but the definition, not the actual protocal
You: sure
Fleep Torok: Tao, will you post your slides on Slideshare or the equivalent?
You: there should be some fields depending on the region type or some container which per default contains the normal SL region stuff
You: I can do that, Fleep 🙂
Expert Market is back
Saijanai Kuhn: so the main grid is extensible, but it doesn’t have to support YOUR portal on every island in teh universe. Only on the ones that you’ve made gateways to your MMORPG at the “edge” of the universe
You: the question how location is handled is also somewhat open I thinnk
Saijanai Kuhn: and YOU or your firends, own those portal grids
You: where to place regions on a map
You: and how interconnections of region domains will look like
Saijanai Kuhn: the map they proposed is 64 bit I think
Expert Market imagine a huge gateway : Stairway to World of Warcraft
Lem Skall: I hate the map
Saijanai Kuhn: that’s a lot of 256 x 256 squares with .01 resolution
You: well, there are many things I’d like to have, being able to have optiobnal portals, being able to stack sims on top of each other, having a region which is more like a sphere, like planet etc.
Dale Glass: right, and I’m sure Blizzard will be really glad to publish the protocol specs 😉
Troy McLuhan is fascinated by how Croquet manages the idea of “location”. You could make contiguous regions, but then there is this idea of portals, sort of like permanent teleport gateways
Miya Watanabe: I’m taking off. Thank you for the meeting, Tao, and everyone.
You: thanks for coming, Miya 🙂
Saijanai Kuhn: Blizzard could host their own gateway grid
Saijanai Kuhn: sim
You: it’s maybe not very likely that Blizzard wants to interconnect 😉
Saijanai Kuhn: but someone might
Tillie Ariantho: They would charge you for entering. .)
Ciemaar Flintoff: dunno Tao, entrance sims could be very compelling
You: I hope so
Soft Linden: Thanks for holding this, Tao! gtg
You: thanks for coming, Soft 🙂
Expert Market: A sperical sim (like a planet) could be a lot of fun too 🙂
Tillie Ariantho: bye Mr Linden 🙂
You: so maybe we should do such meetings more often
Ciemaar Flintoff: replace the chat rooms waiting for games to start etc.
Saijanai Kuhn: take care soft
Dale Glass: later 🙂
Saijanai Kuhn: exactly
Saijanai Kuhn: I wanted to do that with Croquet, but croquet is LAN-only right now without a LOT of work
You: it would also be nice to get people from e.g. Multiverse to participate
You: metaplace would be nice, too, but it seems that Raph Koster has some problem with LL 😉
Expert Market: a SL client using crytek engine coule be nice too 😀
Saijanai Kuhn: LL is on a mission. Money is secondary in a sense.
Tillie Ariantho: Definitely 🙂
Tara5 Oh: croquet is actually fairly easy to put on a WAN
You: yep, it seems really a mission
Expert Market: crytek is definitivelyu the kind of engine made for openspace like SL
You: but it’s all about surviving, too
Saijanai Kuhn: other groups are on missions that may or may not be compatible with LL’s
Lem Skall: Tao, thanks for holding the meeting
You: they mostly have their business missions and they mostly mean “walled gardens” unfortunately
Fleep Torok: I think the marketplace for virtual worlds is so undefined, there is a lot of incentive to keep your dev to yourself and hope to be THE emergent player.
Tara5 Oh: and i think considering whjere the open architecture grid (server based) fits with this p2p No server model would be a very good thing to consider!
Saijanai Kuhn: the nice thing about the LL view is that it DOES allow for other players
You: Tara: this might be a question of how to implement regions maybe..
You: you might also consider taking regions to sleep when there is nobody on them
Saijanai Kuhn: p2p doesn’t scale well, and as people have pointed out, even Croquet isn’t true P2P
Expert Market: P2P won’t work
Expert Market: huge overhead
You: thus having a better resource usage
Tara5 Oh: yes i agree but multicore processing could change that
Expert Market: and P2P is all but reliable
Saijanai Kuhn: and the roadmap for Crquet calls for work on central asset servers, etc, anyway
You: it’s maybe also about bandwidth
Tillie Ariantho: I don’t like P2P anyway. :p All I see it is used for is breaking laws.
Tara5 Oh: and an mixture of server based open archotecture and p2p may well be the way things go?
Expert Market: prfff
Fleep Torok: ??? P2P is used for much more than breaking laws. 😛
You: Tara5: the good thing here might be that people can experiment with solutions like that
Tillie Ariantho: yes, breaking more laws.
Tara5 Oh: well i doubt if there will be much DRM in p2p!!!!
Dale Glass: people could break laws just fine with plain FTP
Saijanai Kuhn: I still see a use for Croquet as a private sandbox for testing/playing purposes, but you have to keep in mind the univseral asset nature of Croquet when you write the interface
Tao Takashi does not buy DRMd products anyway
Expert Market: P2P is a technology, not a “way of life”
Tara5 Oh: but inetl and croquet have started a small collaboration i saw in the “news”
You: but I also do not copy them illegally, I want to add 🙂
Fleep Torok: nod Tara5, I posted about it on my blog, they gave a brief demo –
http://fleep.wordpress.com
Saijanai Kuhn: Rray tracking? The Squeak/groquet graphics guy is a real go-getter
Tara5 Oh: i don’t think croquet will necessarily remain private sandbox
Saijanai Kuhn: ray tracing?
Fleep Torok: Intel and Qwaq that is, not Croquet
Saijanai Kuhn: until they get scaling via central servers, it can only be a small group thing
You: I hope at some point somebody can demo croquet to me 😉 I once tried it but I never got the interface 😉
Tara5 Oh: aQwak is Croquet
Saijanai Kuhn: its old school smalltlak
Expert Market: /mr frown
Tara5 Oh: they just didn’t call it that in the blogs
Saijanai Kuhn: aQwak is still LAN based as far as I know
Tara5 Oh: Nope qwa is WAN
Tara5 Oh: had the croquet folks hep me get out of the firewall here once
Saijanai Kuhn: getting two macs or two pcs to talk to each other over the internet is hard. Getting a mac AND a PC to talk to each other is almost impossible
Tara5 Oh: not with croquet
Saijanai Kuhn: You gotta go through their KAT open server
Expert Market: i hope LL keep this in mind : Don’t forget to implement IPv6 with open architecture
Tara5 Oh: it is platform blind
Tara5 Oh: that is the whole point
Saijanai Kuhn: I’ve been playing with Croquet for months. It is quite platoform specific
Tara5 Oh: absolutely not in design
Saijanai Kuhn: you had to get help from a commercial group to get it to work. Scale that to 100,000 newbie users
Tara5 Oh: that is the heart of what they want to do why they chose smalltalk etc
Saijanai Kuhn: they say specifically that they haven’t ironed out the issues with WAN yet
Saijanai Kuhn: I never said it was designed that way–only tha tthat is the current reality
Gareth Ellison: hi there
Tara5 Oh: they have no applications released yet’ it is only available for developers
You: Hi gareth
Saijanai Kuhn: right.
Tara5 Oh: spokthey demo it on WANs now
Boroondas Gupte: yeah, croquet isn’t end-user ready, yet
Gareth Ellison: sorry Fleep
Saijanai Kuhn: Qwak has private stuff for business clients, but that stuff hasn’t been folded into the opensource version yet
Fleep Torok: Hehe no prob
Tara5 Oh: it is far from user ready
Gareth Ellison: ooh, Croquet
Gareth Ellison: interesting system
Malburns Writer: I think Lendens also need to consider in all this that this type of environment is probably be going to be the future of the desktop – it needs to be cross-platform and something the end-user can easily configure foir use with local files
Tara5 Oh: but that doesn’t mean one shouldn’t think about what this kind of architecture may mean in the furture
Boroondas Gupte: sure
You: Malburns: Maybe the viewer can also act as agent domain
Tara5 Oh: foolish not too
Expert Market: don’t think so
Saijanai Kuhn: Hey I’m a bnig fan of croquet. I tried to patch the SL client to send SL chat to croquet a few months back
Gareth Ellison: have people here been reading the wiki? was told that this was a less technical discussion
Saijanai Kuhn: but its hard to test when you have to spend hours workign with someone to connect over the internet
You: it turned out to get somewhat technical nevertheless 😉
Ciemaar Flintoff: it was Gareth, but it’s not anymore…
Gareth Ellison: one idea i proposed there is turning the core of the viewer into a lightweight library
Gareth Ellison: with a scripting language
Tara5 Oh: yes get them on the phone and they will give better instructions
Malburns Writer: yes – a lot of the tech is above me but i catch the drift – it is the future of the interface itself that concerns me most – an offline privatre system that connects to various grids
Saijanai Kuhn: its actually on the user interface roadmap.
Expert Market: libsecondlife
Gareth Ellison: that would make it possible to port very easily
You: Gareth: Steve Linden also mentioned such an idea
You: yes, on the Viewer Roadmap
Gareth Ellison: libsl linden style
Tillie Ariantho: I’d like to have a python scripting interface in the client. 🙂
You: libsl without .NET probably
Saijanai Kuhn: it will take time. First they need to normalize the GUI so they CAN separate it into a library
Fleep Torok: For the latecomers and transcripts, here is a list of all the links from this discussion:
Gareth Ellison: Tillie: funny, that’s what i was thinking of adding
You: I’d like to have a python interface, too 😉
Fleep Torok:
https://wiki.secondlife.com/wiki/Agent_Domain
http://boroon.dasgupta.ch/persistent_inventory/persistent_inventory.pdf
https://wiki.secondlife.com/wiki/Developer_communication_tools
https://lists.secondlife.com/cgi-bin/mailman/listinfo/sldev
http://wiki.secondlife.com/wiki/SLDev
https://wiki.secondlife.com/wiki/Office_Hours

Thanks to all the attendees!

tags technorati : slga secondlife architecture lindenlab taotakashi meetup talk