[2007/11/20 9:14] Burhop Piccard: Besides teh AWG stuff, my company has some stuff going on with a university.
[2007/11/20 9:15] Burhop Piccard: Strife, is that your sign?
[2007/11/20 9:15] Morgaine Dinova: It's going to be very interesting how we end up handling objects extensibly ... VWs are going to be so object and media rich.
[2007/11/20 9:15] Saijanai Kuhn: Finally got the login python code working. We cangrab the same stuff the client does, and move to the next level and extract any of hte standard capabiltiies from the "seed" cap
[2007/11/20 9:16] Burhop Piccard: Morgaine - yes. Interesting part is to talk to people outside SL. They have lots of reasons to come here with with a few improvements.
[2007/11/20 9:16] Saijanai Kuhn: Donovan spent over 2 horus working on the code. First hour he tried to help me, then he just grabbed it and found the bug.
[2007/11/20 9:17] FWord Utorid: i think i will be looking to do login with c code soon
[2007/11/20 9:18] Strife Onizuka: most new users have ideas on how to improve SL i have found
[2007/11/20 9:18] Burhop Piccard: Strife -yes... I think since we are used to SL, we sometimes loose sight of this (at least a little bit)
[2007/11/20 9:18] Morgaine Dinova: Strife: pity that av's aren't clickable. I think every object should have the ability to react like prims do, including av ones.
[2007/11/20 9:20] Burhop Piccard: So... I'm on vacation this week... Thinking of getting the viewer code. Any words of advice before I start (and will it work on my Vista Laptop with VS2005?)
[2007/11/20 9:20] Burhop Piccard: Thinking about running a Linux VM too.
[2007/11/20 9:21] Cel Edman: Hi there, client crashed landing here
[2007/11/20 9:21] Saijanai Kuhn: you need 2003 VS I think. You should join SLDEV mailing list and ask that kind of thing there. Lots of people working on getting later compilers to work
[2007/11/20 9:22] Vincent Nacon: 2005 can convert the save project file from older version I believe
[2007/11/20 9:22] Saijanai Kuhn: from what little I have read, there are issues, though
[2007/11/20 9:24] Burhop Piccard: Fword, we should shop together. Nice Suit :-)
[2007/11/20 9:24] Saijanai Kuhn: in the long run, yes. What we're working on now is a "test harness" to test the new communications stuff that will be used in the meta-grid 2 years from now
[2007/11/20 9:28] Saijanai Kuhn: Zero designed the Het-grid to handle different kinds of Sims. I iimagine that he'll create a special "side" login server that will let you login to the beta grid and later, teh main grid, using the new procedures
[2007/11/20 9:29] Saijanai Kuhn: when Zha gets here, you can ask her. She's been working hard on this stuff
[2007/11/20 9:29] Tao Takashi: just FYI: I finally managed to upload the chat script from last time
[2007/11/20 9:29] Tao Takashi: I doubt that Het-Grid can handle these new sort of sims
[2007/11/20 9:29] Saijanai Kuhn: the sims won't change overnight
[2007/11/20 9:30] Vincent Nacon: may have to trottle the server on lower power/speed to test the traffic since having bunch of users on their own net than having bunch of users on the same net... which isn't a true test to start with.
[2007/11/20 9:32] Saijanai Kuhn: has images of her fumbling with massive charts for today's presentation...
[2007/11/20 9:32] Ahzzmandius Werribee: Has anyone build a system for managing hundreds of clients to test these networks with?
[2007/11/20 9:32] Dr Scofield: let me check whether i can see her internally
[2007/11/20 9:33] Dr Scofield: i'm working on a bot system that could do that
[2007/11/20 9:33] FWord Utorid: why would you want hundreds of clients? that's crazy!
[2007/11/20 9:33] Morgaine Dinova: Ahzz: the libsl folks might have something like that. We're also building a test tool for that ourselves.
[2007/11/20 9:33] Dr Scofield: but any libsecondlife client could do that
[2007/11/20 9:33] Dr Scofield: nope, no zha on sametime either
[2007/11/20 9:33] Vincent Nacon: my problem with the testing is not just the number of users but network load
[2007/11/20 9:33] Ahzzmandius Werribee: heh, well, i'm working on a "grid" of sorts that can manage thousands across many machines.
[2007/11/20 9:34] Saijanai Kuhn: one of the usecases for a specailized client and sim is the concert-TV client, where only an inivisble camera exists
[2007/11/20 9:34] Morgaine Dinova: We need to load it up with thousands of clients. Scalability analysis can go only so far. You have to see the descaling trends in practice.
[2007/11/20 9:34] Dr Scofield: saij, you've been looking into the login stuff?
[2007/11/20 9:34] Saijanai Kuhn: in theory you could have thousands or even millions of "avatars" present to watch the same lecture
[2007/11/20 9:34] FWord Utorid: sai, that is a very interesting premise
[2007/11/20 9:35] FWord Utorid: it reminds me of the admin function of a quake server
[2007/11/20 9:35] Saijanai Kuhn: yeah. Took the libsl python script and got it to step 2: grabbing the seed-cap and getting other caps
[2007/11/20 9:35] FWord Utorid: so essentially one could log in and view events without an avatar?
[2007/11/20 9:35] Saijanai Kuhn: so now we know how its done currently
[2007/11/20 9:35] Strife Onizuka: for that many users you would want some time of load distribution, no server in existance could handle it alone, let apart do the physics etc
[2007/11/20 9:35] Ahzzmandius Werribee: there's some backend work that the login server does as well.
[2007/11/20 9:36] Morgaine Dinova: The Use_Cases page on the AWG wiki gives details of such massively-scaled scenarios. They're end cases, but important.
[2007/11/20 9:36] Saijanai Kuhn: no physics, no nuttin. Just the 3D scene graph downloaded to the client
[2007/11/20 9:36] Ahzzmandius Werribee: it checks with your last known sim location to make certain that it's logged you out.
[2007/11/20 9:36] FWord Utorid: that defies my goal of spying on everyone :(
[2007/11/20 9:36] Saijanai Kuhn: the other end from that are chat-only clients, or clients that just log in to check your LInden/whatever account
[2007/11/20 9:37] FWord Utorid: libsecondlife has examples of this
[2007/11/20 9:37] Morgaine Dinova: The "LCCs" ... Limited Capability Clients.
[2007/11/20 9:37] FWord Utorid: has anyone here ever tried the polymorphism proxy?
[2007/11/20 9:38] Saijanai Kuhn: I'm examining libsl code all the time for this stuff. The issue is that libsl, while far cleaner, is no more flexable than any other compiled library
[2007/11/20 9:38] Morgaine Dinova: code.google.com is still getting around to responding to that on its 4040 / ZX-81
[2007/11/20 9:38] Saijanai Kuhn: not to mention its in C3 and I'm on a Mac.
[2007/11/20 9:38] Saijanai Kuhn: Not an easy thing to work with
[2007/11/20 9:38] FWord Utorid: sai, there will be hundreds of languages attempting to log into the grid
[2007/11/20 9:38] Ahzzmandius Werribee: so what exactly is the goal of this login service, in a short simple statement?
[2007/11/20 9:39] Saijanai Kuhn: to allow logging in from any grid and register your "presence" to all of them, more or less
[2007/11/20 9:41] Saijanai Kuhn: hopefully still will
[2007/11/20 9:42] Ahzzmandius Werribee: which seperates auth from service completely.
[2007/11/20 9:42] Saijanai Kuhn: the current (and probably future) capability is of the form url:port/cap/UUID
[2007/11/20 9:42] Ahzzmandius Werribee: Personally I don't see why the login service needs to be able to do anything more than just authenticating and serving up a token.
[2007/11/20 9:43] Morgaine Dinova: LLSD is restricted to LSL datatypes. That probably means that it won't figure highly in an interop/extensible setting.
[2007/11/20 9:43] Saijanai Kuhn: where UUID is a random-looking number that the server uses to provide services to you
[2007/11/20 9:43] Dr Scofield: the URL is for waht?
[2007/11/20 9:43] Dr Scofield: the agent domain/server?
[2007/11/20 9:43] Saijanai Kuhn: currently its to the login server
[2007/11/20 9:43] Rex Cronon: i don't like the term "capability", is belive is kind of missleading. i suggest to be repleace with something else
[2007/11/20 9:43] Dr Scofield: but that could go to an agent server
[2007/11/20 9:43] Saijanai Kuhn: eventually, will be an agent server, I think
[2007/11/20 9:44] Morgaine Dinova: Rex: why don't like? It reflects what it is
[2007/11/20 9:44] FWord Utorid: morgaine: llsd works for registration api
[2007/11/20 9:44] Dr Scofield: it's a capability in the security sense
[2007/11/20 9:44] Rex Cronon: no it doesn't reflect what it is
[2007/11/20 9:44] FWord Utorid: llsd could work for identifying resources to be transferred to other grid servers as well
[2007/11/20 9:44] Saijanai Kuhn: "capabiltiy" is a standard term. Zero likes it. Zha likes it. She gave a 2 hour lecture on how it works. DId it ever make it to the wiki chat logs?
[2007/11/20 9:45] Tao Takashi: maybe I will make some blog post about capabilities :)
[2007/11/20 9:45] FWord Utorid: i love them nearly as much as (insert her name here)
[2007/11/20 9:45] Rex Cronon: it implies that there is an entitye that has all those capabilities
[2007/11/20 9:45] Saijanai Kuhn: whatever happened to that 2 horu chat log of Zha's
[2007/11/20 9:45] Tree Kyomoon: it allows you to give access to a service without having to pass along your session informaton
[2007/11/20 9:45] Ahzzmandius Werribee: I see the caps-seed return as more of a kerberos token + DNS location equivilant. a blend of both auth and location.
[2007/11/20 9:45] Saijanai Kuhn: there's a connection point to those services via that single cap url
[2007/11/20 9:46] FWord Utorid: ahzz: they are looking at doing considerably more than just sharing auth
[2007/11/20 9:46] Tree Kyomoon: IE face book apps SHOULD run that way
[2007/11/20 9:46] Dr Scofield: kerb token + loc = cap
[2007/11/20 9:46] Saijanai Kuhn: right now, the "seed-cap" returned during login is a standard sim-cap URL. The sim that you log into
[2007/11/20 9:46] FWord Utorid: ahzz: they are looking at maneuvering avatar data and inventories and L$ transactions
[2007/11/20 9:47] Saijanai Kuhn: but that is not what they want to do for the meta-grid since its too grid-specific
[2007/11/20 9:47] FWord Utorid: ok meta-grid is an unofficial term for this new quasi-grid of near-grid goodness?
[2007/11/20 9:47] Dr Scofield: it should go to an agent server
[2007/11/20 9:48] FWord Utorid: i like omni-grid or omega-grid
[2007/11/20 9:48] Dr Scofield: meta-grid, open grid
[2007/11/20 9:48] Ahzzmandius Werribee: sai, that's why I was sayign that login needs to deal with auth pure and simple, then pass the client off to a "services" server that deals purely with directing the client to whatever services it needs/requests.
[2007/11/20 9:48] Saijanai Kuhn: something like that. Its my term for a collection of grids that all interact with SL in a SL compatible way
[2007/11/20 9:48] Saijanai Kuhn: thats what most of us think, Ahzz
[2007/11/20 9:49] Ahzzmandius Werribee: k, i'm a bit behind the curve here in the discussion. :-)
[2007/11/20 9:49] Saijanai Kuhn: login gives you yoru first cap, which takes you to the next stage
[2007/11/20 9:49] Morgaine Dinova: With the growth of AWG, perhaps more people could look at specific areas? Our VAG list is pretty threadbare. Interop, for starters. It's a huge area.
[2007/11/20 9:49] Dahlia Trimble: gould use a comprehensive glossary
[2007/11/20 9:49] Saijanai Kuhn: little or no extra info is given to you save that first cap
[2007/11/20 9:49] Morgaine Dinova: Dahlia: we have an AWG glossary
[2007/11/20 9:50] Saijanai Kuhn: Zero wanted to give us "default info" but we convinced him otherwise. Too many kinds of "clients" to predict what is a good "default" set of info
[2007/11/20 9:51] FWord Utorid: ok let's take what ahzz said for just a moment
[2007/11/20 9:51] FWord Utorid: there will need to be multiple tiers of login capabilities
[2007/11/20 9:51] Ahzzmandius Werribee: right now some of the major issues with libsl uses involve assumptions by the sim/services on what the client needs.
[2007/11/20 9:51] FWord Utorid: some clients may need to be simple and facilitate low level informational access
[2007/11/20 9:52] FWord Utorid: other clients may need to perform more complex tasks such as high level avatars and dahlia watchers
[2007/11/20 9:52] Ahzzmandius Werribee: fword, no, foget tiers of login, instead think along login -> service request
[2007/11/20 9:52] Rex Cronon: isn't this interesting. there is no "capability" in the glossary:)
[2007/11/20 9:52] Morgaine Dinova: Rex: well spotted :-)))))
[2007/11/20 9:52] Ahzzmandius Werribee: the client should request a generalized type of service from the services server after logging in.
[2007/11/20 9:52] Ahzzmandius Werribee: THEN the services server can make some assumptions of what the client will need.
[2007/11/20 9:52] Tree Kyomoon: I think capabilities are still conceptual..does anyone know of them being used somewhere?
[2007/11/20 9:52] Saijanai Kuhn: Ahzzmandius I think you misunderstood: login will give you the seed-cap and nothing else
[2007/11/20 9:53] Dr Scofield: tree, the current grid uses caps already
[2007/11/20 9:53] Ahzzmandius Werribee: sai, I understand that, I was respondign the "tiers of login" :-)
[2007/11/20 9:53] Saijanai Kuhn: login uses a seed-cap right now, but hands you bunches of SL specific data
[2007/11/20 9:53] FWord Utorid: the premise i am putting forth is to simplify the login as much as possible to enable entry level developers the opportunity to connect easily and therefore screw things up
[2007/11/20 9:53] Saijanai Kuhn: well, there's a login/registration you would need to go through to establish a presence on a given grid and a given sim, but thats afte rthe initial login
[2007/11/20 9:54] Ahzzmandius Werribee: here's what I see: client sends login to login server, login server passes back initial seed cap, client requests services type from seed cap url, service server returns default data based on service type requested. can request multiple service types. im, presence, etc.
[2007/11/20 9:54] Tree Kyomoon: is suspect of the security of capabilities...and the ability to exploit them in an open source environment
[2007/11/20 9:54] Saijanai Kuhn: so in that sense, ther emight be a "tier" of logins, but only one layer, if that
[2007/11/20 9:55] Saijanai Kuhn: Ahzz, right. That's pretty much what happens now, but with extraneous STUFF
[2007/11/20 9:55] Saijanai Kuhn: we want to get rid of the stuff
[2007/11/20 9:55] Ahzzmandius Werribee: if there's more than one login domain, then let the login server handle forwarding the login request to the proper authority in the backend transparently to the client.
[2007/11/20 9:55] Strife Onizuka: exploitation of capabilities is a sersious problem
[2007/11/20 9:56] Ahzzmandius Werribee: sai, actually right now the initial login pass 1 gives the extraneous, split the login request into 2 parts THEN contact the sim seed caps url as part 3.
[2007/11/20 9:56] FWord Utorid: any multigrid environment has questionable security if the commodities can be transferred between one server and another
[2007/11/20 9:57] FWord Utorid: i'm being very broad here when i say that i do not foresee a means for inventory transfer or object transfer between one simulator and another represents an IP drm rift
[2007/11/20 9:57] Ahzzmandius Werribee: when you hit the seed cap, you send an XML data set containing what kinds of services you need for the client, such as IM, inventory for grid-X, presence for grix-X
[2007/11/20 9:57] Dr Scofield: stuck in the elevator?
[2007/11/20 9:57] Tree Kyomoon: we are staging a mutiny :)
[2007/11/20 9:57] Zha Ewry: Sorry, got way sucked into a discussion with the guru of GPL and things legal
[2007/11/20 10:00] Dr Scofield: or agent server, that knows how to locate services for an av
[2007/11/20 10:00] FWord Utorid: provisions for multiple types of clients in the login service will facilitate informational access for a variety of borg
[2007/11/20 10:01] Zha Ewry: No point in getting told all about things you can't use anyway
[2007/11/20 10:01] Dr Scofield: or don't want to use
[2007/11/20 10:01] Tao Takashi: btw, the chatlog from last meeting is uploaded now, Zha
[2007/11/20 10:01] Tao Takashi: I think many things are in there already
[2007/11/20 10:01] Dr Scofield: what did you learn about caps, zha?
[2007/11/20 10:01] FWord Utorid: there should be like grade school for clients
[2007/11/20 10:01] Ahzzmandius Werribee: that's why I'm saying send only what pertains to the types of services that client requests.
[2007/11/20 10:01] FWord Utorid: a first grade app can only get the alphabet
[2007/11/20 10:01] Tao Takashi: when I have time I might consolidate this and Zero's OH about login/inventory into some post
[2007/11/20 10:01] Ahzzmandius Werribee: the DNS system is a perfect example of what i'm talking about.
[2007/11/20 10:02] Tao Takashi: and I hope we get the caps server soon :)
[2007/11/20 10:02] FWord Utorid: a college level app can perform physics
[2007/11/20 10:03] Zha Ewry: So... I have been watchibg over Saij's shoulder. I think, from what he's been seeing, that we do need a handshake step, after the first bit
[2007/11/20 10:03] FWord Utorid: is it possible that future login services can avoid a multiple protocol complication?
[2007/11/20 10:03] Zha Ewry: I see no reasonw hy we should really get more than the seed cap back. I could be had, on sending a short list of servcies one wants caps to, in the first login, but it doesn't buy you much
[2007/11/20 10:04] Morgaine Dinova: Zha: Rex noticed that Capabilities isn't in our Glossary. :P And come to think of it, they aren't discussed properly anywhere in AWG or Groupies space.
[2007/11/20 10:06] Zha Ewry: But get/set./post/delete
[2007/11/20 10:06] Dr Scofield: you can use the cap to get invoke a service
[2007/11/20 10:06] Dr Scofield: but it's not the service itself
[2007/11/20 10:07] Zha Ewry: Think, as much as possible in terms of nouns we can set/get as the services, rather than SOAPy things where you are calling a method
[2007/11/20 10:07] Tao Takashi: so is there any news as when LL wants to release their caps server? :) if it would be around I might start coding ;-)
[2007/11/20 10:07] Zha Ewry: REST tries to avoid SOAP/RPC styl;e methods as they are very hard to make sense of, and don't really transfer state
[2007/11/20 10:08] Dr Scofield: noteverything fits into REST, though
[2007/11/20 10:08] j3rry Paine: soap is what you should wash your mouth with - never seen anything fill a vacuum so fast and still suck since X Windows
[2007/11/20 10:08] Tao Takashi: Was nicht passt, wird passend gemacht ;-)
[2007/11/20 10:10] Saijanai Kuhn: it will likely be the login module for the modular client
[2007/11/20 10:10] FWord Utorid: the advent of a webpage that could control an avatar would be remarkable imho
[2007/11/20 10:10] Zha Ewry: Well.. depends on what you mean by control
[2007/11/20 10:10] FWord Utorid: and this with a very limited amount of data transfer
[2007/11/20 10:11] FWord Utorid: well i can see no reason why i couldn't make a javascript to log in an avatar, maneuver it, make it communicate, and log out
[2007/11/20 10:11] FWord Utorid: the present udp is really the restrictive element here
[2007/11/20 10:12] FWord Utorid: i'll just call you a paine, j3rry ;)
[2007/11/20 10:13] Zha Ewry: But. the major thrust.. for most of this. is going to be to float on pretty generic web services. which should make things like that a lot eaiser
[2007/11/20 10:14] Ahzzmandius Werribee: and I agree that login/services should avoid UDP.
[2007/11/20 10:14] Saijanai Kuhn: udp makes sense for what, key presses?
[2007/11/20 10:14] Ahzzmandius Werribee: that's up to the individual end services.
[2007/11/20 10:14] Zha Ewry: Anything.. where the next packet.. replaces fully, the lost one?
[2007/11/20 10:15] Dahlia Trimble: I thought udp reduced server memory requirements
[2007/11/20 10:15] Ahzzmandius Werribee: let's not worry about how the end services communicate for now.
[2007/11/20 10:15] Strife Onizuka: hmmm, maybe with hte login request, a version number for the style of login, so we can support old style logins and then increase the complexity as our caps system evolves?
[2007/11/20 10:15] Zha Ewry: Trivially, on the grand scale of things
[2007/11/20 10:15] Zha Ewry: (the memory difference)
[2007/11/20 10:15] FWord Utorid: sai: reviewed that link, and yes, i have experienced these applications, and yes, they aren't really as cool as they could be because of the udp requirements
[2007/11/20 10:15] Strife Onizuka: (sorry for being off topic)
[2007/11/20 10:15] Ahzzmandius Werribee: strife, that can be encoded as a part fo the UUID
[2007/11/20 10:15] FWord Utorid: and keystrokes could be done by methods other than udp as well
[2007/11/20 10:16] Zha Ewry: What is nice about UDP.. is that it has slightly better behavior in the face of stalls and packet loss
[2007/11/20 10:16] Saijanai Kuhn: not as part of the UUID, but as whatever is accessed by the UUID
[2007/11/20 10:16] Zha Ewry: But.. as soon as you care, at all
[2007/11/20 10:16] Zha Ewry: about reliability... UDP.. basically gets in the way
[2007/11/20 10:16] Dr Scofield: well... as long as you don't end up reimplementing TCP
[2007/11/20 10:16] Zha Ewry: you end up re-creating tcp,
[2007/11/20 10:16] j3rry Paine: ;-) what tcp is for lol
[2007/11/20 10:16] Ahzzmandius Werribee: sai, but the UUID represents a specific login session yes?
[2007/11/20 10:19] Ahzzmandius Werribee: because if this is restfull, then you "GET" a redirection or set of data for a specific capability
[2007/11/20 10:19] FWord Utorid: also, one thing i haven't mentioned that i was considering, if it were possible to log in, post data, and log out, in a single hit
[2007/11/20 10:19] FWord Utorid: for the aforementioned midgetbot javascript microapps.
[2007/11/20 10:19] Zha Ewry: You'll pretty much never do that with a capability.. because
[2007/11/20 10:19] Saijanai Kuhn: might have a dedicated server for that, but its not partr of hte regular login
[2007/11/20 10:19] Zha Ewry: you need to do the two step.. to get a capabiilty
[2007/11/20 10:20] FWord Utorid: this is why i was prescribing a tiered login architecture
[2007/11/20 10:20] FWord Utorid: so large complex apps can perform negotiation
[2007/11/20 10:20] Morgaine Dinova: Zha: what's the current state of play with your end of things? We need to start analysing services soon I think ... but need you to prime the pump on REST services.
[2007/11/20 10:20] FWord Utorid: and simple apps can do simple things
[2007/11/20 10:20] Saijanai Kuhn: the modular client cold have its own "login" service for a given erson. Its always "on" you login, request yoru linden accoutn via your iPHone, and log off
[2007/11/20 10:20] Zha Ewry: Simlpe apps, don't have a secure place, or way to go
[2007/11/20 10:21] FWord Utorid: some information is truly superflouous yet entertaining to query
[2007/11/20 10:21] Zha Ewry: The whole point of capabitlies is that they get security into one place.. and then put the services, into a secure, and re-directable place
[2007/11/20 10:21] Saijanai Kuhn: But that would be YOUR custom server, not the generic login
[2007/11/20 10:22] Zha Ewry: Login says "I am Zha Ewey" and gets me one or more seed caapbitlites
[2007/11/20 10:22] Zha Ewry: I then.. can use those.. for some length of time, in normal REST fashion
[2007/11/20 10:22] Zha Ewry: So.. if I for example, want to login, then do a bunch of short hits
[2007/11/20 10:22] Zha Ewry: I could grab the capabilties, and issue requests against them
[2007/11/20 10:23] Zha Ewry: But.. the mainline path for clients to talk into a space like this.. is probably going to be biased in that direction. That.. most interactions will be for enough time to amortize the setup cost of getting the caps
[2007/11/20 10:24] j3rry Paine: efficiency depends on minimizing network traffic. can'tcha do it in fewere hits?
[2007/11/20 10:27] Morgaine Dinova: We're not assuming sanity in the clients. That would have an unhappy ending.
[2007/11/20 10:27] FWord Utorid: if anyone knows what i am driving at
[2007/11/20 10:28] Zha Ewry: If you have two apps which share the capabilities and send conflicting requests...
[2007/11/20 10:28] Zha Ewry: I think, at some level, you deserve what you get
[2007/11/20 10:28] Saijanai Kuhn: banning on a parcel, sim and grid level would be an option for an avatar OR for the identify behind the avatar if things got bad enough
[2007/11/20 10:34] Zha Ewry: that.. there is a single agent.. which has a single client it talks to.. and the client talks to one or more region simulators.. which coordinate the 3d aspects of things
[2007/11/20 10:34] Dr Scofield: all the client that use a multiplexing proxy that is
[2007/11/20 10:34] Zha Ewry: And.. if you log in, from one place.. you would be saying "I want my client here"
[2007/11/20 10:35] Zha Ewry: and if you log in from another your saying "Here now"
[2007/11/20 10:35] Morgaine Dinova: If its a capability-based RESTful system, it's driven by messages. If you can hand over any capabilities you have to another of your clients, that's fine from a REST perspective. It's not the server's business to dictate how your client is structured ... you may be distributed all over the planet.
[2007/11/20 10:35] Zha Ewry: You can do whetever you want upstream of that endpoint
[2007/11/20 10:37] FWord Utorid: i think i made the reasoning behind the request clear... coherency isn't the issue, it's access from multiple points and transferrence of primary me
[2007/11/20 10:37] Saijanai Kuhn: a real person might have multiple identise on the grid. Each identity can have one or more avatars in one or more places, but only one place per avatar on a given grid
[2007/11/20 10:38] FWord Utorid: someone just suggested a test server for this, and i hope everyone heard
[2007/11/20 10:38] Ahzzmandius Werribee: restrciting presence to one sim per grid would be up to the presence service of that specific grid to enforce.
[2007/11/20 10:38] Zha Ewry: Unless there is a compelling, deep, and unavoidable reason to do it, we're pretty much commted to usign web building blocks unchanged
[2007/11/20 10:38] Ahzzmandius Werribee: we keep talking about merging of everything when we're trying to seperate. 8-P
[2007/11/20 10:38] Zha Ewry: The web model says that if you have the URL, and the stuff needed to seal the https session
[2007/11/20 10:39] Zha Ewry: then you get to issue the request
[2007/11/20 10:39] Zha Ewry: and we get to attempt to fulfill it
[2007/11/20 10:39] Zha Ewry: If you pass your keys and endpoints around to six apps
[2007/11/20 10:39] Saijanai Kuhn: true, but itwould be rather confusing to have the same inventory point in two places in the same grid. Which gets to rez a no-copy item first?
[2007/11/20 10:39] Zha Ewry: and they do odd things
[2007/11/20 10:39] FWord Utorid: i just want to spy on dahlia. but the other stuff is cool too. when there's a test server i would love to experience it
[2007/11/20 10:40] Zha Ewry: Since.. you could be using a proxy, or a distribted app cluster to drive the interaction.. we jjust assume you're clever, and follow along
[2007/11/20 10:40] Zha Ewry: In terms of multiple logins and aves
[2007/11/20 10:40] Zha Ewry: that's actually 90% goign to show up in a few places
[2007/11/20 10:41] Zha Ewry: 90% of it would be unchanged
[2007/11/20 10:41] Zha Ewry: but.. we would have some real tricky spots
[2007/11/20 10:41] Dr Scofield: nobody would prevent fword from doing a value add login server though
[2007/11/20 10:41] Ahzzmandius Werribee: why not have multiple inventory servers, and when you want to rez something, you request the inventory server to rez at sim XYZ, which it then does by talking directly to the sim XYZ.
[2007/11/20 10:41] Rex Cronon: zha, were u going to do a presentation today, or i missunderstood?
[2007/11/20 10:41] Ahzzmandius Werribee: that solves the rezing multiple copies of a no-copy item and concurency
[2007/11/20 10:41] Saijanai Kuhn: I think this IS the presentation, Zha-style
[2007/11/20 10:41] Zha Ewry: I had hoped ot go through my pretty capabitily foils Bt.. Instread.. I spent 2 extra hours gettigg my head shapped by Eben
[2007/11/20 10:41] Morgaine Dinova: Zha: what's the current state of play with your end of things? We need to start analysing services soon I think ... but we need you to prime the pump on REST services.
[2007/11/20 10:45] Zha Ewry: Oh.. and by the way.. if you think Eben's insight valuble.. he charges us with the goal of allowing as much fo a digital persona's assets and control to be kept in thier own servers as thety desire
[2007/11/20 10:45] Morgaine Dinova: Well AWG did have Gareth for prototyping the services early, but he's gone to a competitor. So now we need someone to do server-side prototyping.
[2007/11/20 10:45] Zha Ewry: I'm thinking that in terms of an early prorotype
[2007/11/20 10:52] Tao Takashi: UML would be nice ;-)
[2007/11/20 10:52] Saijanai Kuhn: righ, I used Python cause I'm tryng to get a LInden job, and Python is the LInden's favoriate scripting language and I need practice
[2007/11/20 10:52] Tao Takashi: so one knows where it starts and how the flow goes
[2007/11/20 10:52] Zha Ewry: is aklso pretty good for this sort of protoyping
[2007/11/20 10:52] Dr Scofield: python as speci'ing language
[2007/11/20 10:52] Zha Ewry: Or is getign a godo statrt
[2007/11/20 10:52] Saijanai Kuhn: easy to read compared to many scritping lagnuages
[2007/11/20 10:52] Ahzzmandius Werribee: sai, i'm working on a library called botmaster that will let me run thousands of libsl clients in a networked grid across many servers.
[2007/11/20 10:53] Zha Ewry: I think there is a missing
[2007/11/20 10:55] FWord Utorid: they facilitate regapi for that
[2007/11/20 10:55] Zha Ewry: All premium, no doubt :-)
[2007/11/20 10:55] Morgaine Dinova: Ahzz: very valuable both for testing protocols and robustness, and also to help in scalability analysis.
[2007/11/20 10:55] Ahzzmandius Werribee: I had to beat up hurliman and a couple others to get them to restructure some of libsl to make it multi-instance stable. :-)
[2007/11/20 10:56] FWord Utorid: that would be nice
[2007/11/20 10:56] Saijanai Kuhn: WOld reduce server lag at least a bit for NPCs
[2007/11/20 10:56] Zha Ewry: Sigh. I need to run, or New York State is going to get cranky about my car's inspection.. and while Zha can TP everwhere..
[2007/11/20 10:56] FWord Utorid: as i understand it sl ships all of the data for npcs except media
[2007/11/20 10:56] Zha Ewry: IRL.. I have problems with tp
[2007/11/20 10:58] Rex Cronon: the map search no longer works:(
[2007/11/20 10:58] Cel Edman: I think you got a point, Saij. from just an irc-chat server you dont need to send all 3D data. But like if you got top of the line 3D card, you can send more data to client if they selected that as well
[2007/11/20 10:59] Saijanai Kuhn: don't post to AWG without getting permission from Zero. He's cranky
[2007/11/20 10:59] Morgaine Dinova: Ahzz: me too, on amd64 Gentoo ... Windlight looks nice. A bit flakey, but pretty :-)
[2007/11/20 10:59] Saijanai Kuhn: instead, use AW Groupies. We're nice
[2007/11/20 10:59] Ahzzmandius Werribee: bah, I can't edit the wiki, i'll post the dia to my web server and someone else with auth can grab and add it.
[2007/11/20 10:59] Dr Scofield: rex, seems to still wrok on the windlight cleint
[2007/11/20 10:59] Dr Scofield: ahzz, just login with your SL creds
[2007/11/20 11:00] Rex Cronon: for example i type "Brampton", and it can't find it:(
[2007/11/20 11:02] Saijanai Kuhn: wants that svg on a prim
[2007/11/20 11:02] Ahzzmandius Werribee: I'll leave it for now. it can be returned if it's considered trash. ;)
[2007/11/20 11:03] FWord Utorid: i think it's cool
[2007/11/20 11:03] FWord Utorid: i just think there are too many operations
[2007/11/20 11:03] Ahzzmandius Werribee: you can't avoid the level of ops realy when you want to cleanly seperate services.
[2007/11/20 11:03] Ahzzmandius Werribee: especially if you want to have multiple login servers by different companies and multiple services by different companies.
[2007/11/20 11:04] FWord Utorid: it's always debatable. my notion was varied levels of complexity for varied forms of transactions.
[2007/11/20 11:04] FWord Utorid: which is not simplistic either :O
[2007/11/20 11:04] Ahzzmandius Werribee: i mean look at getting on the net and editing an authenticated wiki today> :-)
[2007/11/20 11:04] Rex Cronon: tp://slurl.com/secondlife/Brampton/110/148/26
[2007/11/20 11:04] Saijanai Kuhn: is that a working example of how to obtain the linden balance right now, or a porposal on how to od it on the metagrid
[2007/11/20 11:06] FWord Utorid: i want to make a flash interface
[2007/11/20 11:06] Saijanai Kuhn: which makes sense if its being sent as an update to the client
[2007/11/20 11:06] Ahzzmandius Werribee: if you look at the returned list of capabilities from the seed-caps url you can see what all happens via caps right now.
[2007/11/20 11:06] j3rry Paine: oh god. first python, then flash.
[2007/11/20 11:06] Tree Kyomoon: sorry if anyone tried to talk to me, we had a surprise visitor in RL
[2007/11/20 11:06] Saijanai Kuhn: but if you're just asking for a specific on-shot balance...
[2007/11/20 11:17] j3rry Paine: i have no default configuration
[2007/11/20 11:18] Ahzzmandius Werribee: if these lawyers ever settle on wordings then I get to have a full time job once more and learn how to be an effective manager. ;-P
[2007/11/20 11:18] Ahzzmandius Werribee: I can stop beating payments out of clients.
[2007/11/20 11:20] Saijanai Kuhn: was jk, Tree, wasn't sure if it was a band or what
[2007/11/20 11:20] Ahzzmandius Werribee: if you want your appearance to NOT be ruthe, you also have to send the wear commands. libsl has a setPreviousAppearance(false) method to do that in one call.
[2007/11/20 11:20] Saijanai Kuhn: KK. Will look into it. Trying to implement it in python
[2007/11/20 11:21] Ahzzmandius Werribee: you'll need some packet processing in UDP then.
[2007/11/20 11:21] j3rry Paine: python makes me boob nervous
[2007/11/20 11:21] Saijanai Kuhn: I don't actually have to DO anything with them, just establish the connection, right?
[2007/11/20 11:21] Ahzzmandius Werribee: I had to learn all this back last March when I attempted to rewrite libsl ;-P
[2007/11/20 11:22] Ahzzmandius Werribee: well, you do have to establish the UDP link.
[2007/11/20 11:22] Saijanai Kuhn: smiles happily at his new best friend
[2007/11/20 11:22] Ahzzmandius Werribee: you'll also have to ACK reliable packets.
[2007/11/20 11:22] Ahzzmandius Werribee: or the sim will keep sending them and eventually dump you.
[2007/11/20 11:22] j3rry Paine: is struck dumb by ahzz's erudition. and a good thing too.
[2007/11/20 11:23] Saijanai Kuhn: ah, so for NPC client, at least currently, you need to maintain some UDP presence with the server. Makes sense, I guess
[2007/11/20 11:23] Saijanai Kuhn: would be interesting to explore a NPC client presence with Zero later today
[2007/11/20 11:24] Ahzzmandius Werribee: think of the sim as a proxy for your interactions with all the backend services in SL
[2007/11/20 11:24] Saijanai Kuhn: a storebot or a quest NPC shouldn't need to get constnat updates from the server. Just local chat
[2007/11/20 11:24] Ahzzmandius Werribee: the login server just get's you the basic starting URL and ip:port as well as basic information on what to request from the sim.
[2007/11/20 11:24] Ahzzmandius Werribee: agreed, and I've voiced this concern several times.
[2007/11/20 11:25] Ahzzmandius Werribee: that the sim should only send what's requested.
[2007/11/20 11:25] Saijanai Kuhn: straright to Zero?
[2007/11/20 11:25] Ahzzmandius Werribee: but the curent system is designed around the sim controlling the client. with the client being only mouse/keyboard input and a draw mechanism.
[2007/11/20 11:25] Ahzzmandius Werribee: no, I don't think I've talked with zero before.
[2007/11/20 11:26] Saijanai Kuhn: right. And in fact and NPC doesn't need to know how to draw itself. Its a glorified chat client with gestures
[2007/11/20 11:26] Ahzzmandius Werribee: I haven't been very pushy with anyone other than libsl crew up till now. ;)
[2007/11/20 11:27] Saijanai Kuhn: best way to get a job is to show up at office hours and show you know what you're talking about (or at least pretend really well)
[2007/11/20 11:28] Zha Ewry: Well. we have. within reason, rights to pester ;-)
[2007/11/20 11:28] j3rry Paine: i love to pretend!
[2007/11/20 11:28] Ahzzmandius Werribee: I dunno if LL would be willing to pay me what I think i'm worth, or let me telecommute full time. ;)
[2007/11/20 11:28] Tree Kyomoon: you guys want to work at linden lab?
[2007/11/20 11:28] Saijanai Kuhn: I've interviewed with 5 linends now I think
[2007/11/20 11:28] Morgaine Dinova: HAHAHAHA --- [8] (the wiki SVG converter is really bad at SVG font conversion) :PPPP
[2007/11/20 11:28] Saijanai Kuhn: I'm looking for a remote telecomute job as AQ
[2007/11/20 11:28] Ahzzmandius Werribee: tree, i'm not chasing it, but I wouldn't outright reject an offer.
[2007/11/20 11:30] Saijanai Kuhn: had 2 interviews last week and one the week before and one the week before and one the week before and one the week before...
[2007/11/20 11:30] Saijanai Kuhn: on back to the beginnign of the universe
[2007/11/20 11:30] Ahzzmandius Werribee: my wife and my parents are pestering me lately to quit the consulting gigs and get a day job for stability.
[2007/11/20 11:30] Ahzzmandius Werribee: btw if anyone hadn't guessed, i'm a guy.
[2007/11/20 11:36] Morgaine Dinova: Ahzz: I have a computer room upstairs, it's a right mess, but at least it's all racked. 3 racks, mostly gigabit. No machinery downstairs, just quiet monitors etc via Cat5 KVM extenders.
[2007/11/20 11:36] FWord Utorid: i'm the one or ahzz is the one
[2007/11/20 11:36] FWord Utorid: i have a laptop with spacenet
[2007/11/20 11:36] j3rry Paine: i have ones as well as zeroes
[2007/11/20 11:36] Ahzzmandius Werribee: morgaine, i'd thought about using a high end cat5 KVM extender myself. but I run multihead and that would get expencive.
[2007/11/20 11:37] Ahzzmandius Werribee: so instead I opted for an ultra quiet workstation and a server closet for the servers.
[2007/11/20 11:37] Ahzzmandius Werribee: I jsut need to ditch the two 19" CRTs for LCDs.
[2007/11/20 11:37] Morgaine Dinova: Ahzz: not just the expense .... it's also annoying that all the Cat5 KVM extenders I've tried have bugs of one kind or another. Very annoying
[2007/11/20 11:38] Ahzzmandius Werribee: also build in a KVM switch for the kb/mouse so that as I swivel the chair to each monitor it automaticly switches
[2007/11/20 11:42] Ahzzmandius Werribee: when my sister started working at lockheed martin she got pestered with "oh no not another Wolfe!" due to my dad's lasting impressions upon the staff there after 30 years.
[2007/11/20 11:43] Saijanai Kuhn: FWord was problematic for me. He and seg didn't get along at all well. He likes to play semantic games that grate on some peole
![[8]](http://wiki.secondlife.com/w/images/d/d9/Ahzzmandius-newcaps-0.svg){kind=link}
![[9]](http://pictures.terrabox.com/newcaps.jpg){kind=link}