AW Groupies/Chat Logs/AWGroupies-2009-10-20

From Second Life Wiki
Jump to navigation Jump to search
  • [2009/10/20 9:31] Zha Ewry: Welcome Rex
  • [2009/10/20 9:32] Rex Cronon: greetings
  • [2009/10/20 9:32] Rex Cronon: hi
  • [2009/10/20 9:33] Zha Ewry: Heads up for next week..the topic will be "edits on the VWRAP/OGPX intro document" focused on some of what I'm going to be talking about today
  • [2009/10/20 9:33] Rex Cronon: btw zha. u should convert that tiny molecle to sculpties. it would use a lot less prims:)
  • [2009/10/20 9:33] Rex Cronon: molecule*
  • [2009/10/20 9:34] Zha Ewry: The one over the lake?
  • [2009/10/20 9:34] Rex Cronon: yes
  • [2009/10/20 9:34] Zha Ewry: the beasty is built programtically, based on RL data, so mapping 3d endopints of bonds to prims is easy, creating scupltmaps hard ;(
  • [2009/10/20 9:35] Zha Ewry: (Once could clearly take the logical tree, and map it into a set of scultmaps, but bleh)
  • [2009/10/20 9:36] Zha Ewry: Gonna give people abotu 3 more minutes
  • [2009/10/20 9:37] Lillie Yifu: sorry was adjusting a new pair of sandals
  • [2009/10/20 9:38] Zha Ewry: So, what I'm goign to talk about, is fodder for next weeks more document editing focused meeting, I hope
  • [2009/10/20 9:38] Zha Ewry: Which is trying to pin down words for and a rought agreement on what the actual leverage points we have for managing policy are, in the specs
  • [2009/10/20 9:38] Zha Ewry: with.. an emphasis on making sure we keep the bulk of th mechanisms in the main specs, and any specifics about any one policy out
  • [2009/10/20 9:39] Rex Cronon: u know zha. nyx has a list of topics for his OH. we could also have a list of topics for this weekly meeting:)
  • [2009/10/20 9:39] Zha Ewry: That requires me to be organized ;-)
  • [2009/10/20 9:39] Pixel Gausman: dares Zha to hold a single groupies meeting without using the word 'Policy'
  • [2009/10/20 9:39] Zha Ewry: Next week's is goign to be trying to get some discussino on new sections for the intor doc outlined and discussed
  • [2009/10/20 9:40] Zha Ewry: So.. I'll actually post a notice to theg roup,a nd maybe the wiki
  • [2009/10/20 9:40] Zha Ewry: Scare people.
  • [2009/10/20 9:40] Zha Ewry: Eccept
  • [2009/10/20 9:40] Lillie Yifu: can we use the word %$(#@!!! instead?
  • [2009/10/20 9:40] Zha Ewry: Except
  • [2009/10/20 9:40] Pixel Gausman: Lilie: yes
  • [2009/10/20 9:40] Zha Ewry: Blast. I won't be here next week, unless my kids tooth surgery goes faster thane xpect
  • [2009/10/20 9:40] Zha Ewry: Sheeesh
  • [2009/10/20 9:40] Zha Ewry: It cannot be 10/20/2009 already
  • [2009/10/20 9:40] Zha Ewry: kicks the calendar
  • [2009/10/20 9:41] Zha Ewry: Sigh. I may post a re-schedule on next week
  • [2009/10/20 9:41] Lillie Yifu: It is 10/21/09 in China if that makes you feel any better
  • [2009/10/20 9:41] Zha Ewry: Oh, good, maybe they can tell me what the lottery in NY is tonight then;-)
  • [2009/10/20 9:41] Zha Ewry: So.. for fun.
  • [2009/10/20 9:42] Zha Ewry: To be horribly formal, anything the IETF level spec is going to care about in terms of policy is giong to be expressed as what LLIDL moves to or from an endpoint
  • [2009/10/20 9:43] Zha Ewry: the spec currently has 2 types of endpoints, message oriented ones (think UDP) and Caps ones (think http(s))
  • [2009/10/20 9:43] Zha Ewry: There's lots of other spots people can code policy deicsions, but they won't show up in the spec
  • [2009/10/20 9:43] Zha Ewry: so.. as an example, a "nanny style" chat service
  • [2009/10/20 9:43] Lillie Yifu: hmmmmm
  • [2009/10/20 9:43] Lillie Yifu: two points
  • [2009/10/20 9:43] Lillie Yifu: one is that the typing anim looks particularly silly on that chari Zha
  • [2009/10/20 9:43] Lillie Yifu: the other is
  • [2009/10/20 9:44] Zha Ewry: Might have a "policy" of replacing "bad" words with asterisks
  • [2009/10/20 9:44] Zha Ewry: laughs hard
  • [2009/10/20 9:44] Lillie Yifu: why not have a section of the spec on policy itself?
  • [2009/10/20 9:44] Zha Ewry: Ah, we will
  • [2009/10/20 9:44] Zha Ewry: In fact severla, I think
  • [2009/10/20 9:44] Zha Ewry: I'm sneaking up on proposing some ways of tackling that
  • [2009/10/20 9:44] Rex Cronon: u mean like right now if u type more than one space is truncated to one?
  • [2009/10/20 9:45] Zha Ewry: right
  • [2009/10/20 9:45] Mystical Demina: shouldn't endpoint consist of 3 things, message format, protocal and transport, so you do't lock yourself into one way of doing things
  • [2009/10/20 9:45] Zha Ewry: The current spec, maps all its endpoints onto URIs which encapsulate two types of endpoints
  • [2009/10/20 9:45] Zha Ewry: (Caps like ones and message like ones)
  • [2009/10/20 9:46] Zha Ewry: They don't acutally say HTTP(s) and UDP
  • [2009/10/20 9:46] Rex Cronon: i think message format is part of the protocol
  • [2009/10/20 9:46] Zha Ewry: but at the moment, that's how they map
  • [2009/10/20 9:46] Zha Ewry: nothing says you can't build a LLIDL ovr XMPP spec
  • [2009/10/20 9:46] Zha Ewry: except, for the tricky bit of how TLS woudl fit in
  • [2009/10/20 9:46] Mystical Demina: ok
  • [2009/10/20 9:46] Zha Ewry: (Which kind of makes me unhappy, but, security, in general and X.509, and crypto always may everyone unhappy)
  • [2009/10/20 9:47] Zha Ewry: So.. for one more moment of digressin about "nany policy"
  • [2009/10/20 9:47] Zha Ewry: Its a nice example of somethign that may be policy of a region
  • [2009/10/20 9:47] Zha Ewry: but doesn't change the LLIDL we emit or consume in a way we care about very much
  • [2009/10/20 9:48] Zha Ewry: You might well have a propery someplace which announces "This is a PG text censoring regoin which will attempt to remove offenseive workds like "L I N D E N" from discussion
  • [2009/10/20 9:48] Zha Ewry: but, the LLIDL message formats won't change because of that
  • [2009/10/20 9:49] Zha Ewry: (And of couse, the simple expedient of saying things like L1nden and L*nden" will suffice to defeat the policy)
  • [2009/10/20 9:49] Zha Ewry: I raise this, because, people are talking about policy incredibly broadtly, and what is likely to end up in a spec is much narrower, namely what actually shows up in messages and endopint behaviots
  • [2009/10/20 9:50] Lillie Yifu: hmmmm
  • [2009/10/20 9:50] Zha Ewry: So, a "policy" of "This is a PG region" may encode in terms of what the protocol says, in very small ways, in terms of most services the region offers
  • [2009/10/20 9:50] Lillie Yifu: should there be in the protocol some kind of message passed back when a message affected by local plicy is received
  • [2009/10/20 9:50] Zha Ewry: That's a very fair question
  • [2009/10/20 9:51] Zha Ewry: and.. Its also really hard to answer, because "in what way" becomes messy
  • [2009/10/20 9:51] Lillie Yifu: so if I type "Linden Labs Lags my avatar"
  • [2009/10/20 9:51] Lillie Yifu: and the message is chagned to "I LOVE PHILLIP"
  • [2009/10/20 9:51] Zha Ewry: Suppose, for some reason, I have a policy which says "No dark red prims on my region" and I actually look at all the texture maps which go in
  • [2009/10/20 9:51] Lillie Yifu: the region should pass back
  • [2009/10/20 9:52] Lillie Yifu: "your message Liden Labs Lags my avatar was changed to ILOVE PHILIP by region policy."
  • [2009/10/20 9:52] Zha Ewry: and fix any dark red textures to the more pleasing "beige" texture I like, do I have toflag that
  • [2009/10/20 9:52] Lillie Yifu: Or would that crate too much back and forth verbosity?
  • [2009/10/20 9:52] Zha Ewry: I *think*
  • [2009/10/20 9:52] Zha Ewry: and this is very much a "I've thought about this a little bit, no more"
  • [2009/10/20 9:52] Lillie Yifu: right now for example on LL search
  • [2009/10/20 9:52] Lillie Yifu: it tell s you if you try to search on an adult term
  • [2009/10/20 9:52] Zha Ewry: I think that the most I would be happy with would be if you asked for a cap for the "text message cap"
  • [2009/10/20 9:52] Mystical Demina: policy is a pretty ambiguous term, not sure the distinction between policy and protocol, seems the end point should only be concerned with the workflow and finite state automata of what it gets and what it returns when
  • [2009/10/20 9:52] Lillie Yifu: when adult content is not checked
  • [2009/10/20 9:53] Zha Ewry: and you got back a "nannybot enforcing text messaging cap" you wuold be told.
  • [2009/10/20 9:53] Lillie Yifu: that might be a better way
  • [2009/10/20 9:53] Zha Ewry: Which is to say, one could imagine saying 'The semantic of the messaging cap is it echos text unchanged"
  • [2009/10/20 9:53] Zha Ewry: The semantic of the nannycap is "It changes text"
  • [2009/10/20 9:53] Lillie Yifu: get a cap from the region that is a log of asy the last minute's worth of changes
  • [2009/10/20 9:54] Zha Ewry: and allow the region by policy to giv you back a nany cap
  • [2009/10/20 9:54] Lillie Yifu: if you wnat to know, check it
  • [2009/10/20 9:54] Lillie Yifu: and it truncates quickly
  • [2009/10/20 9:54] Zha Ewry: if the region cares to share
  • [2009/10/20 9:54] Lillie Yifu: there could be a vareity fo levels of verbosity
  • [2009/10/20 9:54] Lillie Yifu: with the default meing just a link to the rd's policy list
  • [2009/10/20 9:54] Zha Ewry: Since, the region deployer and code implemntor is always free to say "Nope, I never show rude words"
  • [2009/10/20 9:54] Lillie Yifu: but more and more levels of verbosity on request
  • [2009/10/20 9:54] Lillie Yifu: which may or may not be iplemented, but if they are, are implemented by a stanrdard format
  • [2009/10/20 9:55] Zha Ewry: so.. The reason for this dirgession into nany text, and similar
  • [2009/10/20 9:55] Zha Ewry: is to suggest a very natural split in my mind
  • [2009/10/20 9:55] Zha Ewry: between the tools and leverage points needed to permit policy
  • [2009/10/20 9:55] Zha Ewry: and the specific policies
  • [2009/10/20 9:55] Zha Ewry: And also highlight, that theo nly place things like this get captured, is in the messages and behaviros of formal endpoints
  • [2009/10/20 9:56] Zha Ewry: So.. for example, when deciding to let or not let someone into a region, due to a policy decision
  • [2009/10/20 9:56] Zha Ewry: how much of that shows up in the protocol
  • [2009/10/20 9:56] Zha Ewry: I'd argue two bits
  • [2009/10/20 9:57] Mystical Demina: i would say just the return message and its attribributes, everything else behind it should be decoupled
  • [2009/10/20 9:57] Zha Ewry: Any things that end up needing to be speced to help in the decision, and any replies which might be changed due t policy
  • [2009/10/20 9:57] Zha Ewry: Pretty mch Mystical
  • [2009/10/20 9:57] Zha Ewry: Inputs needed to help make the decision
  • [2009/10/20 9:57] Zha Ewry: and outputs which may be effected by it
  • [2009/10/20 9:58] Zha Ewry: At which point, I would *think* that one way to try and peal the onion is to
  • [2009/10/20 9:58] Zha Ewry: keep a minimal set of those inputs and messages in the core spec
  • [2009/10/20 9:58] Mystical Demina: so guess the protocol can identiy all the possible returns and possible suppotr a way for peope to extend the message for custom things
  • [2009/10/20 9:58] Zha Ewry: and then, populate the policy specs with "and here is how policy exposes it"
  • [2009/10/20 9:58] Zha Ewry: Well, since we're not psychic
  • [2009/10/20 9:59] Zha Ewry: (at least I'm not)
  • [2009/10/20 9:59] Zha Ewry: We probably take that sort of extension point approach
  • [2009/10/20 9:59] Zha Ewry: "Here's the terms that we got to group consensus on"
  • [2009/10/20 9:59] Zha Ewry: "Here's how we define new ones"
  • [2009/10/20 9:59] Zha Ewry: "Here where you can extend an "Denied by policy" response to include a new reason"
  • [2009/10/20 10:00] Zha Ewry: It parallels, closely,the split between mechansim and policy I've been harping on like a broken record player
  • [2009/10/20 10:00] Zha Ewry: "Mechanism, Mechanism for doing policy, policies"
  • [2009/10/20 10:01] Pixel Gausman: plugs ears
  • [2009/10/20 10:01] Fleep Tuque: Am I late or is no one rezzing. ;)
  • [2009/10/20 10:01] Mystical Demina: another work for mechanism is protocol? the order of things and what is allowed when?
  • [2009/10/20 10:01] Zha Ewry: About 10 people here
  • [2009/10/20 10:01] Mystical Demina: word*
  • [2009/10/20 10:01] Zha Ewry: Well
  • [2009/10/20 10:01] Zha Ewry: To an extent, yes
  • [2009/10/20 10:01] Rex Cronon: a little late
  • [2009/10/20 10:01] Fleep Tuque: Hope I'm not sitting on anyone then!
  • [2009/10/20 10:01] Zha Ewry: I think of mechanism as a combination of policy and expected semanic
  • [2009/10/20 10:02] Zha Ewry: Sorty
  • [2009/10/20 10:02] Zha Ewry: typoe
  • [2009/10/20 10:02] Zha Ewry: I think of mechanism as a combination of protocol and expected behavior/semantic
  • [2009/10/20 10:02] Zha Ewry: So..
  • [2009/10/20 10:02] Zha Ewry: Its "here's how you express "Please rez this prim"
  • [2009/10/20 10:02] Zha Ewry: the protcool
  • [2009/10/20 10:02] Zha Ewry: and the expcted behavior
  • [2009/10/20 10:02] Zha Ewry: "A prim will be added to the region's scenegraph at the indicated location"
  • [2009/10/20 10:02] Zha Ewry: along witht he follwoing side effects"
  • [2009/10/20 10:03] Zha Ewry: "Any script within the prim will be started, if the prim is physical, it will begin to be acted upon by the region's simulation engine..."
  • [2009/10/20 10:03] Zha Ewry: Whereas the policy might be
  • [2009/10/20 10:03] Zha Ewry: "We only rez objects from trusted asset stores"
  • [2009/10/20 10:04] Zha Ewry: and fleep, you're not sitting on anyone, but you're on a "tiny" beanbag, which is making ou look al ittle odd
  • [2009/10/20 10:04] Fleep Tuque: laugh
  • [2009/10/20 10:04] Fleep Tuque: It's pretty comfy though!
  • [2009/10/20 10:04] Zha Ewry: I think your AO is keeping you from folding up like a tiny
  • [2009/10/20 10:04] Mystical Demina: so policy would be for the most part only seen when i reject something you send me and you tellin gme why, otherwise i will get the expected return response
  • [2009/10/20 10:04] Fleep Tuque: haha
  • [2009/10/20 10:05] Zha Ewry: I thnk yes, Mysictal, tha's prety much the main effect
  • [2009/10/20 10:05] Zha Ewry: and, thus, harkening back to Lillie's comments
  • [2009/10/20 10:05] Zha Ewry: it makes a lot of sense, if I have a policy to use a "nanny chat service"
  • [2009/10/20 10:05] Zha Ewry: to note at cap grant time, that I granted you that, not the default one
  • [2009/10/20 10:06] Mystical Demina: so an interface will want to be able to querty you policcies, which is like querying your capabilities or services you support
  • [2009/10/20 10:06] Zha Ewry: Likewise, if as a "I don't know you very well" avataar, I grant you no rezzing rights, that may be expressed in the lack of a "rez object" cap
  • [2009/10/20 10:06] Zha Ewry: well
  • [2009/10/20 10:06] Zha Ewry: sighs deeply
  • [2009/10/20 10:06] Zha Ewry: In theory, "sure"
  • [2009/10/20 10:06] Zha Ewry: In practice?
  • [2009/10/20 10:07] Zha Ewry: Policy expressoin languages are a deep horrible black hole
  • [2009/10/20 10:07] Zha Ewry: I expect we'll see some agreed to string tokens, and certainly, the option for people to publish lots of information
  • [2009/10/20 10:07] Rex Cronon: it is possible that some servers will be kind of anti-social:)
  • [2009/10/20 10:07] Fleep Tuque: That could be shortened to "policy is a deep horrible black hole" ;)
  • [2009/10/20 10:07] Lillie Yifu: In the end policy needs to be presnted to the user or decisions
  • [2009/10/20 10:07] Mystical Demina: sure and they get complicated but better to query it you support something before i try to use it, than to use it and see if you fail
  • [2009/10/20 10:08] Lillie Yifu: because it's a people deciosn not a computer one
  • [2009/10/20 10:08] Zha Ewry: The history of dsitributed computing, tho, is littered with tons of annotations which could just as easily have been string matches
  • [2009/10/20 10:08] Lillie Yifu: nods so just send the text over like a EULA
  • [2009/10/20 10:08] Lillie Yifu: and let peopel decide if they are going to accept it
  • [2009/10/20 10:08] Zha Ewry: Right, or tell people what they need to do to be permitted access
  • [2009/10/20 10:09] Lillie Yifu: because really that's what it is, a terms of service
  • [2009/10/20 10:09] Zha Ewry: In a *lot* of cases, i expect that the path is more like "If you want to come to this sim, you need to go sign our TOS, and get on our list as a trusted avatar"
  • [2009/10/20 10:09] Zha Ewry: Or
  • [2009/10/20 10:09] Zha Ewry: even worse, at some level
  • [2009/10/20 10:09] Fleep Tuque: imagines having to sign a new ToS at every sim border crossing.
  • [2009/10/20 10:10] Zha Ewry: "If you want to come here, use one of the ADs which which we have a legal agreement
  • [2009/10/20 10:10] Fleep Tuque: Hm
  • [2009/10/20 10:10] Zha Ewry: Right
  • [2009/10/20 10:10] Zha Ewry: Not going to happen fleep
  • [2009/10/20 10:10] Mystical Demina: yeah, policy donesn't have to be online query, it can be text that tells you how to define you endpoint before calling into a system
  • [2009/10/20 10:10] Zha Ewry: I expect about three clusters of actual services
  • [2009/10/20 10:10] Zha Ewry: Ones which grant access to everyone in the Agent Domain who holds huge sets of avatars
  • [2009/10/20 10:10] Zha Ewry: Very open places, which simply let everyone in
  • [2009/10/20 10:11] Zha Ewry: and grumpy places who are only letting in special people, and they will, just like web pages
  • [2009/10/20 10:11] Zha Ewry: which pop up TOS screens and demand regitratoins
  • [2009/10/20 10:11] Zha Ewry: be a pain
  • [2009/10/20 10:11] Fleep Tuque: nod and hopefully some kind of aggregator service that a bunch of sims can all agree to the same policies and you agree once and move on
  • [2009/10/20 10:11] Zha Ewry: Right
  • [2009/10/20 10:11] Zha Ewry: That's the Agent Domain, broad share region domani model
  • [2009/10/20 10:12] Mystical Demina: can be lots of policies, like how many objexts you are allowed to create, how much cpu resources you can use, almost everything that can have limits or needs to be controlled
  • [2009/10/20 10:12] Zha Ewry: Allow the TOS to be signed by the person running the AD, and the promist to behave well and take down people who are bad, and then the region say "ok, my legal issue are covered' Ugly, but pragmatic
  • [2009/10/20 10:12] Zha Ewry: Right and lots of those issues, mysictal, merely require makign sure we can say
  • [2009/10/20 10:12] Zha Ewry: "Nope, you rezzed your prim quota"
  • [2009/10/20 10:12] Lillie Yifu: programmers are the unacknowledged legislators of makind
  • [2009/10/20 10:12] Zha Ewry: on the reply
  • [2009/10/20 10:13] Lillie Yifu: pace Shelley
  • [2009/10/20 10:13] Fleep Tuque: Though it is pretty common on edu sims on this grid to have a list of rules or guidelines pop up when you first TP in, I imagine you're talking about something more technically in depth
  • [2009/10/20 10:13] lonetorus Habilis: im here!
  • [2009/10/20 10:13] lonetorus Habilis: late, but im here
  • [2009/10/20 10:13] lonetorus Habilis: XD
  • [2009/10/20 10:13] lonetorus Habilis: see sai, we care :)
  • [2009/10/20 10:13] Dahlia Trimble: hiya lone
  • [2009/10/20 10:13] Rex Cronon: welcome
  • [2009/10/20 10:13] Zha Ewry: So, right Fleep, what I'd expect we'd see in a lot of those cases
  • [2009/10/20 10:13] Zha Ewry: is compacts, basedo n things liek shiboleth
  • [2009/10/20 10:13] Fleep Tuque: nod
  • [2009/10/20 10:13] Mystical Demina: yeah, i can understand the need for the policy stuff now, and it will have to let people extended polices for specialize things they are supporting
  • [2009/10/20 10:14] lonetorus Habilis: (just returned from hospital, my wife splintered her ancle in a horse riding accident this morning)
  • [2009/10/20 10:14] Dahlia Trimble:  :(
  • [2009/10/20 10:14] Fleep Tuque: (youch)
  • [2009/10/20 10:14] Zha Ewry: "I'll let in any AD which I have a shiboleth toeken I can grok" in to my sim, with the understanding they signed off on the TOS for the education domain
  • [2009/10/20 10:14] Zha Ewry: and Ouch
  • [2009/10/20 10:14] Fleep Tuque: that makes sense to me, Zha
  • [2009/10/20 10:15] Lillie Yifu: I am very sory to hear that lone, that is terrible
  • [2009/10/20 10:15] Zha Ewry: The fun bits are expressingg back to users, in wayts they understand, the consequences of policy
  • [2009/10/20 10:15] Fleep Tuque: I could certainly see educational consortia wanting to do something like that
  • [2009/10/20 10:15] Zha Ewry: (especially when it becomes an I8N issue)
  • [2009/10/20 10:15] Rex Cronon: may she get well soon:)
  • [2009/10/20 10:15] Zha Ewry: And back in the saddle
  • [2009/10/20 10:16] Zha Ewry: My tween assures me that its not gettign well, that counts, for her friends, its back on the horse
  • [2009/10/20 10:16] Zha Ewry: Oh, is I8N totaly obscure?
  • [2009/10/20 10:16] Zha Ewry: Internationalizaion
  • [2009/10/20 10:17] Zha Ewry: Which is to say, it's pretty rude of SL to say "Access to that region is denied" in english to the germans, and do on.
  • [2009/10/20 10:17] Lillie Yifu: less obscure than say Harold Thorne
  • [2009/10/20 10:17] Zha Ewry: chuckles
  • [2009/10/20 10:18] Zha Ewry: Harold Thorne shows up a ton of obscure google hits
  • [2009/10/20 10:19] Lillie Yifu: nods
  • [2009/10/20 10:19] Zha Ewry: So, what I'm going tot ry to do to get the policy stuff more cleanly into the introductroy document is try and see if we can sift between "We need this affordance to permit these sorts of policy decisions" which tends
  • [2009/10/20 10:19] Zha Ewry: towards wanting to bei n the base documents and
  • [2009/10/20 10:19] Fleep Tuque: So then, in terms of policy, will there be some sort of checklist or template to help guide folks as to what policies they need to determine? Is it a mix of technical and behavioral issues? Feel free to ignore this ? if it's stupid. :)
  • [2009/10/20 10:19] Lillie Yifu: I've actually seen a copy of his book on Contract Bridge
  • [2009/10/20 10:19] Zha Ewry: "here is how you indicate to remote servcie that you hit a policy constraint"
  • [2009/10/20 10:19] Zha Ewry: and then
  • [2009/10/20 10:19] Lillie Yifu: I'm mentioning this because I think that the policy section could have "common conventions"
  • [2009/10/20 10:20] Fleep Tuque: nod Lillie, that's kind of what I was getting at, I think
  • [2009/10/20 10:20] Lillie Yifu: to communicate concisely common policy enforcement means
  • [2009/10/20 10:20] Zha Ewry: "Here is a informaional spec, for how to do "rez limits"
  • [2009/10/20 10:20] Zha Ewry: Hmm.
  • [2009/10/20 10:20] Zha Ewry: I think that's really likely
  • [2009/10/20 10:21] Zha Ewry: I expect a *lot* of this ends up encoded in
  • [2009/10/20 10:21] Lillie Yifu: hmmmm
  • [2009/10/20 10:21] Fleep Tuque: I mean some things will have defined parameters
  • [2009/10/20 10:21] Fleep Tuque: and these are your choices
  • [2009/10/20 10:21] Lillie Yifu: if the client has a sandbox
  • [2009/10/20 10:21] Zha Ewry: shared documents, and then in horrible things liek X.509 certs which back them up
  • [2009/10/20 10:21] Lillie Yifu: it would be possible to code an echo
  • [2009/10/20 10:21] Lillie Yifu: where the region echos final state of an action
  • [2009/10/20 10:21] Lillie Yifu: and the client can compare it to expected final state based on default
  • [2009/10/20 10:22] Lillie Yifu: Use: Do this
  • [2009/10/20 10:22] Lillie Yifu: region: I did this
  • [2009/10/20 10:22] Lillie Yifu: User: compares expected do to actual this
  • [2009/10/20 10:22] Zha Ewry: That's interesting, but really hard to be sure of what would be expected
  • [2009/10/20 10:22] Lillie Yifu: because the User sandboxes this
  • [2009/10/20 10:22] Lillie Yifu: it's debugging
  • [2009/10/20 10:23] Lillie Yifu: and whether the difference is policy or something else
  • [2009/10/20 10:23] Mystical Demina: there are many ways to express these policies and may end up with one or more way, can be a list of features codes you get, could be list of interfaces that have defines methods they support, could be a list of messages you support
  • [2009/10/20 10:23] Zha Ewry: The question is can you possibly know the expected tesult in a lot of cases
  • [2009/10/20 10:23] Zha Ewry: Echo chat is easy
  • [2009/10/20 10:23] Zha Ewry: "Sit" is not
  • [2009/10/20 10:23] Lillie Yifu: in many cases no
  • [2009/10/20 10:23] Lillie Yifu: in many cases yes
  • [2009/10/20 10:23] Zha Ewry: The region may deny your sit request for lots of reasons, including someone sitting before you
  • [2009/10/20 10:23] Lillie Yifu: sit is easy
  • [2009/10/20 10:24] Lillie Yifu: but physics is not
  • [2009/10/20 10:24] Zha Ewry: and someone removign the chair before you sat
  • [2009/10/20 10:24] Zha Ewry: And physics, unless you go the Forterra (IPR warning) synced phyiscs engineapproach is really hard
  • [2009/10/20 10:25] Lillie Yifu: but some replies will be easy to parse
  • [2009/10/20 10:25] Lillie Yifu: and in any event if the user is getting unexpected results
  • [2009/10/20 10:25] Lillie Yifu: it's data
  • [2009/10/20 10:25] Zha Ewry: (IPR warning: I am aware Forterra claims IPR in the space of sycned phsyics engines as part of virtual world creation, end IPR warning)
  • [2009/10/20 10:26] Lillie Yifu: I'm thinking particularly of beign able to determine the region actively not doing something, with it not doing something within the user's time frame
  • [2009/10/20 10:26] Lillie Yifu: like say pressing the arrow key, getting no response for a long time
  • [2009/10/20 10:26] Lillie Yifu: and then having a host of moves forward come at once.
  • [2009/10/20 10:27] Zha Ewry: Worth some thought
  • [2009/10/20 10:27] Zha Ewry: It would depend, mightly, on how you defined the expected behavior of sending regoin the "I have stepped forward" message
  • [2009/10/20 10:27] Lillie Yifu: an echo of "yes I actually ave proceessed message blah blah blah" would allow users to know when something has "failed" in quotes
  • [2009/10/20 10:28] Zha Ewry: And, any type of QOS/ expected response time stuff is edge of what I expect to define here, but.. inetresting
  • [2009/10/20 10:28] Zha Ewry: right
  • [2009/10/20 10:28] Zha Ewry: In THEORY, you have only two fails in the VWRAP world
  • [2009/10/20 10:28] Zha Ewry: Caps, which don't returns after being posted
  • [2009/10/20 10:28] Zha Ewry: and messages which aren't recieved, or aren't promptly managed
  • [2009/10/20 10:29] Zha Ewry: Mind you, in pracice, int he current SL protocol, you can look at acks to see what's been done
  • [2009/10/20 10:29] Zha Ewry: rather in tehory
  • [2009/10/20 10:29] Zha Ewry: in practice some stuff falls into balck holes
  • [2009/10/20 10:29] Lillie Yifu: but since this is a synchronous or sort of synchronous in expectation environment, an action not acted on within some user time is as good as a fail.
  • [2009/10/20 10:29] Lillie Yifu: se chat lag
  • [2009/10/20 10:29] Lillie Yifu: see*
  • [2009/10/20 10:29] Zha Ewry: I've been arguing that it would be clever to define any messages, as opposed to caps as
  • [2009/10/20 10:30] Lillie Yifu: this would allow differentiation between #fail and user perception of fail.
  • [2009/10/20 10:30] Zha Ewry: things which are discardable with no loss, and keep anything where a loss is dangerous in the caps frameowrk, wih an idempotent REST like cap behavior in hand to make it clean
  • [2009/10/20 10:30] Zha Ewry: its the quasi syncronous nature of the inetrnet that makes it so hard
  • [2009/10/20 10:31] Lillie Yifu: nods
  • [2009/10/20 10:31] Zha Ewry: In my office, between my laptop and an OpenSim, I can pretty trivailly know how long it shoudl take to roundtrip a packet
  • [2009/10/20 10:31] Lillie Yifu: I am thinking about this because for user feedback
  • [2009/10/20 10:31] Lillie Yifu: it is often hard to know why an action is not producing desired results
  • [2009/10/20 10:31] Zha Ewry: At home, over wireless across the web to a Linden Simulator
  • [2009/10/20 10:31] Lillie Yifu: for example, being blcoked by a wall that has not rezed yet
  • [2009/10/20 10:31] Zha Ewry: the ping times vary from 40ms to over 1000
  • [2009/10/20 10:32] Zha Ewry: Or, in some sense, even worse, "not blocked by a wall which hasn't rezzed"
  • [2009/10/20 10:32] Zha Ewry: The one eventually becomes explicable, the other, just is "Oh, heck SL being odd"
  • [2009/10/20 10:32] Lillie Yifu: welcome to what one techie friend of mine calls "the perils of the limited verizon"
  • [2009/10/20 10:33] Lillie Yifu: yes it is common to walk forward, then hav the simulator send messages that snaps you bavck to some previous position
  • [2009/10/20 10:33] Zha Ewry: So, at the spec level, I think we can reasonably say "So, here is the expected behaviro of a servcie getting this message"
  • [2009/10/20 10:33] Zha Ewry: What's harder is to quantify the range of #fails.
  • [2009/10/20 10:33] Lillie Yifu: first capture the data
  • [2009/10/20 10:34] Mystical Demina: i would argue that this is more a of a classification of scope, and that just saying policy is too abiguouis. Policy is inplied for something to complete and conditions have to be met. but i see levels to the policies, to even know i can send a specific message is one. the message itself can have policies
  • [2009/10/20 10:34] Lillie Yifu: then start fidnign clever ways of quantifying common fails
  • [2009/10/20 10:36] Zha Ewry: I think that's likely
  • [2009/10/20 10:37] Rex Cronon: .
  • [2009/10/20 10:37] Fleep Tuque:  ;)
  • [2009/10/20 10:37] Zha Ewry: So.. Next week, my plan is to have actually posted some prose on this topic, in the context of "I think this would work in the intro"
  • [2009/10/20 10:38] Zha Ewry: and discuss it in context and in detail
  • [2009/10/20 10:38] Zha Ewry: Anyone who has words they think would be good there, is of course, strongly encouraged to pipe up onthe OGPX/VWRap mailign list
  • [2009/10/20 10:38] Zha Ewry: or in a notecard
  • [2009/10/20 10:38] Zha Ewry: or to my gmail
  • [2009/10/20 10:38] Zha Ewry: or get a swallow to bring it to me
  • [2009/10/20 10:38] Fleep Tuque: heh
  • [2009/10/20 10:40] Fleep Tuque: I'll look forward to reading that, I'm always coming from an end user/educational user perspective, so I don't really feel qualified to do much beyond give feedback.
  • [2009/10/20 10:40] Zha Ewry: I love feedback :)
  • [2009/10/20 10:40] Zha Ewry: and I even more lovereal user's use cases
  • [2009/10/20 10:40] Mystical Demina: a swallow, hehe
  • [2009/10/20 10:40] Zha Ewry: The "We need X" is incredibly helpful
  • [2009/10/20 10:40] Rex Cronon: would a pigeon work:)
  • [2009/10/20 10:40] Zha Ewry: Yes, eithe African or European
  • [2009/10/20 10:40] Fleep Tuque: nod well, I can certainly do that. :) We have lots o need. hehe
  • [2009/10/20 10:41] Fleep Tuque: Afraid I have to run to the next gig, good to see everyone though, wish I could make these more often.
  • [2009/10/20 10:41] Zha Ewry: 2:00 is our normal stop, and I'm sort of wound down, attm
  • [2009/10/20 10:42] Rex Cronon: tc
  • [2009/10/20 10:43] Wrapp Seiling: ty...
  • [2009/10/20 10:44] Lillie Yifu: beijos
Retrieved from "https://wiki.secondlife.com/w/index.php?title=AW_Groupies/Chat_Logs/AWGroupies-2009-10-20&oldid=825892"