User:Zero Linden/Office Hours/2008 August 19

[13:00] Tao Takashi: Hi
[13:00] Morgaine Dinova: Hi Tao
[13:00] Soundless Smalls: ty sai
[13:00] Soundless Smalls: wow late response, srry for that
[13:00] Lightsaber: releasing: controls
[13:01] Saijanai Kuhn: for them that is new, here's a link to previous office horus and related chatlogs: https://wiki.secondlife.com/wiki/Category:Grid_Interoperability_Chat_Logs
[13:02] Saijanai Kuhn: hey ZEro
[13:02] Soundless Smalls: hi Zero
[13:02] Morgaine Dinova: I wonder how it would look for 3D flat av's to be rendered truly in 2D in clients.
[13:02] Morgaine Dinova: 'Afternoon Zero
[13:02] Zero Linden: hello all
[13:02] Saijanai Kuhn: You'd need a particle effect that would stay in the avie location
[13:03] Morgaine Dinova: Ie. such that the av always presented the same aspect to all viewers.
[13:03] Saijanai Kuhn: like I said, a linked particle effect
[13:03] G2 Proto: hello Zero!
[13:03] Saijanai Kuhn: ah Zero, added a whole buncha links you might want to look at: https://wiki.secondlife.com/wiki/AW_Groupies#Capabilities
[13:04] Errafel Eccleston: oh like Facing sprite? You can also use a lot of flat planes with a texture, like how they do cheesy trees
[13:04] Zero Linden: I already feel scattered enough - I don't want to be rendered by a particle effet!
[13:04] G2 Proto: lol
[13:04] Free Radar: HUD v1.1 by Crystal Gadgets
[13:04] Soundless Smalls: lol
[13:04] Morgaine Dinova: Hehe
[13:04] Conover's Flight-Helper: 6.3.3 (WEAR ME!): Flight-helper is ready and operational.
[13:04] Zero Linden: recites "Through three cheese trees, three free fleas flew."
[13:04] Saijanai Kuhn: there's a language called E that incoprorates capabilities
[13:04] Saijanai Kuhn: guy named Kevin Reid wrote a MUD in it
[13:05] Zero Linden: Yes, I know the E folks (not much of a surprize, I suppose)
[13:05] Saijanai Kuhn: ah, OK. nothing new under the sun
[13:05] Bulli Schumann: Hi all, sorry bout the late showup
[13:05] Soundless Smalls: Hi Bulli
[13:05] Tao Takashi: Hi Zero and folks!
[13:06] Latha Serevi: Does anyone know if E is respectable in academia?
[13:06] Mirt Tenk: it's not something we use
[13:06] Morgaine Dinova: E's statement that you can't provide intra-process security through an API is curious ... seeing as that's exactly what Erlang does.
[13:06] Mirt Tenk: seems respectable
[13:06] Zero Linden: yes, I think is considered a fascniating area of research
[13:07] Zero Linden: I'm pretty sure I can't parse that fragment from E without further context
[13:07] Morgaine Dinova: Aye, they provided no justification for it, I agree.
[13:07] Saijanai Kuhn: I think E assumes that not all communications points cn be considered trustworthy
[13:07] Dahlia Trimble: so E is the hot language ju'jour?
[13:08] Morgaine Dinova: Nah
[13:08] Saijanai Kuhn: https://wiki.secondlife.com/wiki/AW_Groupies#Capabilities
[13:08] Saijanai Kuhn: just ended chatting with some of them in irc today
[13:08] Saijanai Kuhn: and there's a nice set of lniks via the links on the wiki
[13:08] Zero Linden: oh? who from E?
[13:08] Zero Linden: Ping?
[13:09] Saijanai Kuhn: Kevin Reid
[13:09] Morgaine Dinova: Yet another "One True Language" project. Pretty sad, from my engineering perspective.
[13:09] Dahlia Trimble: lol
[13:09] Latha Serevi: among Sai's many links related to E and the capability stuff, I found this 50-page masters thesis to be a very readable intro to "capabilities in SL", well worth half an hour -- [1]
[13:09] Saijanai Kuhn: eh, at the least, their rationale for design is worth looking at. They appear to consider a lot of issues
[13:09] Dahlia Trimble: 50 pages in 1/2 hour?
[13:10] Latha Serevi: I skim.
[13:10] Talarus Luan: Depends on how many pictures there are and the font size. :D
[13:10] Morgaine Dinova: Hehe
[13:10] Rex Cronon: hello everybody
[13:10] Dahlia Trimble: hi Rex
[13:10] Morgaine Dinova: Hi Rex
[13:11] Rex Cronon: hii
[13:11] Tao Takashi: still likes headers ;-)
[13:12] Zero Linden: that thesis doesn't actually include discussion of how we use capabilities -- oddly enough
[13:12] Zero Linden: any how - -----
[13:13] Zero Linden: Since E isn't going to be of much practical use....
[13:13] Zero Linden: Agenda Items?
[13:14] Saijanai Kuhn: dog dys of summer apparently
[13:14] Zero Linden: I see
[13:14] Bulli Schumann: LOL.. just my first meeitng
[13:14] Zero Linden: well, we could all tell avatar ghost stories....
[13:14] Bulli Schumann: More of a listener today :)
[13:14] Soundless Smalls: is here to listen
[13:15] Whump Linden: "and after the region crossing, they found a prim hook attachment on their car."
[13:15] Morgaine Dinova: Well, could give us a status report, if no agenda items
[13:15] Latha Serevi: (I should've said, w.r.t. that thesis, "intro to how we might use capability-based security to implement open grid trust, after we figure out what we want to implement."
[13:15] Saijanai Kuhn: lots of status reports
[13:15] Infinity Linden: blergh. capabilities are a way of communicating trust, not describing it
[13:15] Saijanai Kuhn: I know about python based AD, an IDL, new version of OGP draft...
[13:16] Infinity Linden: we still need a model for describing whom we trust and for what operations and under which conditions
[13:16] Latha Serevi: Infty, yes yes!
[13:16] Infinity Linden: or
[13:16] Zero Linden: feels a giant XML schema coming on...
[13:16] Morgaine Dinova: Yes, that's a very precise statement, Infinity
[13:16] Saijanai Kuhn: PEL ... ;-)
[13:17] Tao Takashi: somebody can start to write a UI for that XML schema then ;-)
[13:17] Saijanai Kuhn: thnk thats a MPEG 29 thing
[13:17] Infinity Linden: i would recommend a conceptual view first
[13:17] Tao Takashi: reminds me of the huge Zope permission screen
[13:17] Infinity Linden: 29? or 21?
[13:17] Latha Serevi: Zero, I don't think we're nearly ready to write down the right answer, much less in a finicky form like a schema. This month I'm just hoping for one or two very very rough drafts and some discussion. Infty++
[13:17] Saijanai Kuhn: hmmm thought 29 butyou may be right
[13:17] Morgaine Dinova: Well, at least the model that describes who we trust will be small ;-)))))
[13:17] Tao Takashi: yes, we should not start with the UI or XML but with a concept :)
[13:18] Tao Takashi: and see how much we actually need to define
[13:18] Infinity Linden: also... moving from concepts to "object model" to XML schema might be the way to go
[13:18] Infinity Linden: as we may simply want to use XML as a serialization format for communicating security assertions
[13:18] Infinity Linden: uh oh... i'm sounding like one of the SAML guys
[13:19] Tao Takashi: sounds all very complicated ;-)
[13:19] Saijanai Kuhn: which gets us back to OGP and pyogp. Once we're a bit further along, I can see implementing many variations for testing using a test asset server or 20
[13:19] Zero Linden: yes - actually, I agree with this - and I'm actually not advocating starting with a giant taxonomy
[13:19] Infinity Linden: we could probably start with "things tha should be trusted"
[13:19] Infinity Linden: "what they're trusted to do"
[13:20] Tao Takashi: maybe we mostly need to make sure we can move the permission scheme along maybe with some simple additions
[13:20] Infinity Linden: and "under what circumstances"
[13:20] Zero Linden: I even hope that, for the most part, we can keep as much of this as possible out of the direct spec
[13:20] Saijanai Kuhn: You'vejust set off Prokofy's alarm bells, Zero
[13:20] Morgaine Dinova: A giant taxonomy would be out of date before the file is closed in the editor.
[13:20] Whump Linden: bonus points if you can get to a Borgesan classification.
[13:21] Tao Takashi: what is allowed and not is more likely defined in contracts I think
[13:21] Tao Takashi: so that you are only allowed to connect to e.g. the SL grid if you are such a nice guys like me and this is written down in some legal form ;-)
[13:21] Zero Linden: Well, Sai - I think I would set off those alarms either way:
[13:22] Infinity Linden: there is a general belief in the PKI world that "security" and 'expressability" are often at odds
[13:22] Morgaine Dinova: If the basis for a world is a legal form, we're dead already.
[13:22] Tao Takashi: I also think there is no way around making Prok's alarm bells ring ;-)
[13:22] Morgaine Dinova: Heh
[13:22] Infinity Linden: lol
[13:22] Dahlia Trimble: lol
[13:22] Tao Takashi: but I would more concentrate on the rest of the community actually
[13:22] Morgaine Dinova: Maybe making them ring actually tells you that you're on the right track :P
[13:23] Dahlia Trimble: or totally tangential
[13:23] Saijanai Kuhn: sees a blog entry coming on to address Morgaine's remakr
[13:24] Tao Takashi: well, there will most likely be some TOS in place
[13:24] Latha Serevi: Zero, regarding keeping some of the trust stuff "out of the direct spec", do you mean that we could separate out the how-to-transmit-and-verify-authority aspects from the how-to-transfer-object-data aspects, and keep the OGP stuff focused on the latter? If they could decouple and we could describe the coupling....
[13:24] Morgaine Dinova: Going back to permission etc, I'd take the opposite approach, and define nothing except an extensible architecture for permissions. Then let evolution take its course.
[13:24] Tao Takashi: but I agree if too much laweyrism comes into a world (I remember the TM issue) it might as well break it..
[13:24] Saijanai Kuhn: I think that is the main intent, Morgaine, but een "permissions" is to specific
[13:25] Infinity Linden: Morgaine... there was a system for email that did this
[13:25] Tao Takashi: but nevertheless there will be some sort of contract between LL and RDs/ADs who want to connect I would assume
[13:25] Infinity Linden: it was called MOSS
[13:25] Infinity Linden: or Message Oriented Security Something
[13:25] Bulli Schumann: Remember the sensitive subject of ownership, though
[13:25] Infinity Linden: and there were such problems with interoperability, it was eventually abandoned
[13:25] Tao Takashi: as said at the AWGroupies meeting, I would like to not concentrate on ownership but licenses you attach to an object
[13:25] Strawberry Fride: was thinking of the other kind of MOSS for a moment there - eugh
[13:25] Infinity Linden: and replaced in general usage by S/MIME and PGP
[13:26] Tao Takashi: and having ownership more being a technical term to describe in which person's inventory an object resides
[13:26] Bulli Schumann: Cool
[13:26] Infinity Linden: the problem was, hte system was so flexible that adherence to the spec meant nothing.
[13:26] Saijanai Kuhn: I tried to expaln it to Prokofy thusly: The AWG isn't just designing "SL for the web." It's designing the [hopefully
[13:26] Tao Takashi: parts of this license might be defined by the existing permissions but some maybe is just legal text, like a CC license
[13:27] Zero Linden: What I don't want is for the OGP spec to become one of those behemouths where the interaction runs along the lines of:
[13:27] Talarus Luan: Including having extensibility for interactions which don't exist yet.
[13:27] Zero Linden: "Hello B, I'm A. Let's have a be negotiation for how we decide we are each other"
[13:27] Tao Takashi: Zero: "at this point in time"
[13:27] Zero Linden: "Okay, we agree on method X... now let's use it"
[13:27] Infinity Linden: seconds that statement
[13:28] Zero Linden: "Now we can have a big negotiation to talk about which concepts our trust domains are about...."
[13:28] Zero Linden: etc... etc.. etc..
[13:28] Morgaine Dinova: Why Zero? (Just curious) I see both pros and cons to it
[13:28] Tao Takashi: nobody would implement it I'd assume
[13:29] Saijanai Kuhn: well, could *support* it as an option, but not do much work on implementation..
[13:29] Zero Linden: Because historically it has led to very very large specifications that only a very few players can code interoperably to
[13:29] Latha Serevi: Perhaps we can get away with something minimal within OGP like the ability to send messages labeled "authentication request" or "my credentials" or "security exception, I don't know you"
[13:29] Zero Linden: and results in an "open standard" that is really just an agreement between a few large players
[13:29] Infinity Linden: it greatly increases the support burden and limits the number of people you can communicate with
[13:29] Talarus Luan: Simplicity and elegance go hand in hand.
[13:29] Zero Linden: what she said!
[13:29] Tao Takashi: it will most likely result in some alternative OGP
[13:29] Infinity Linden: *cough* CORBA *cough*
[13:30] Saijanai Kuhn: right. Call it the MS sandbox method of security and leave it totally opaque
[13:30] Zero Linden: So instead, I'd rather see if, once we understand some basic structure around security and trust
[13:30] Latif Khalifa: Infinity, I was just thinking about CORBA lol
[13:30] Tao Takashi: we probably all thought about CORBA ;-)
[13:30] Morgaine Dinova: Ah, but that would be a bad implementation then, if there's a universal spec. The *good* thing about such an approach is that there is no universal spec --- instead, all agreements are 2-party consensual, and therefore short.
[13:30] Zero Linden: if instead all OGP will have to do is define those points at which a system can simply deny access
[13:30] Zero Linden: The mechanism by which it decided to deny (or allow) needn't be part of the protocol
[13:30] Zero Linden: and IF some operator
[13:31] Zero Linden: decides that it needs a negotiation with the other party to make that determination
[13:31] Tao Takashi: the question for me as user would be with such a huge spec if I trust the spec because I don't understand it
[13:31] Zero Linden: that can happen outside the spec....
[13:31] Latha Serevi: I might add "define some way of messaging your counterpart" to the mix, Zero, so the negotiationh can start.
[13:31] Zero Linden: until such time as we realize that some particular negotiation is what everyone needs to do
[13:31] Zero Linden: for example --- authentication of remote party is one such negotiation and that should be in the spec
[13:32] Tao Takashi: there might be many external ways of checking if you can trust some component and to which extent
[13:32] Tao Takashi: like some rating service
[13:32] Tao Takashi: (if you trust that service ;-) )
[13:32] Zero Linden: OGP does defined an extensible resource system that allows two players to attempt additional, non-spec communications
[13:32] Infinity Linden: reputation systems are pretty cool, but in the past they haven't been utilized very much
[13:32] Tao Takashi: doesn't trust them
[13:33] Infinity Linden: (well. .not when mixed with public keys and so forth... yelp! _is_ somewhat popular)
[13:33] Morgaine Dinova: But there is no possible spec for "everything that everyone wants to do". There will be worlds handling and interchanging items that SL2 won't have any hope of interpreting, yet it still needs to be able to talk to such worlds in a subset fashion. So, single, universal, exhaustive specs are NOT the way to go.
[13:34] Tao Takashi: hence service discovery
[13:34] Morgaine Dinova: Yep. And part of that is negociation
[13:34] Tao Takashi: and a modular spec, which is I think where we are heading to
[13:34] Zero Linden: And RDBLs are rather popular -- or at least some are
[13:34] Talarus Luan: Actually, a metadata system with extensibility could more or less be universal.
[13:34] Latha Serevi: Is there some clever way that people design "portable into the future" specs, by providing for extension data blocks that are guaranteed to be ignored and not to break old versions of the protocol? Perhaps that what Zero means about the extensible resource system.
[13:35] Zero Linden: Latha - in the OGP spec you can ask a seed cap (from some end point you are talking to) for the capability for a function by name....
[13:35] Zero Linden: that namespace is extensible
[13:35] Talarus Luan: I deign to point out OpenGL spec's "extension" capability, as it is rather primitive on some levels and not entirely applicable here, but it demonstrates that such a thing is possible, at least.
[13:35] Morgaine Dinova: Aye, there is probably "extensible X" for all values of X. :-)
[13:35] Zero Linden: and can be extended without central authority (they are URIs)
[13:36] Infinity Linden: zero's mention of "namespace" reminds me of Knuth's comment that all computer science boils down to sorting and namespace management.
[13:37] Tao Takashi: how is the namespace actually defined? I wonder how does a service know in which namespace place_avatar is?
[13:37] Whump Linden: "naming is hard, let's go shopping!"
[13:37] Tao Takashi: do we assume the OGP namespace as long as we don't provide the full URL?
[13:37] Zero Linden: and all algorighms are table lookups and indirection?
[13:37] Morgaine Dinova: The desire for extensibility goes way back, in numerous areas. Eg. I remember the transputer's (a CPU) instruction set being extensible, which was quite impressive. And that was some time in the Early Jurassic. :P
[13:37] Zero Linden: Tao - yes - and there is a lurking annoyance there in the spec,
[13:37] Zero Linden: if the name is a relative URL, then the base URI is provided in the spec ---
[13:38] Zero Linden: and in theory an implementation should tread both the relative URL form and the full form identically
[13:38] Zero Linden: but I bet none does
[13:38] Tao Takashi: ic
[13:38] Infinity Linden: comes close to being unglued at the mention of transputers and the fear that someone out ther is working on OGP language bindings for Occam
[13:38] Morgaine Dinova: Hahaha
[13:38] Tao Takashi: well, I can implement this in our AD ;-)
[13:38] Tao Takashi: but you are right, so far I only test for name=="place_avatar"
[13:39] Zero Linden: mutters "If I had a L$1 for every time Infinity mentioned Occam or Forth....."
[13:39] Morgaine Dinova: Well, CSP will be back in vogue before long I reckon. And really Erlang is just a modern version of CSP and Occam
[13:39] Infinity Linden: or CORBA
[13:39] Infinity Linden: or Smalltalk
[13:39] Infinity Linden: blergh. pi calculus
[13:39] G2 Proto: lol
[13:39] Zero Linden: Ya know - I'm gonna put out one of those jars where you have to pay everytime you say a swear word
[13:40] Infinity Linden: FORTH is not a four letter word
[13:40] Tao Takashi: so then I wonder what the full URL will be in the final spec
[13:40] Infinity Linden: but CORBA is
[13:40] G2 Proto: lol'
[13:40] Boroondas Gupte: starts counting letters
[13:40] Tao Takashi: has a computer to do the counting ;)
[13:41] Saijanai Kuhn: My first programming boss did his masters work in astronomy writing the multiple mirror telescope OS in FORTH
[13:41] Morgaine Dinova: Zero: E will be hard to discuss then, because they have two implementations: one primary swear-penalty material (Java), and the other karma++ (LISP) :P
[13:41] G2 Proto: my first job was programming with bit switches
[13:41] Saijanai Kuhn: USAF?
[13:41] Zero Linden: E will be hard to discuss because their web site appears to be down
[13:41] G2 Proto: im robo-old
[13:41] Infinity Linden: "my favorite programming language is solder." -- Steve Ciarcia
[13:41] Morgaine Dinova: Hehehe
[13:41] G2 Proto: USN
[13:41] Talarus Luan: Is up to me: [2]
[13:42] G2 Proto: I was a radar tech soldered alot
[13:42] Saijanai Kuhn: same diff. PUts you 5 years later than the comercial version of those systems though
[13:42] G2 Proto: lol
[13:42] Tao Takashi: yeah, it seems only down for parts of the world ;-)
[13:42] Saijanai Kuhn: late 70's?
[13:42] Zero Linden: bizarre - I can't DNS that
[13:42] Rex Cronon: java has a "sear-penalty"? lol
[13:42] G2 Proto: late 80's thankfully, USN has some od gear, tubes even
[13:42] G2 Proto: old
[13:42] Opensource Obscure: zero, try [3]
[13:43] Saijanai Kuhn: taking a page from the old soviet anti-EMP systems ;-)
[13:43] Zero Linden: that one works
[13:43] Talarus Luan: Coral cache 4tw! :D
[13:43] G2 Proto: haha
[13:43] Infinity Linden: ugh. it's only a matter of time before someone mentions ADA
[13:43] Saijanai Kuhn: ether that or just too cheap to upgrade teh contracts
[13:43] Infinity Linden: d'oh!
[13:43] Zero Linden: mind you, I've read the whole site a while ago
[13:43] Zero Linden: and I know Ping
[13:43] G2 Proto: until you have burnt your hand fixing tubes you have not lived
[13:43] Infinity Linden: Is Ping Yi doing erights?
[13:43] Latif Khalifa: i still use fortran libs daily
[13:43] Infinity Linden: i thought it was Miller and Szabo
[13:43] Latif Khalifa: FFT
[13:43] G2 Proto: haha
[13:44] Saijanai Kuhn: Kevin Read seems to be maintaining the page, but not sure
[13:44] G2 Proto: Fast Fourier Transform?
[13:44] Saijanai Kuhn: Reid
[13:44] Latif Khalifa: G2, yup
[13:44] Infinity Linden: haven't seen those guys since the Foresight Institute days
[13:44] G2 Proto: used it alot as an aerospace engineer to measure VSWR and Insertion Loss
[13:44] Saijanai Kuhn: INifnity [4]
[13:44] Zero Linden: KaPing Yee is one of the active contributors to the Open E Project
[13:45] Latha Serevi: You could try [5] instead of erights.org, LL folks, and see if you can get around the DNS issue...
[13:45] Morgaine Dinova: Heh. In the current lawyer-ridden environment, if an EMP went off, someone would probably try to collect royalties for each party affected.
[13:45] G2 Proto: lol
[13:45] Saijanai Kuhn: they own the patent to it/
[13:45] Latif Khalifa: G2, we used it these days for analyzing output of gene sequencing machines
[13:45] G2 Proto: Electro Magnetic Pulse?
[13:45] Infinity Linden: hmm.. but the sponsorship opportunities are incredible...
[13:45] G2 Proto: wow Latif cool!
[13:45] Morgaine Dinova: Aye
[13:45] Saijanai Kuhn: wottya mean, my son owes royalties for his DNA???
[13:46] G2 Proto: YOU HAVE TO PUT YOU COMM GEAR IN BATTLE MODE IN CASE OF emp
[13:46] Infinity Linden: "this nuclear holocaust brought to you by the good people of Mutual of Omaha"
[13:46] G2 Proto: LOCKS THE RELAYS IN
[13:46] G2 Proto: oops
[13:46] G2 Proto: sorry
[13:46] Strawberry Fride: it's the rain :)
[13:46] Talarus Luan: Caps locks, too
[13:46] G2 Proto: yes im about to sink, really
[13:46] G2 Proto: river is almost to the house
[13:46] Morgaine Dinova: EMP got his CapsLock ;-)
[13:46] Strawberry Fride: kayaks at the ready, from what I hear :)
[13:46] G2 Proto: lol
[13:46] G2 Proto: they literally are
[13:46] G2 Proto: its that bad
[13:47] Saijanai Kuhn: ykes
[13:47] G2 Proto: and round 2 is coming lol, had to move the car down the road we are flooded
[13:47] G2 Proto: if i get chopper rescued look for the guy holding 3 desktops, a couple of cats and 5 lcds
[13:47] Strawberry Fride: lol!
[13:48] G2 Proto: hehe
[13:48] Strawberry Fride: is gonna be glued to the telly now
[13:48] Zero Linden: So..... steering back somewhere on the ground
[13:48] Morgaine Dinova: How about a progress update on PyOGP then, if no other AWG topics?
[13:48] Saijanai Kuhn: right arm
[13:48] G2 Proto: oh sorry, ill survive im sure
[13:49] Saijanai Kuhn: we got ininity and tao here
[13:49] Saijanai Kuhn: infinity*
[13:49] Tao Takashi: pyogp is in the middle of some refactorting right now
[13:49] Saijanai Kuhn: I'm just working on an external GUI for controlling the tests
[13:49] Tao Takashi: we are looking forward to make some release when draft 3 with the auth changes is out
[13:49] Infinity Linden: and i'm in the middle of madly editing specs so we can get them reviewed internally so they can be released for external review
[13:49] Tao Takashi: and some cleanups have been made
[13:49] Dahlia Trimble: tk Saij?
[13:50] Saijanai Kuhn: right, universal install. Remarkably like the SL viewer GUI for some reason
[13:50] Morgaine Dinova: What was the state of play prior to refactoring?
[13:50] Saijanai Kuhn: all essential bits working, just not necessarily at the same time
[13:50] Dahlia Trimble: lol
[13:50] Saijanai Kuhn: OGP login and TP, UDP/caps messaging
[13:50] G2 Proto: haha
[13:50] Morgaine Dinova: Sounds funny, but yeah, that's cool
[13:51] Saijanai Kuhn: AD and sims presence
[13:51] Tao Takashi: yes, it was basically working
[13:51] G2 Proto: awesome
[13:51] Tao Takashi: but we are looking now on how people would use it
[13:51] Tao Takashi: and make sure it's compatible with different types of e.g. networkign libs
[13:51] Tao Takashi: like eventlet or twisted or threads
[13:51] Tao Takashi: so it's not too much predefined by the lib
[13:51] Saijanai Kuhn: and I want to shoehorn in a legacy (XMLRPC) login option so it can be used as a lightweight client on the regular grid
[13:52] Tao Takashi: it's supporting creating and serialization of messages and theother way round as well
[13:52] Tao Takashi: and managing circuits
[13:52] Latif Khalifa: Saijanai, you could use legacy LLSD as well ;)
[13:52] Saijanai Kuhn: true. same diff
[13:53] Saijanai Kuhn: rpobably better to use that only
[13:53] Tao Takashi: and I also checked in the Agent Domain part which is now online at [6]
[13:53] Tao Takashi: this needs some review, docs and tests though
[13:53] Tao Takashi: and some additions like supporting TP ;-)
[13:53] Infinity Linden: doesn't consider XMLRPC to be "lightweight"
[13:54] Saijanai Kuhn: well, its part of standard pythono distro so
[13:54] Zero Linden: is it really?>
[13:54] Tao Takashi: yep
[13:55] Strawberry Fride: xmlrpc is much lighter than something like soap
[13:55] Infinity Linden: hmm... section 18.24: xmlrpclib
[13:55] Zero Linden: funny- I don't think there are any maintained C libs for it
[13:55] Saijanai Kuhn: was going to say, I didn't even know how to install new modules when I wrote first version of python login so
[13:55] Tao Takashi: I am not saying it's maintained.. ;-)
[13:55] Tao Takashi: but maybe it also does not need to be maintained
[13:55] Latif Khalifa: Zero, did you change the license of LLSD c++ lib yet? ;)
[13:56] Tao Takashi: and the question might be if it's still part of the lib in Python 3000
[13:56] Saijanai Kuhn: I found at least one bug in it. Randomly sends non xml stuff in header
[13:56] Tao Takashi: Guido wants to think about what's in the stdlib and what not as usually it's problematic to have things in the stdlib
[13:56] Zero Linden: Grrrrr... no I haven't yet gotten that out
[13:56] Zero Linden: fie on me
[13:56] Zero Linden: a million poxes
[13:56] G2 Proto: lol
[13:56] Latif Khalifa: lmao
[13:56] Saijanai Kuhn: shies away slightly
[13:56] Latif Khalifa: any year now ;)
[13:57] Latha Serevi: I'm wondering how we're doing on defining what an AD does and how it can either host my assets or delegate to some other asset server. Tao, what is Syntronik about anyhow, and based on what spec?
[13:57] Dahlia Trimble: I just had an encounter with LLSD and python and it isnt pretty :(
[13:57] Saijanai Kuhn: I think the indra stuff is non-GPL licensed dahlia
[13:58] Infinity Linden: Latha... btw... that might be out of scope for OGP... thinking that OGP is about communicating between "viewer" and "agent domain"
[13:58] Morgaine Dinova: Wasn't xmlrpc one of those things always giving trouble because of incompatible library proliferation? I seem to remember we use -epi, which isn't compatible with the -c or others.
[13:58] Saijanai Kuhn: we already talked about that sorry
[13:58] Dahlia Trimble: it's working now...
[13:58] Infinity Linden: some of what you're discussing MAY be a private implementation issue
[13:58] Infinity Linden: btu
[13:58] Infinity Linden: that being said
[13:58] Tao Takashi: syntronik.de is an Agent Domain which does not much except letting you register an account and handling login and place_avatar
[13:58] Tao Takashi: it's based on the OGP spec plus what needed to be found out by using tcpdump and asking Sai
[13:58] Infinity Linden: it would be great if we could find some standard protocol for it
[13:59] Saijanai Kuhn: Infinity, I think that AD to RD and AD to asset server pretty much tie into AWG work
[14:00] Saijanai Kuhn: It might not be strictly "OGP" but its essential to getting it to work
[14:00] Infinity Linden: oh oh... i was thinking we were talking about having "composite" services where the functions of the AD were implemented with a "cluster" of different machines
[14:00] Saijanai Kuhn: wow major local chat lag
[14:00] Zero Linden: (one the xmlrpc front - this might stem from the somewhat moving target that the xmlrpc spec was for quite some time)
[14:00] Tao Takashi: the next thing I wanted to try with it is putting some XRDS document inside and play around with an pyogp client to do service discovery with that
[14:00] Infinity Linden: and technically.. how an AD represents inventory is not as important to the spec as how it communicates what it's got and what you're trying to do with it
[14:00] Tao Takashi: so the simplest thing would be that you don't need to know the exact AD login URI but you simply enter syntronik.de and the XRDS discovery will find it for you
[14:01] Infinity Linden: XRDS being a great example of things that are outside teh scope of OGP authentication
[14:01] Tao Takashi: yes, to the spec only the API is important
[14:02] Zero Linden: this is the LLSD python stuff, and it is MIT licensed: [7]
[14:02] Infinity Linden: but you definitely want to make sure you don't "specify yourself into a corner" htat disallows several popular options
[14:02] Saijanai Kuhn: thought so, thanks zero
[14:02] Dahlia Trimble: ty Zero :)
[14:02] Tao Takashi: depends, maybe you want to discover which services an AD implements, like what version of inventory, what group services etc. then you might want some discovery
[14:02] Latha Serevi: I'm wondering, though, whether the AD will pass thru all info like a proxy, or be able to hand-off tasks via URI's.
[14:02] Zero Linden: that all said, I've wondered if we should put a "discovery" method into the seed-cap - since it bascially can answer that question easily
[14:03] Tao Takashi: I wonder if the client cannot directly communicate with such services than instead of proxying everything
[14:03] Saijanai Kuhn: that might be implementaton dependent, but I think we need to design some standard protocols to talk to external systems (used or not)
[14:03] Infinity Linden: so while i think service discovery is outside the scope of the auth spec, i wouldn't want to make something that couldn't be used with XRDS
[14:03] Infinity Linden: or, precludes it's use in addition to the auth spec
[14:03] Zero Linden: well all
[14:03] Zero Linden: it is 2pm and I've got to run
[14:03] Infinity Linden: ditto
[14:03] Zero Linden: (to go review spec. docs with Tess and Infinity!)
[14:03] Morgaine Dinova: This is off-topic to infrastructure, but seeing the link off Slashdot and the video made me think of our avatars and SL machinima. Interesting times ahead: [8]
[14:03] Saijanai Kuhn: hooray
[14:03] Zero Linden: thanks for all coming
[14:03] Rex Cronon: bye zero, linden
[14:03] Tao Takashi: also each user might have it's own XRDS spec describing which service the user has
[14:03] Zero Linden: later

User:Zero Linden/Office Hours/2008 August 19

Navigation menu

Search