AW Groupies/Chat Logs/AWGroupies-2008-10-21

From Second Life Wiki
Jump to: navigation, search
  • [9:34] Saijanai Kuhn: wow. big crowd
  • [9:34] Latha Serevi: Hi, Sai. Never seen it so dead.
  • [9:35] Saijanai Kuhn: I'm surprised I'm here
  • [9:36] Zha Ewry: checks for crickets
  • [9:37] Zha Ewry: We are meeting, in theort, but, mighty crickety here
  • [9:39] Saijanai Kuhn: well, I was behind on my spam. still trying to get UDP dictionary for pyogp right.
  • [9:39] Saijanai Kuhn:  :-(
  • [9:39] Latha Serevi: So... my topic areas are a bit narrow; group management and non-LL-controlled spec writing. Zha, anything I could help you with in those narrow areas? :-)
  • [9:40] Zha Ewry: smiles
  • [9:40] Zha Ewry: OK
  • [9:40] Zha Ewry: So.. the later is clearly interesting
  • [9:41] Zha Ewry: At the moment, the entirely of spec writing ha sbeen done by Linden
  • [9:41] Zha Ewry: They take some feedback and input
  • [9:41] Zha Ewry: but at the end of the day
  • [9:41] Zha Ewry: They've done all the actual "Here is spec" writing
  • [9:41] Zha Ewry: Which given thiergoals is fine
  • [9:41] Zha Ewry: But.. it doesnt' server the OpenSim or broader coumminbity at all
  • [9:42] Zha Ewry: or, not as well as it would if we had a broader approahc
  • [9:42] Latha Serevi: Had a chat with Baba Yamamoto who runs openmv.org; he'd be willing to host a credible spec'ing effort, but he struck me as pretty vague and idealistic, and wanting to generalize way away from SL.
  • [9:42] Latha Serevi: More promising perhaps might be to hit up the opensim-dev list and see if anyone there is craving a spec to write towards / share.
  • [9:43] Zha Ewry: well, in the long term, I think that's the goal, being somewhat more general
  • [9:43] Latha Serevi: They're at least intensely specific and do-it-now.
  • [9:43] Zha Ewry: but, I tend to cast it as very much super set
  • [9:43] Zha Ewry: not.. away from
  • [9:43] Zha Ewry: and, end of t day, the problem with away from is the client, and will be as long as the client is essentially a bottleneck
  • [9:44] Latha Serevi: good points.
  • [9:44] Latha Serevi: Hey, what about this possibly wrong-headed idea of mine: adapters.
  • [9:44] Zha Ewry: In what sense?
  • [9:44] Saijanai Kuhn: seems to me that the AD is an adapter
  • [9:45] Zha Ewry: Sort of a facade in a lot of ways
  • [9:45] Saijanai Kuhn: so is the code that Zha wrote for OpenSim, and likely the code for the SL sims too
  • [9:45] Latha Serevi: What if I had an adapter that could take 3-d input from some other 3-d world, and munge it into SL form to run on the conv entional SL client? It would be a (performance-dragging) way to establish connectivity.
  • [9:45] Zha Ewry: Ahmmm.
  • [9:45] Zha Ewry: Hmmm.
  • [9:45] Zha Ewry: The problem with the client, at the moment
  • [9:45] Saijanai Kuhn: well, could have an asset server do that once per asset...
  • [9:45] Zha Ewry: in a lot of ways is that irs not a stand along client, at all, but very co-dependent
  • [9:46] Zha Ewry: There is a *lot* as in *scary* amount of coupling in odd places
  • [9:46] Latha Serevi: My mental image was more at the sim level -- "region wants to show scene to incompatible viewer"
  • [9:46] Zha Ewry: Some of which has really desirable effetcs
  • [9:46] Saijanai Kuhn: in fact, the obj translator people want would be such an adapter
  • [9:47] Latha Serevi: Perhaps I shouldn't have mentioned the SL viewer; it might be too hairy for this kind of approach.
  • [9:47] Zha Ewry: Not for real time, tho
  • [9:47] Zha Ewry: (The object translator)
  • [9:47] Zha Ewry: Hmm.
  • [9:47] Zha Ewry: That is less painful, Latha although
  • [9:47] Latha Serevi: Well, I'm thinking "real time" as in a frame every few seconds so I can visit a foreign sim, and basically establish a target for later compatible-viewer efforts.
  • [9:48] Zha Ewry: Then.. the challange becomes that the Sims are sort of equally oddly tangled with the client, and expect it to do some odd things
  • [9:48] Zha Ewry: At a .5 FPS level, you can almost just
  • [9:48] Zha Ewry: grab a jpg and render it
  • [9:48] Zha Ewry: Of course,t hat's hardly what we want to do
  • [9:48] Latha Serevi: Yah, but I was trying to get at the underlying commonality of a 3-d representation.
  • [9:48] Zha Ewry: nods
  • [9:49] Zha Ewry: right and part of the problem there
  • [9:49] Zha Ewry: is that Linden's constructive prim geometry is pretty wonky next to 90% of the worlds meshy stuff
  • [9:49] Zha Ewry: Collada is *just* starting to get someplace on being able to handle htings like real constructive prims
  • [9:50] Latha Serevi: We've got a few levels at which we can try to establish lingua franca -- agent existence, IM, "place", protocols, 3-d objects, 3-d scenegraph ... I don't have a solid handle, but wish we had a short list of convincing ways-to-interact-via-a-general-protocol that we're tyring to develop.
  • [9:51] Zha Ewry: nods
  • [9:51] BigMike Bukowski: I think group chat might be dead Saij.
  • [9:51] Saijanai Kuhn: I wish we could get the openviwer and realxtend people more involved in this aspect
  • [9:51] Latha Serevi: The abstract stuff non-3d is where we're focusing, rightly, at the moment with OGP ... but the 3-d is the inspirational stuff.
  • [9:51] Zha Ewry: It is, Latha, and that is a good point
  • [9:52] Zha Ewry: One thing I'd love to see, just as a proof of concept, if nothign else
  • [9:52] Zha Ewry: would be a totally new prim type, added to both the client and sim side, for OpenSim, to show that we can actually get seperate from the pure Linden development pipe
  • [9:52] Zha Ewry: A high fidelity meshed prim, for example, with <1024 vertices
  • [9:52] Saijanai Kuhn: gareth has his gpl viwer on a webpage working at about .1fps, btw
  • [9:53] Zha Ewry: nods and how many CAD leters per week?
  • [9:53] Zha Ewry: *letters
  • [9:53] Zha Ewry: or am I being too cynical?
  • [9:53] Saijanai Kuhn: not sure I follow
  • [9:53] Saijanai Kuhn: its a SL webcam basically
  • [9:54] Zha Ewry: I'm sort of joking buit C&D lstters seem to follow gareth
  • [9:54] Saijanai Kuhn: oh, right
  • [9:54] Latha Serevi: Sai, render-to-web is a good example of a minimal kind of adapter -- 3-d to 2-d rendering -- that we might want to keep in mind as a very useful compatibility level, and try defining a protocol for ...
  • [9:54] Saijanai Kuhn: yeah. He's a borderline autistic savant I think and doesn't quite know where to draw the line
  • [9:54] Latha Serevi: "2-d window onto incompatible 3-d world xyz"
  • [9:54] Zha Ewry: Not entirely incompatible 3-d of course, but yes
  • [9:56] Zha Ewry: One thing we're just startintg to see modled at all in any of the dicussion is a good way of caturing the sim-> client interactoin as a web rather than deeply obscure UPD pipe model
  • [9:56] Latha Serevi: ... put up a video feed from WoW, and a reverse direction feed, and get a minimal kind of connectivity w/o solving 3-d problems directly ...
  • [9:56] Latha Serevi: Oh ... Zha, you're saying, how to control an avatar from a browser?
  • [9:57] Latha Serevi: Well, ahem, adapter proxy?
  • [9:58] Zha Ewry: well a full model for taking about the connection
  • [9:58] Zha Ewry: right now, its not at all webish
  • [9:58] Zha Ewry: its "Here is a magic stream of UDP packets you can parse"
  • [9:58] Saijanai Kuhn: the agentmovement packet isn't that obscure though its a bit cumbersome to compose on-the-fly
  • [9:59] Zha Ewry: but there is *no* sense of a model thhat describe it in reasonable terms
  • [9:59] Zha Ewry: its all
  • [9:59] Zha Ewry: Parse a packet
  • [9:59] Zha Ewry: not "here are resoruces, which get updated" or anything close
  • [9:59] Saijanai Kuhn: I'd like to get an events handler into pyogp that accepts GUI/commandfile input and turns it into portions of that packet
  • [10:00] Latha Serevi: can't imagine web interface without a proxy listening for lots of packets, accumulating a viewer-ish local model, and providing web-ish resources. Make sense? Alternatives?
  • [10:00] Zha Ewry: In practice, that's how Linden codes it
  • [10:00] Zha Ewry: Its a endpoint you throw the updates at
  • [10:00] Zha Ewry: and then the client pulls from it
  • [10:00] Zha Ewry: The problem is
  • [10:00] Saijanai Kuhn: right. BUT, for end-user coding, its a bitch to work with
  • [10:00] Latha Serevi: That's pretty much the libsecondlife model, too, isn't it.
  • [10:00] Zha Ewry: that's really hopeless to model
  • [10:00] Zha Ewry: and, pretty much
  • [10:00] Zha Ewry: What you'd like
  • [10:01] Zha Ewry: is a clear set of resources on the sim, for the client to build an interest list
  • [10:01] Rex Cronon: hello
  • [10:01] Saijanai Kuhn: hey rex
  • [10:01] Zha Ewry: and then model where the sim publishes to an endpoint, which would be responsible for mdeiating
  • [10:01] Latha Serevi: Ooh, I'm hearing you, Zha. Ambitious but good.
  • [10:01] Rex Cronon: hi
  • [10:01] Tao Takashi: Hi
  • [10:01] Zha Ewry: if you're not a peer
  • [10:01] Zha Ewry: (if you are a peer, you don't need the proxy)
  • [10:02] Saijanai Kuhn: is getting Locklainn's UDP dictionary parser to work for the missing packets today, so this is quite serundipitous (sp)
  • [10:02] Latha Serevi: Perhaps we could define those resources and interests, and be able to provide them from either a hacked viewer (the proxy approach), or, later, from a hacked server (the interest list / services approach).
  • [10:03] Latha Serevi: Wups, lost Zha for now.
  • [10:03] Saijanai Kuhn: not sure where they belong right now. Agent movement stuff is all in one packet, and the interest list for the servers isn't defined anywhere on the wiki
  • [10:04] Tao Takashi: what's the topic for today? sorry, I missed the beginning
  • [10:04] Saijanai Kuhn: and not sure how they should be factored out/together
  • [10:04] Rex Cronon: there might be problems on the grid righ now, when i tried to log in it said that "www.secondlife.com" doesn't exist
  • [10:04] Saijanai Kuhn: non-LL spec writing for ogp is sorta the topic
  • [10:05] Latha Serevi: Tao, I brought up "non-Linden spec writing", and we've been talking generally about how to define open grid protocols via adaptyers; 3-d to 2-d web stuff; communications models (flurry of packets vs. services and subscriptions)...
  • [10:05] Audoa Giha: thats not good
  • [10:05] Tao Takashi: ic
  • [10:05] Saijanai Kuhn: to summarize: Lindens are writing everything and OpenSim isn't contributing to the OGP spec [though they might have stuff they are doing on their own -ed]
  • [10:05] Tao Takashi: I started spec writing here: [1] ;-)
  • [10:06] Latha Serevi: Any interest in being the nexus of a non-LL OGP cabal, Tao?
  • [10:06] Tao Takashi: I think the problem is usually that spec writing is not for everybody. you know coders and documentation ;-)
  • [10:07] BigMike Bukowski: I would if I wasn't so busy.
  • [10:07] Saijanai Kuhn: is lousy at both :-/
  • [10:07] Latha Serevi: To me, the key for success would be getting some fraction of the active coders excited about tehe prospect of using a spec that other people are likely to use.
  • [10:07] Zha Ewry: staggers back after a total crash
  • [10:07] Saijanai Kuhn: grats
  • [10:07] Tao Takashi: well, I should make clear that this alternative version is really just a thinking experiment
  • [10:07] Rex Cronon: wb
  • [10:07] Tao Takashi: wb Zha
  • [10:08] Tao Takashi: not really adding to OGP but trying to make it more using things used on the web today
  • [10:08] Zha Ewry: nods
  • [10:08] Latha Serevi: Zha, we were just getting Tao up to speed on our topic today and he mentioned his alterna-OGP webpage.
  • [10:08] Tao Takashi: and bringing it together with Data Portability goals.
  • [10:08] Zha Ewry: I think one of the fair questoins, for that effort Tao, is whether you *need* skin/facade existing standards, or you can just plug them. I lean to the former, but I'm not 100% sure its true
  • [10:09] Tao Takashi: I don't get this sentence ;-) what do you mean?
  • [10:09] Zha Ewry: OK
  • [10:09] Tao Takashi: what does skin/facade mean?
  • [10:09] Zha Ewry: Do you need to put a skin/facade in front of
  • [10:09] Zha Ewry: for example Jabber, or OpenId
  • [10:09] Zha Ewry: Whihc is another way of asking how deeply all the bits are co-dependent
  • [10:09] Zha Ewry: The example, which comes to mind to drive that alittle
  • [10:09] Tao Takashi: well ,it depends on how "secure" this service is going to be
  • [10:10] Tao Takashi: if you only want properly identified AD agents on there then you need some way of checking
  • [10:10] Zha Ewry: is how easily can you pass an inventory item in IM, if you just use Jabber, and how likely are you to be able to tell it is really one on the recieving end
  • [10:10] Tao Takashi: if you don't care but simply reuse e.g. an IRC users with other usages, then you don't need all that
  • [10:10] Tao Takashi: how can I pass inventory right now, btw?
  • [10:10] Tao Takashi: with IM
  • [10:11] Tao Takashi: I always here this but honestly IM transfer should be separate
  • [10:11] Zha Ewry: done under the covers with notices
  • [10:11] Zha Ewry: And. you need a way to do it, in general, when you aren't adfjancet
  • [10:11] Tao Takashi: yes, but IMHO that's not IM to me, that's a different protocol
  • [10:11] Zha Ewry: when I drop an item on your profile
  • [10:11] Zha Ewry: I'm doing that
  • [10:11] Zha Ewry: and, often, I want to do it inline with chat
  • [10:11] Zha Ewry: we do it all the time in text chat, like Yahoo and such
  • [10:11] Zha Ewry: "here's a photo"
  • [10:12] Tao Takashi: of course if you want to give somebody in a group IM some object you need to be able to identify this person and it's AD url to get some service URL for dropping that object
  • [10:12] Zha Ewry: "Here's snippet of XML"
  • [10:12] Zha Ewry: and so on
  • [10:12] Tao Takashi: yes, but passing XML additionally in IM shouldn't be the problem.. your client inside the viewer can decode it
  • [10:12] Zha Ewry: But, you need a clean way of konwing where to look for it in the client
  • [10:12] Tao Takashi: and I am mainly talking about group IM right now, btw
  • [10:12] Tao Takashi: you mean where to get the object?
  • [10:13] Zha Ewry: You don't really want to parse every line of IM for possibel XML which may be in there
  • [10:13] Tao Takashi: isn't jabber XML anyway? ;)
  • [10:13] Zha Ewry: But, not nested, and we will not discus the pain of nesting in XML unless you want to see me wimper a lot
  • [10:13] Tao Takashi: but checking for some markup in an IM line shouldn't be too much effort for the viewer
  • [10:14] Tao Takashi: I assume libraries will do proper encoding of it :)
  • [10:14] Tao Takashi: that's also the reason why I'd rather use existing libraries/servers instead of writing my own
  • [10:14] Zha Ewry: Sure, but now, you're actually layeing two specs, right, and hoping the client can find the XML in the packets
  • [10:14] Zha Ewry: (and XML nesting, is messy, even with libraries, because XML is messy)
  • [10:14] Tao Takashi: right, I am putting a spec on top, but that should be an easy spec
  • [10:15] Tao Takashi: well, I can type here <mytag attr="2" />, so at least SL IM seems to encode it properly and I assume I can do the same in a jabber client
  • [10:15] Zha Ewry: You can
  • [10:15] Tao Takashi: but it will raise some issues of course, like people then being able to type it manually
  • [10:15] Zha Ewry: but when you go to parse the IM, its just plaintext
  • [10:15] Zha Ewry: and
  • [10:15] Zha Ewry: yeah
  • [10:16] Tao Takashi: then again I am not against maybe extending standards and for XMPP there seem to be some around already (extensions that is)
  • [10:16] Zha Ewry: you really would prefer not to have people being able to sneak those in like that, tho.. there are ways to avoid that
  • [10:16] Tao Takashi: but even with extending it you would have a starting point and you might be able to use unextended service as well automatically
  • [10:17] Tao Takashi: you can maybe do service discovery in front to see what they support
  • [10:17] Tao Takashi: we probably need to push the boundaries in some areas simply because we are doing something new
  • [10:18] Tao Takashi: but in general I would say, that the AD really is mainly yet another social network but maybe one which has sharing automatically build in
  • [10:18] Tao Takashi: as one goal seems to be to move only one agent from world to world
  • [10:18] Tao Takashi: not like today's SNs where you need to have an agent on each
  • [10:18] Zha Ewry: I think that's key, Tao
  • [10:18] Zha Ewry: That its 90% the same old same old
  • [10:18] Zha Ewry: and its even worse
  • [10:18] Zha Ewry: because in fact
  • [10:19] Zha Ewry: what we see, all the time is people using all of the SN tools in conjunction with SL
  • [10:19] Tao Takashi: so I see a chance here to directly solve the Data portability problem aswell.. but for that we need at leas somewhat established standards
  • [10:19] Zha Ewry: Yeps
  • [10:19] Tao Takashi: yes, people are also very active in bridging them in
  • [10:19] Zha Ewry: right, and LL so far, is not showing much interest
  • [10:19] Tao Takashi: while it would be so much easier if there simply would be some web services available for sharing it
  • [10:19] Tao Takashi: instead of going via e.g. LSL or somesuch
  • [10:20] Tao Takashi: yes, their web service story is not really big
  • [10:20] Zha Ewry: I'd love to see the profile, and such, and some of the IM and state info sharing, just go to things like plurk and twitter and the like
  • [10:20] Tao Takashi: yep, just adding some OpenSocial REST API to secondlife.com would be great
  • [10:20] Zha Ewry: nods
  • [10:20] Tao Takashi: even if it does not yet support everything you can support with 3D, the basics would be great
  • [10:21] Zha Ewry: it might be woth actually writing that down for a topic with Zero
  • [10:21] Zha Ewry: because it would bevery interesting to see how they respond
  • [10:21] Tao Takashi: I have the feeling this Linden Lab these days is really not very much different to facebook etc.
  • [10:21] Tao Takashi: yes, that would be interesting
  • [10:22] Tao Takashi: as well as discussing these discussions about whether LL is really planning on other ADs except their own
  • [10:22] Zha Ewry: They have always said they are
  • [10:22] Zha Ewry: but it isn't entirely clear they aregoing to act on that
  • [10:22] Tao Takashi: well, people are spreading some rumours ;-)
  • [10:22] Tao Takashi: based on blurry explanations somewhere in the knowledge base
  • [10:23] Tao Takashi: so what I am trying to do as time permits is also experimenting with implementing something like I have written down in that spec
  • [10:23] Tao Takashi: not for 3D stuff as this would be quite complex but maybe IM
  • [10:24] Tao Takashi: a proof of concept is always good
  • [10:24] Tao Takashi: already played around with OAuth
  • [10:24] Tao Takashi: and maybe one don't even would need to differ between an IM service which wants to connect or a region
  • [10:25] Tao Takashi: might be the same or at least a very similar process
  • [10:25] Tao Takashi: region might have more issues though like sim crossing
  • [10:25] Zha Ewry: OAugh with some minor bending of the rules
  • [10:25] Zha Ewry: I don't really see OAuth with the
  • [10:25] Zha Ewry: "Ask every time you pass a key on" model working
  • [10:26] Tao Takashi: right, and I don't ask everytime
  • [10:26] Zha Ewry: There is *some* OAuht discussion going on about a standard way of bypassing that requirement
  • [10:26] Tao Takashi: where is this happening?
  • [10:27] Zha Ewry: I'll ask. Soemoen in my building mentioned it last time I complained loudly about OAuth
  • [10:27] Tao Takashi: should really live in SF, there are all the time some *camps around such topics.. ;-)
  • [10:27] Zha Ewry: "Yeah, we know, and there's been some discussion about that"
  • [10:27] Zha Ewry: chuckles
  • [10:27] Zha Ewry: or NY
  • [10:27] Zha Ewry: We manage it too
  • [10:27] Zha Ewry: The other one to think about, a little Tao, would be SIP for presence
  • [10:28] Tao Takashi: but what requirement exactly? I don't get the ask everytime thing
  • [10:28] Tao Takashi: maybe, no idea how complicated SIP is
  • [10:28] Zha Ewry: Well, the OAuth notion is that it is a low use thing
  • [10:28] Tao Takashi: yes, you mean the ask every time thing with Caps I assume
  • [10:28] Zha Ewry: You use it to establish you gmail->twitter linkage once
  • [10:28] Zha Ewry: and then..
  • [10:28] Zha Ewry: you say, ok, that's set
  • [10:28] Zha Ewry: But, in the VW case
  • [10:29] Tao Takashi: but I wonder if there is really a need to ask all the time
  • [10:29] Zha Ewry: you're goign to want to generate secrets betwene tons of components
  • [10:29] Tao Takashi: just add several permissions to that access token
  • [10:29] Zha Ewry: do.. you probably want to do it much less heavily
  • [10:29] Zha Ewry: And. not clear you exactly want to do that
  • [10:29] Zha Ewry: becaus eif you add too many
  • [10:29] Tao Takashi: probably, depends on the components
  • [10:29] Zha Ewry: then people end up having too broad a key
  • [10:29] Zha Ewry: Which is the Oauth bind
  • [10:29] Zha Ewry: When you pass the key beyond the single pair
  • [10:30] Zha Ewry: it now unlocks too much
  • [10:30] Zha Ewry: "This was for twittter/gmail"
  • [10:30] Tao Takashi: well, right now I don't know too much about the protocol like the 3d stuff and object transfers. I will think about it once I learn more about it :)
  • [10:30] Zha Ewry: food for thought
  • [10:30] Zha Ewry: I have alas, a stack of RL meetings today, so I'm goign to call 1 hour of this enough
  • [10:30] Tao Takashi: I also have some work actually ;-)
  • [10:30] Zha Ewry: But, I think that raising OpenSocial with Linden should be interesting and worthwhile
  • [10:31] Tao Takashi: well, at least the REST API
  • [10:31] Tao Takashi: can't be too hard to code
  • [10:31] Tao Takashi: and they have some stuff in world.secondlife.com already
  • [10:31] Tao Takashi: it might be nice to pair this actually with OAuth, like myspace did
  • [10:31] Tao Takashi: so it's not directly public to everybody
  • [10:32] Tao Takashi: is there some Zero meeting today?
  • [10:34] Rex Cronon: is it?
  • [10:34] Tao Takashi: I don't know, just asking ;-)
  • [10:34] Latha Serevi: barks and runs away. Bye!
  • [10:34] Rex Cronon: bye
  • [10:34] Tao Takashi: btw, I also incorporated some comment widget in my alternative spec, so that might be worthwhile for OGP, too maybe
  • [10:34] Rex Cronon: seems like people r asleep here
  • [10:35] Zha Ewry: Sorry, RL phone
  • [10:35] Tao Takashi: it's at least per page.. but then again OGP is only big pages ;-)
  • [10:35] Zha Ewry: I htink Zero is still out today, btu I'm planning on stoping by to see
  • [10:35] Tao Takashi: might do so, as well
  • [10:35] Rex Cronon: bye everybody
  • [10:36] Tao Takashi: ok, I am off, too
  • [10:36] Tao Takashi: cya later