[9:45] Saijanai Kuhn: hoping INfinity's work to define a service query for group IM can be exteended to provide connection points for things like client P2P
[9:45] Zha Ewry: so.. if you tp into a space, and tp out, and back
[9:45] Zha Ewry: everyone sees a new set of local IDs for all your stuff
[9:45] Latha Serevi: (newcomers: Zha's question started out with "Has anyone chewed through how we use local IDs?
[9:40] Ewry: In: particaulr, attachements, and local prims, and such, and how they are asigned"
[9:45] Rex Cronon: the same thing happens when ucross sim borders. right?
[9:46] Tara5 Oh: well any one with interesting ideas let me know I was listenting to google, HP, Cisco and Microsoft debating interoperability of clouds and it sounded like they were a step or too behind AWG!
[9:46] Zha Ewry: I happened to be in a perverse mood last weekend
[9:49] Latha Serevi: off the top of my head: textures at least should be easy enough; any "local texture" (didn't know they existed before now) could have a universal UUID attached, as in "if you 've got this cached, use it"
[9:49] Saijanai Kuhn: but where does that universal UUID come from?
[9:51] Saijanai Kuhn: on_rez the local simassigns its own UUID
[9:51] Zha Ewry: and they are UUIDs at the moent, but clearly you would like them tobe URIs, with stabel URNs so you could fetch them from lots of places
[9:51] Zha Ewry: (If URI, URN, and such actually were ever used)
[9:51] Latha Serevi: ...and it's fine for it to assign its own UUID, as long as it doesn't disconnect the new object from its old identity altogether
[9:51] Zha Ewry: Its all the rezzed prims and all the fdynamic texture
[9:54] Zha Ewry: I'm inclined to want them to be on the Agent Domain name space
[9:54] Latha Serevi: Are we talking about some kind of a tradeoff between "indirect reference to a possibly sluggish home asset server where the object originated" and "direct reference to a harder-to-recognize locally maintained asset on a sim" ?
[9:54] Morgaine Dinova: Zha, how wide is the handover zone at a sim edge? Ie. how close to it does an av have to be for handover preloading to start? (or is it the camera position that triggers handover?)
[9:54] Zha Ewry: the asset rezzed into a scenegraph is not the same resource as the asset in stroage
[9:56] Latha Serevi: Item rezzed is not the item stored. Except that most of the time, it is exactly the item stored, modulo a few position-like variables.
[9:56] Tara5 Oh: oops gotta run Veterans day and kid home from school cyas all soon !
[9:57] Latha Serevi: Right - but don't we want to deal with the important common case of "exactly equal modulo position and rotation", for isntance? For caching reasons?
[9:57] Morgaine Dinova: What triggers handover preloading, av position entering the 10m zone, or av's viewer camera position entering it?
[9:57] Zha Ewry: and chewing on this, is really subtle and not baked yet
[10:01] Zha Ewry: I've been chomping on some ideas for this at the web level
[10:01] Morgaine Dinova: Unusual name. Our finger actions are cached and not under voluntary control mostly, and caching of unusual names seems to be borked.
[10:01] Zha Ewry: Including asking if URIs and forwards and rediects might help
[10:03] Latha Serevi: I was thinking the opposite; region has all of the rezzed objects fully available locally, but each has a "source URI" and "what has changed map" of 1 or more bits TBD
[10:03] Zha Ewry: is that if we have a URL, in partricular
[10:04] Zha Ewry: you tend to end up bound onto the very physical server
[10:04] Zha Ewry: (well, the FQDN, you can hide behind that)
[10:04] Sheet Spotter: If everyone needs to rez your attachments, it may be more...convenient(?) to fetch them from a cache on the sim.
[10:04] Morgaine Dinova: Is this handover scheme going to continue to work with portals, eg. a "mirror" in the middle or the zone into which you can walk for TP, but into which you can also look, and see objects on the other side of the TP point?
[10:12] Zha Ewry: Exactly how we model things like local assets, into web space
[10:12] Zha Ewry: probably has moderate impacts in how we slowly evolev away from naked UUIDs
[10:12] Morgaine Dinova: The PHB-developer design pattern is wrong, inverted. The main operator should be a "Tell me what I want" operation.
[10:12] Zha Ewry: has the urge to buy fuzzy litlle URI wrappes for all the UUIDs
[10:13] Zha Ewry: If it actually was doable, I'd love to at least go to:
[10:13] Saijanai Kuhn: would it make sense, at least for the avatar-dominated ones like baked textures and attachments? Or is that too many conventions..
[10:28] Saijanai Kuhn: more out of order messages for me
[10:28] Zha Ewry: I look forward tot he next generation of that toy
[10:30] Morgaine Dinova: How far we are from handling objects in OGP? The world really revolves around objects, not login or teleport or ADs etc, so I'm concerned that OGP may be hardwiring in things that may be wrong or inefficient for object handling.
[10:30] Saijanai Kuhn: at this point, there's so little to the OGP its hard to see how
[10:30] Missile Defence!: v0.10: Alisha Matova, get ready for wave 1!
[10:31] Zha Ewry: I have been very gentl prodding Linden on more of this
[10:31] Saijanai Kuhn: When INfinity has a chance, she'll send me what she's done on the group IM stuff. SHould provide us with samples for client => AD and generic service discovery techniques
[10:32] Alisha Matova: a friend made it, neat use of that function , ithink
[10:33] Saijanai Kuhn: both of which are lacking from the spec and would make the AD's useable already if it were available
[10:34] Zha Ewry: OK, RL is about to suck up my full attention
[10:34] Zha Ewry: One thing I would love to see brought up at Office Hours, and people thinking about
[10:34] Zha Ewry: is the questoin fo when does OGP actually refer to an asset, and such
[10:35] Morgaine Dinova: But seems to be very hard to get anyone to talk about it
[10:36] Zha Ewry: They are in a pause, while they do a plannign cycle I think
[10:38] Saijanai Kuhn: there seems to be a hiring freeze also, but not sure
[10:38] Morgaine Dinova: I sometimes wonder if the AWG is actually doing anything, since we're always waiting for delivery from LL. The original vision of a design group is so far back in the past now. We seem to be no more than commentators on things that LL dumps over the wall to us.
[10:38] Zha Ewry: That is a very good qustion Morgaine
[10:38] Zha Ewry: One way to fix that, would be coding in parallel in OpenSim
[10:39] Zha Ewry: the challange there, is the client
[10:39] Morgaine Dinova: Yes, they're doing a fantastic job.
[10:39] Saijanai Kuhn: well, once we get the service discovery pattern and related output CAP pattern, we can start adding things ourselves
[10:40] Rex Cronon: does it have to be 3d client like the present viewer, or somthing simpler?
[10:40] Zha Ewry: Well, to test? No, but to be useful? Ueah
[10:41] Saijanai Kuhn: I've got pyogp to the point it can display generic packets in a nice GUI. Tao has his limited AD and barxjander i working on a different AD based on google apps
[10:41] Saijanai Kuhn: and Dahlia has been working on a pyogre renderer for the pyov client
[10:47] Morgaine Dinova: The out of order messaging is curious. I like decoupling, but decoupling messages from time ordering constraints is taking it too far.
[10:48] Saijanai Kuhn: its an odd thing. It usually doesn't happen in local chat as that is sim local and coming in via a different kind of packet.
[10:48] Rex Cronon: it can also lead to very interesting conversations:)
[10:48] Saijanai Kuhn: but, given that eerything seems to be laggy, such as rezzing the drawing prims out of order, I guess its to be expected
[10:49] Zha Ewry: OK, RL demands my full attention...
[10:49] Zha Ewry: So.. see most of you at Office hours...