[9:58] Pixel Gausman: well, i think it looks for clamped values for region_x and region_y. i need to deeply ponder that check to understand what it's truely checking
[9:58] Zha Ewry: Oh, and amusingly enough, due to enough bugs and oddity, since the sim was on the map, and mapped to vivoxx, voice worked on the OpenSim, as long as you were in voice whentyou tpd
[10:05] Latha Serevi: Remedial code-fu: I the server code on opensimulator.org ; what client(s) are we working with?
[10:06] Zha Ewry: "Avatar name@agent domain authenticating" and uuid@uuid of authenticator"
[10:06] Dahlia Trimble: currently the Linden client is the main client for opensim but there are others in the works
[10:06] Zha Ewry: The agent domain, which voices for this avatar's ID and UUID, and uniqueness seems the right thing
[10:06] Zha Ewry: But. That's a guess, hardly a certainty
[10:06] Tess Linden: what we need is some way to associate an agent from one agent domain with an agent from another agent domain
[10:07] Zha Ewry: OpenSim, normal, runs on the 1.19(4) and the 1.20(11) RC just fine
[10:07] Zha Ewry: For the interop test, we were running on a hacked version of the 1.20(9) RC
[10:07] Goldie Katsu: Wouldn't the agent just be in one domain and trust would allow the agent to rez in the new domain?
[10:07] Latha Serevi: What code repository should I use if I want to squint at a client and see what it does?
[10:08] Rex Cronon: u need a new table in your database, for which avatar, it lists the name/uuid/gridName
[10:08] Zha Ewry: I *think* we need to assume that some avatars are going to be associated with mlutiple agent domains
[10:08] Tess Linden: Goldie: when other people from the new domain see you, who do they think you are? _EXTERNAL? or can you someway tell them that you are also 'Tess Linden' on an OpenSim grid?
[10:09] Zha Ewry: right now it's the hacked _EXTERNAL
[10:09] Zha Ewry: my next spin of the patch should map the IP of the agent domain, to a string
[10:09] Goldie Katsu: Ah yes, there is the avatar on agent domain a needs to talk with avatar on agent domain b and exchange items/see items while in region domain c.
[10:09] Zha Ewry: so it should be "tess Linden@aditi.lindenlab" or such
[10:09] Goldie Katsu: That is true. Wasn't sure whether we were talking about avatar in multiple agent domains or two avatars interacting.
[10:14] Zha Ewry: I think, a region domain, is likely to be a colletcion of regions which share trust, and can show memebership
[10:14] Anamolie Poitier: i'm worried that if we tp into grids/sims from external networks without already having an identiry account in our destination network then it may be harder to code around the authority hacks later on. wouldn't it be easier to just have the recieving network check a 3rd party trust server to vouch that the avatar belongs to the same user on both networks?
[10:14] Zha Ewry: so... you'd be able to say "I'm in the ZhaTest" region domain
[10:15] Pixel Gausman: Zha: how is region domain different from grid?
[10:15] Zha Ewry: I think a region is less than a grid
[10:24] Zha Ewry: that asset servers, will serve up to lots of region domains
[10:24] Zha Ewry: federation, but the trust links stay short
[10:24] Latha Serevi: I'd be interested in refining my picture of where the asset servers sit. My inventory will be in an asset server gatekeeped-by my AD? But there will also be assets, say, rezzed in a region, gatekeeped-by its RD?
[10:27] Goldie Katsu: unless the inventory pointers get borked in which case you just don't see it any more.
[10:27] Zha Ewry: "Failed to find clothing named old frebie jeans in database"
[10:27] Latha Serevi: The idea of a "no copy" object disappearing from your inv when it's rezzed, seems like it may have to change a bit in the move to open grids. I'm not gonna want an asset server to delete assets just cuz they're being rezzed off in Joe's sim.
[10:28] Zha Ewry: Well, I think that's going to depend on the permissions model of the object
[10:28] Latha Serevi: Perhaps more like bookkeeping bits, not "delete it"
[10:30] Latha Serevi: In the future, they're objects wearing a unique "dog license". Just off the top of my head.
[10:30] Rex Cronon: if u were to register your no copy object with a "special" service you might be allowed to rezz multiple copies? could that work?
[10:31] Latha Serevi: So it's not the bits that you have only one of, but the right to use only one license.
[10:31] Zha Ewry: If the license of an object is "I only want you to be able to have one copy rezzed at a time and I don't want it to escape the Linden Trust fomain"
[10:31] Zha Ewry: Then, the quiestion is.. can we support such a model, and hopefully not lose the user's assets in the face of sim crashes
[10:31] Latha Serevi: We never lose assets if we don't delete the bits. You may have to "reclaim" your license
[10:32] Zha Ewry: if you look at the c-hhtp and escrow owrk, which linden has done
[10:32] Zha Ewry: that's probably part of the puzzle
[10:32] Zha Ewry: reliable ways to prove you've copied an object and in an acid transaction, deleted the orginal
[10:32] Latha Serevi: What's the approach with escrow? Delete the bits but keep a backup copy under lock and key?
[10:37] Zha Ewry: In fact, I'd say tha's a base usecase for all this
[10:37] Latha Serevi: Yucko. I'm still gonna drag my feet on that. Seems wrong somehow, to remove the possibility of fixing things later. Maybe if the law requires, but surely not just as the implementation of a license-to-use?
[10:38] Zooby's: All: three labrador models have been completely redone. As you know all updates are FREE! Please rerezz your old pet to receive the latest update.
[10:38] Zha Ewry: Lataha, I have some personal e-mail goign back to 1980. I'm with you, but.. I also delete my e-mail that's routine and older than 2 years, at work.
[10:38] Latha Serevi: Archival of email, and management of an asset DB, may not be identical issues
[10:38] Anamolie Poitier: i think the license-to-use stuff will be more for functional content and must-delete stuff will probably be more for confidential data
[10:39] Zha Ewry: I'm fairly laissez faire on prtocol design
[10:39] Zha Ewry: I'm willing to assume we can design it to span the cases and let people build licences they think will sell
[10:40] Goldie Katsu: I think we have to assume that the protocols we design will be used as the developers see fit.
[10:40] Zha Ewry: if it turns out that 90% of people *hate* licenses which result in lost assets, then stores and creators will likely evolve to more open ones, but.. I hate to gyuess
[10:40] Zha Ewry: People, in general, will bend our work to thier needs, and the easier, we make that, the happier we will be
[10:41] Latha Serevi: I guess I'm just thinking that "secure transactions involving reliable deletion of bits" is an unnecessary precondition for "license-preserving interaction between sims". Maybe it won't be too hard to have that as part of the first impl, but I don't see a requirement for it really.
[10:41] Pixel Gausman: zha: let's just make sure we dont flex ourselves into a security hole
[10:41] Zha Ewry: I'd much rather have a design which allows it to be clear that an obejct is being deleted, than.. one where its happenign anyway
[10:41] Zha Ewry: Well, I assume that some people will, build totally open domains, and that they won't ever get hooked up to some grids, and that's ok. Thr protocol should, if possible
[10:42] Zha Ewry: handle everythign from totally permissive policies, to ones which will make the prims cringe
[10:42] Pixel Gausman: Zha: is there a list of use cases that we anticipate?
[10:42] Goldie Katsu: And I think there will be some places where "suggested best practices" might be useful but can't be defined as part of the standard.
[10:42] Zha Ewry: "No, no, don't rez me on that sim, they are mean to me"
[10:43] Latha Serevi: Zha, this is a big enough issue that I'll stick with it for a few more minutes. If we get stuck in the idea that "rights-preserving sim" is identical to "delete-on-transfer sim", I think we're shooting ourselves in the foot.
[10:43] Goldie Katsu: I think the implementation isn't one we can control
[10:43] Goldie Katsu: what we need to define is a way to know what the sim will be doing.
[10:43] Zha Ewry: I think we can anticpiate that there will be a large cry for a set of polcies which closely follow the current Linden rules, as best we can make that happen
[10:44] Zha Ewry: Protocol, is better at marking behavior than enforcing it
[10:44] Latha Serevi: Shouldn't we try hard to retain open to the possibility of a sim implementing the LL rules in a different way than the knee-jerk one?
[10:44] Zha Ewry: I assume we'll see lots of variations
[10:44] Zha Ewry: Goldie's been very cogent, in suggesting layering, and policy as they way to go here
[10:44] Zha Ewry: if the protocol supports a range of options
[10:45] Zha Ewry: the greate ecosystem will fill out those which work
[10:45] Latha Serevi: That means, don't put in a requirement for delete-the bits, but rather respect-the-license. But it would be easy to make that mistake of casting "delete-the-bits" in stone. Let's not.
[10:45] Zha Ewry: and gradually evolve away form deadends
[10:45] Goldie Katsu: and Goldie is trying to make it though the documents and define the layers that she sees based on what she is reading. :)
[10:45] Zha Ewry: I think you need a "KILL THIS OBJECT or be in violation of TOS" abilitiy
[10:46] Rex Cronon: the bits support the license:)
[10:47] Zha Ewry: if someone wants to write a license which says "Destroy the bits after 180 days" and we can mark it and support it, and someone has a trust domain permitting it? I'm not going to buy many of those objects, but.. people may
[10:47] Zha Ewry: (ack, RL phone call, one moment)
[10:47] Latha Serevi: OK, but if we then let that become the default permissions for all legacy objects, we've shot ourselves in the foot again.
[10:48] Latha Serevi: regrets having mentioned legacy objects at all
[10:48] Goldie Katsu: Ah well there are two issues there - what do we make the protocols support and what do we do when we transition legacy bits.
[10:48] Latha Serevi: Let's forget legacy for now, forget I said it.
[10:48] Goldie Katsu: Probably isn't ours to decide anyway :)
[10:49] Latha Serevi: What seems more relevant is, making it easy to choose a sensible scheme for newly built objects. One that doesn't require secure delete-theib-ts transactions just to make an object sellable.
[10:49] Rex Cronon: whoever makes it, decides that
[10:50] Latha Serevi: Not if they dont have the choice. And we need to noodge them in an open direction or opensim is toast.
[10:50] Goldie Katsu: Ideally the implementation should make it easy to know what you are deciding.
[10:50] Goldie Katsu: And what you are deciding when you buy.
[10:52] Latha Serevi: Mostly I'm suggesting that we try to promulgate the idea that "rights-preserving" doesn't mean "police state with secure deletion", and that "making money" doesn't require "clamoring for the most restrictive possible system"
[10:53] Zha Ewry: I am going to encourage and support creatyors who offer sane licenses
[10:53] Saijanai Kuhn: WEll, I expresed what I think ar ethe two extremes in teh jira: Creation-grid-only (grid permissions) to any grid (any permissions). The hard part is the middle
[10:53] Zha Ewry: I don't buy non-mod prim clothin, and I support copy/no-xfer as desirable
[10:53] Goldie Katsu: I think it would be beneficial to define the options and look at the implicans of eaopiton.
[10:54] Latha Serevi: How about a disclaimer on the "delete it" option that says "this is probably a bad idea, consider using licenses instead. does the same thing, faster and with less risk of data loss.". :-)
[10:54] Zha Ewry: I think we need a few strong use cases
[10:55] Goldie Katsu: waches her USB bus decend into madness and waits for the panic
[10:55] Saijanai Kuhn: I think there are two issues: designing the system to handle anything we can think of (hard) and making sure it handles what already exists (still hard)
[10:55] Zha Ewry: Actually, Saij, I'd argue.. that getting the prtocoll right is going to be easier, than getting the community to adopt some things
[10:56] BlueWall Slade: whaqt happens when RIAA/ASCAP and such get here?
[10:56] Goldie Katsu: Confidentiality Integrity Availability - people have different priorities on the security triangle.
[10:56] Saijanai Kuhn: well, the community doesn't need to adopt anything. It can be "THis grid only" forever after. The trick is the folks who want to open up may not understand all the implications (certainly we don't, so hw could they?)
[10:57] Zha Ewry: Oh.. and by the way.. for amusement, the prim on the right, with two aves floating.. is the first round trip tp
[10:58] Goldie Katsu: The 8th layer of OSI network model is always the hardest.
[10:58] Saijanai Kuhn: so you went SL => OpenSIm => SL?