Difference between revisions of "AW Groupies/Chat Logs/AWGroupies-2008-04-29"

From Second Life Wiki
Jump to navigation Jump to search
(New page: * [9:27] Saijanai Kuhn: high all and bye * [9:27] Saijanai Kuhn: afk * [9:27] SignpostMarv Martin: boo * [9:29...)
 
Line 373: Line 373:
* [11:03] [[User:Zha Ewry|Zha Ewry]]:  Thanks all
* [11:03] [[User:Zha Ewry|Zha Ewry]]:  Thanks all
*
*
[[Category: Pyogp]]

Revision as of 17:16, 30 June 2008

  • [9:27] Saijanai Kuhn: high all and bye
  • [9:27] Saijanai Kuhn: afk
  • [9:27] SignpostMarv Martin: boo
  • [9:29] SignpostMarv Martin: ^the grid as it looks now
  • [9:29] Enus Linden: yoo much water
  • [9:29] Enus Linden: need to fill that in with some land!
  • [9:29] SignpostMarv Martin: the actual data shows the grid is a fair bit larger
  • [9:29] SignpostMarv Martin: I've cropped it to visible sims
  • [9:30] SignpostMarv Martin: the images are in https://wiki.secondlife.com/wiki/Category:Images_of_Agni
  • [9:33] SignpostMarv Martin: need to mess about with the joystick flycam settings for "native" Wiimote support vs keyboard control
  • [9:34] Zha Ewry: Not a huge crowd
  • [9:35] Rex Cronon: hello everybody
  • [9:36] Rex Cronon: thank you for the tp
  • [9:38] Zha Ewry: SL has been cranky this morning. I'm goign to give it a couple mor eminutes and we'll get started
  • [9:39] Zha Ewry: passes Goldie the hair dryer
  • [9:39] Goldie Katsu: I really should remember to wear a swimsuit
  • [9:39] Davina Glitter: hehe
  • [9:41] Zha Ewry: OK
  • [9:41] Zha Ewry: I htink we'll decalre a quorum
  • [9:42] Zha Ewry: So... Enus asked for some time on test harnesses
  • [9:42] Zha Ewry: and.. that's a really good topic
  • [9:43] Enus Linden: we'd talked about this once, a while back, and am hoping to get moving in the direction of putting something together...
  • [9:43] Zha Ewry: In particular, there are sort of two parts to the topic
  • [9:43] Rex Cronon: are we going to dicsuss how LL does it?
  • [9:43] Zha Ewry: I thnk we're going to dicuss how we all want to do it
  • [9:44] Zha Ewry: For example
  • [9:44] Zha Ewry: I am finding I need to have a test harness, which sends to a Sim, what the Agent Domain sends as a rez-avatar call
  • [9:45] Zha Ewry: and, of course, since that's a several step little dance, it needs to catch back some stuff fromt he sim
  • [9:45] Zha Ewry: As we define more stuff, it would be really nice if we had a general approach to:
  • [9:45] Zha Ewry: spit out well formed requests
  • [9:45] Zha Ewry: And handle simple responses, to allow multistep tests to run
  • [9:46] Zha Ewry: There are sort of three or four related things in here, I think
  • [9:46] Zha Ewry: An easy way of making requests/handlers
  • [9:47] Zha Ewry: Some way, of sharing some scafodling, so we don't keep re-inventing the basic respons stuff
  • [9:47] Zha Ewry: and.. possibly a way of pluging in lots of seperatly authored bits into a bigger test harness over time
  • [9:47] Goldie Katsu: nods
  • [9:48] Zha Ewry: So.. we have, at least three parts to simulate, now
  • [9:48] Zha Ewry: (Client) (agent domain) (region simulator)
  • [9:48] Zha Ewry: and.. over time
  • [9:48] Zha Ewry: (not to distant)
  • [9:48] Zha Ewry: Asset server
  • [9:49] Enus Linden: to simulate being to have an environment to test against zha, or...?
  • [9:49] Zha Ewry: Well
  • [9:49] Zha Ewry: In the second and third proposed steps we want to test with OpenSim
  • [9:49] Zha Ewry: Fetching assets comes into play
  • [9:50] Zha Ewry: At that point, it would be good
  • [9:50] Zha Ewry: to be able to have a simple web service, that acts like the asset cloud and returns some canned objects
  • [9:50] Zha Ewry: (Define a set of 10-20 UUIDs or REST URLS, and have a cnnonical set of test assets)
  • [9:51] Umeko Kawanishi: what about test avatars, zha?
  • [9:51] Zha Ewry: Good point
  • [9:51] Enus Linden: ah yes, best practices
  • [9:51] Zha Ewry: The agent domain, would benefit fromt hat
  • [9:52] Zha Ewry: Test aves, with test inventories
  • [9:52] Zha Ewry: In fact
  • [9:52] Zha Ewry: we could easily set up a smalls et of test resources all linked together
  • [9:52] Zha Ewry: Test Dummy
  • [9:52] Zha Ewry: With a small inv
  • [9:52] Zha Ewry: in the agent domain
  • [9:52] Zha Ewry: A test tim
  • [9:52] Zha Ewry: *sim
  • [9:52] Goldie Katsu: Can we get a test dummy skin for it?
  • [9:52] Zha Ewry: with a small set of assets
  • [9:52] Zha Ewry: and then..
  • [9:52] Zha Ewry: an asset server test harness that maytches
  • [9:53] SignpostMarv Martin: ^isn't there a Ruth-like avatar distributed with OpenSim ?
  • [9:53] Enus Linden: my reservation about Test Dummy is a potential for griefing, i've been wondering how best to do this...
  • [9:53] Zha Ewry: Well
  • [9:53] Zha Ewry: We don't need a main grid asset set for this
  • [9:53] Zha Ewry: Tho..
  • [9:53] Zha Ewry: I suppsoe it woudl be uber cool, if they matched
  • [9:53] Enus Linden: i would like to be able to run the same sets of tests
  • [9:53] Davina Glitter: Grey goo avi... :)
  • [9:53] Enus Linden: on any grid
  • [9:53] Zha Ewry: right
  • [9:54] Goldie Katsu: A way to document the tests would be useful...so that regression testing could be added and verified for its presence.
  • [9:54] Umeko Kawanishi: what about the "ruth-like" avatar that signpostMarv spoke of? i don't know much about it
  • [9:55] Enus Linden: perhaps initially we allow user data to be data driven, and defined by the test environment, that is, a template for the data is included with the harness
  • [9:55] Zha Ewry: Ruth is in the client
  • [9:55] Enus Linden: but the contents ned to be updated with actual agent info prior to execution
  • [9:55] Goldie Katsu: thinks that sounds like a ghost in the machine.
  • [9:55] Enus Linden: not ideal...
  • [9:55] SignpostMarv Martin: it's in assets\Avatar\Newruth
  • [9:55] Enus Linden: but am hesitant to put a username and password anywhere publicaly accessible
  • [9:56] Zha Ewry: Yeah, not one that works on main grid and even aditi
  • [9:56] SignpostMarv Martin: 5 jpeg2k's and a settings file
  • [9:57] Rex Cronon: how about temp passwords, that work for only a limited time
  • [9:57] Zha Ewry: Hard to do
  • [9:57] Goldie Katsu: Is there a way to test asset, agent and sim (and client) indepenedently of each other.
  • [9:57] Enus Linden: new agent for each test is an option...
  • [9:57] Zha Ewry: That's why I want to have test harneses for each part
  • [9:57] Goldie Katsu: Ok, just clarifying.
  • [9:58] Zha Ewry: I am imaginging we'll have
  • [9:58] Zha Ewry: Small basic web server parts for the Agent, Region and grid bits
  • [9:58] Zha Ewry: and a more complex thing for the client
  • [9:58] Zha Ewry: Ideally, some simple snap in framework for the first three
  • [9:58] SignpostMarv Martin: ^svn link to the ruth-like avatar in OpenSim: [1]
  • [9:59] Zha Ewry: Basic REST handler
  • [9:59] Zha Ewry: And a way of snapping in a assembly to handle each request
  • [9:59] Umeko Kawanishi: thanks signpostmarv
  • [10:00] Enus Linden: llpython ftw....
  • [10:00] Enus Linden:  : )
  • [10:00] Zha Ewry: Well
  • [10:00] Zha Ewry: grins
  • [10:00] Zha Ewry: We clearly wnat phthon
  • [10:00] Zha Ewry: we also want some parts in at least one other language
  • [10:00] Zha Ewry: otherwise, we don't drive some of the issues what show up in on the wire/vs code defintinon of interop
  • [10:00] Enus Linden: yes
  • [10:01] Zha Ewry: (Python tends to deliver things in dict order, for example)
  • [10:01] Enus Linden: so repo for this test harness is becoming a priority: i've floated options internally and have come up with LL creating a new branch in svn, and publishing that as if it were a source code drop as needed. if not as familiar with the svn -> awg flow as i should be, i'll fix that. does that sound reasonable, or no?
  • [10:01] Zha Ewry: Unless the spec calls for this
  • [10:01] Zha Ewry: we want to allow them to be non ordered
  • [10:01] Enus Linden: k
  • [10:01] Zha Ewry: And. That would be good. If we got hat route?
  • [10:02] Zha Ewry: I'd like to suggest we try for an ApahceV2 or similar license, like Whihc Linden has done on the c-http work
  • [10:02] Zha Ewry: I'd like to be able to share the core test harness stuff with the OpenSim, LibSL, and any other communities which want to play
  • [10:03] Enus Linden: k, i'll follow up with you and which on that then...
  • [10:04] Zha Ewry: So.. I'd suggest we aim for a structur where we have one basic shell
  • [10:04] Zha Ewry: Which handles caps (and sadly, maybe the UDP pipe)
  • [10:04] Zha Ewry: and then has a very simple top level way of passing off requests to simple handlers
  • [10:04] Zha Ewry: Build out an skeletel Agent Domain, Region Domain, and eventually some of the services
  • [10:05] Zha Ewry: I'm less sure on the client side
  • [10:06] Enus Linden: who? when? which bits? what order? type questions come to mind...
  • [10:06] Zha Ewry: IF we do it right, we also, make it farly easy to plug in handlers in non pyhton languages
  • [10:06] Zha Ewry: grins
  • [10:06] Zha Ewry: Well.. we need:
  • [10:06] Zha Ewry: a simple caps hander
  • [10:06] Zha Ewry: the long queue code
  • [10:06] Zha Ewry: and a very simple way to add message handlers
  • [10:06] Zha Ewry: That would be a good starting point
  • [10:07] Zha Ewry: I think that some of what Which has done (and possibly Donovan) in the c-http work might be adaptable
  • [10:08] Enus Linden: i'll take the task of summarizing and posting to sldev all of the above.
  • [10:08] Zha Ewry: Cool
  • [10:08] Zha Ewry: One related thing which will likely happen, in the mid-term
  • [10:08] Tess Linden: the long queue is mainly useful for messages that go from server -> client
  • [10:09] Zha Ewry: is that I'm going to re-code the caps severer on OpenSim, as a hgiher performance
  • [10:09] Zha Ewry: version
  • [10:09] Zha Ewry: and we might adapt that as asecond test base
  • [10:09] Zha Ewry: Right
  • [10:09] Tess Linden: I dont know if we need any of those yet for login and teleport
  • [10:09] Zha Ewry: We'll need that in at least the region simulator
  • [10:09] Zha Ewry: Well
  • [10:09] Zha Ewry: Long term, we will want it
  • [10:09] Zha Ewry: To do a full up version
  • [10:09] Zha Ewry: I'd be happy, if we don't need it at all at first
  • [10:09] Enus Linden: i'd like to contribute as much as is reasonable, but will primarily be interested in facilitating the work for the group overall.
  • [10:10] Umeko Kawanishi: just a status check--has anyone started test cases in opensim yet?
  • [10:10] Zha Ewry: I am currently ripping open the OpenSim caps handler
  • [10:10] Enus Linden: who in awg is interested in playing with this part of the work? is that know?
  • [10:11] Enus Linden: i'll find out that answer : )
  • [10:11] Zha Ewry: Wel, I am
  • [10:11] Zha Ewry: Tao has expressed repeated interest
  • [10:11] Zha Ewry: Saijanai
  • [10:11] Enus Linden: naturally
  • [10:12] Enus Linden: kk
  • [10:12] Zha Ewry: In moments of red-bull inspired enthusiasm, Gareth Ellison
  • [10:12] Rex Cronon: u could make a group notice enus
  • [10:12] Enus Linden: wings = testing inspiration?!
  • [10:12] Zha Ewry: shakes her head
  • [10:12] Enus Linden: will do rex, ty
  • [10:12] Zha Ewry: its safe to say, Gareth is red-bull powered
  • [10:15] Enus Linden: don did some work over the weekend putting together a python login and caps oriented script
  • [10:15] Zha Ewry: Hello Teravus
  • [10:15] Teravus Ousley: Hi
  • [10:15] Enus Linden: i've going to requisition this for our uses, as well as sai's login work, as first commits to svn... starting points
  • [10:16] Enus Linden: starting to define the harness will come next, i'll see if sai wants to document, or if he has enough on his plate...
  • [10:16] Zha Ewry: Sound great
  • [10:16] Zha Ewry: Any additional test harness topics?
  • [10:17] Rex Cronon: creating an automated testing harness for the viewer might be kind of hard
  • [10:18] Enus Linden: 's action item: to boil this down and post to sldev, talk to zha and which about c-http and svn, and start moving toward docs for harness
  • [10:18] Rex Cronon: as a user has to actually look at the screen and see the results
  • [10:18] Enus Linden: agreed
  • [10:18] Enus Linden: but i think awg's work is not focused on viewers
  • [10:19] Enus Linden: it's oriented toward protocols that define agent-> agetndomain -> sim
  • [10:19] Enus Linden: spelled nicely of course
  • [10:19] Goldie Katsu: or at least spelled nicely sometimes.
  • [10:19] Enus Linden: our test harness should essentially be a client on it's own
  • [10:20] Zha Ewry: Exactly Enus
  • [10:20] Enus Linden: ll has their's, libsl has their's, there are others... but this client/harness is written against protocols collectively defined and freshly implemented
  • [10:21] Goldie Katsu: There might be a few forms of the client on its own as well.
  • [10:21] Enus Linden: mmm new bots
  • [10:22] Zha Ewry: listens for a moment
  • [10:22] Zha Ewry: Yes
  • [10:22] Zha Ewry: I think we really want a clean sheet of paper on the test client
  • [10:23] Goldie Katsu: I think breaking down the functionality to its smallest units may be useful.
  • [10:23] Goldie Katsu: (Smallest testable units)
  • [10:24] Goldie Katsu: Identifying dependencies might be useful too.
  • [10:24] Zha Ewry: I am inclinded to agree Goldie, especialy as we look at regression testing
  • [10:24] Enus Linden: it's certainly blank atm... i'd like to suggest that the tests themselves are written to test the protocols as defined in the public domain
  • [10:24] Enus Linden: and that the harness is structures as zha pointed out in such a way that we can enable using the smallest unit of tests in creative and grrouped/sequenced ways
  • [10:25] Enus Linden: a unit test works on it's own
  • [10:25] Zha Ewry: Right
  • [10:25] Zha Ewry: One thing that we ought to be able to do, is compose up short simple sets of tests
  • [10:25] Goldie Katsu: Testing the protocols definitely seems essential. and the security person in me suggests testing what happens with malformed packets to make sure it handles exceptions correctly.
  • [10:26] Zha Ewry: "Login as Test Dummy. on Domain X, login to a sim on Domain Y, fetch the connonicaal assets frm DFomain Z"
  • [10:26] Enus Linden: yes yes yes
  • [10:26] Rex Cronon: so what do u need for a bare-bone client?
  • [10:26] Goldie Katsu: and now i have to run
  • [10:26] Zha Ewry: Login in
  • [10:26] Rex Cronon: a thread that listens for udp packets
  • [10:26] Goldie Katsu: waves and poofs
  • [10:26] Zha Ewry: And UDP
  • [10:26] Zha Ewry: and.. queue handler
  • [10:26] Rex Cronon: bye goldie
  • [10:27] Zha Ewry: and the ability to do the basic
  • [10:27] Zha Ewry: /bye goldie
  • [10:27] Zha Ewry: stuff to get logged on
  • [10:27] Rex Cronon: another thread for tcp communication
  • [10:27] Rex Cronon: a buffer to store for each thread
  • [10:27] SignpostMarv Martin: needs to log as he's expecting a delivery
  • [10:27] SignpostMarv Martin: toodles peeps
  • [10:27] Rex Cronon: bye singpostmarv
  • [10:28] Zha Ewry: Yeah I think we'll, in time
  • [10:28] Enus Linden: i'll boil all of the above into sldev mail and wiki contributions as i said....
  • [10:29] Umeko Kawanishi: btw, are the test cases we are talking about here functionality-based, or performance-based?
  • [10:29] Rex Cronon: and a class called Test that everybody extends, which has function named runTest, that u overwrite
  • [10:29] Zha Ewry: need to be able to handle connections to the Agent Domain, a region, and child regions
  • [10:29] Enus Linden: functional initially imo
  • [10:29] Enus Linden: performance suites can follow?
  • [10:29] Umeko Kawanishi: i see
  • [10:30] Enus Linden: is a question... really
  • [10:30] Enus Linden: we can do what we collectively want
  • [10:30] Enus Linden: i'd prefer unit tests that make sure what is implemented works
  • [10:30] Umeko Kawanishi: so is the current way of functionality in either opensim or sl very manual?
  • [10:30] Enus Linden: then we can see how well it works
  • [10:30] Umeko Kawanishi: functionality testing, i mean
  • [10:31] Enus Linden: ll has their own internal unit tests
  • [10:31] Enus Linden: zha: opensim?
  • [10:31] Teravus Ousley: OpenSimulator: currently.. but we're actively working to add unit tests and an automated testing routine on commit
  • [10:32] Enus Linden: is gonna brush up on opensim now :)
  • [10:32] Teravus Ousley: Adam in particular is working on the automated testing routine.
  • [10:32] Teravus Ousley: :D just added attachments over the weekend to SVN.
  • [10:32] Teravus Ousley: Prim.State = (byte) attachmentpoint. Sneaky
  • [10:33] Teravus Ousley: :D
  • [10:33] Umeko Kawanishi: looking forward to trying it
  • [10:33] Zha Ewry: I think, on the whole?
  • [10:34] Zha Ewry: Testing is pretty limited on both sides
  • [10:34] Zha Ewry: Its a hard, time consuming process
  • [10:34] Zha Ewry: especialy witht he old UDP message template stuff
  • [10:34] Zha Ewry: I'm hoping that one of the nice things that falla out of this
  • [10:34] Zha Ewry: is making sure that the on wire prrotocls are easier to test
  • [10:35] Zha Ewry: In general, the more seperate, decoupled and well broken out they are
  • [10:35] Zha Ewry: both the better they will behave in use
  • [10:35] Zha Ewry: and the easier to test
  • [10:35] Enus Linden: amen, and amen
  • [10:35] Enus Linden:  : )
  • [10:36] Zha Ewry: tess?
  • [10:36] Zha Ewry: I think you had a topic you wanted to bring up?
  • [10:36] Zha Ewry: sees if Tess is jet lagged, or awake ;--)
  • [10:36] Tess Linden: oh yes, I just came back from a 2 week vacation
  • [10:37] Tess Linden: have we discussed the teleport protocols yet at AWG?
  • [10:37] Zha Ewry: Not in detail, no
  • [10:37] Tess Linden: we have some diagrams to post, but I'd like to discuss that, and ask about naming certain web services
  • [10:39] Zha Ewry: listens
  • [10:39] Zha Ewry: Naming, in what sense?
  • [10:39] Tess Linden: heres the rough protocol
  • [10:39] Tess Linden: I dont think we have enough time to discuss, so I'll just paste it here
  • [10:39] Tess Linden: v --> send me to sb (rez_avatar) --> ad
  • [10:40] Tess Linden: ad ---> will you take v? (will_you_take_v) -------------------> sb
  • [10:40] Tess Linden: ad <... sure, data pipe cap ..................... sb
  • [10:40] Tess Linden: ad --> give up v to pipe --> sa
  • [10:40] Tess Linden: sa --- agent packet (teleport_agent) --> sb
  • [10:40] Tess Linden: sa <... 200 .. seed, info .. sb
  • [10:40] Tess Linden: sa remove v
  • [10:40] Tess Linden: ad <.... 200 .. seed, info ..............sa
  • [10:40] Tess Linden: v <- tp complete, seed- ad
  • [10:41] Zha Ewry: v is?
  • [10:41] Rex Cronon: v=?, ad=?, sb=?
  • [10:41] Zha Ewry: ad=agent domain
  • [10:41] Zha Ewry: yes?
  • [10:43] Tess Linden: v is viewer
  • [10:43] Tess Linden: ad is agent domain
  • [10:43] Tess Linden: sa is sim a
  • [10:43] Tess Linden: and sb is sim b
  • [10:43] Zha Ewry: ahh
  • [10:43] Zha Ewry: Viewer
  • [10:43] Zha Ewry: palm smacks
  • [10:43] Zha Ewry: (or is that head palms)
  • [10:44] Rex Cronon: is this using tcp? right?
  • [10:45] Tess Linden: yes, it will be over caps
  • [10:45] Teravus Ousley: okay, naming?
  • [10:45] Teravus Ousley: isn't quite sure what we're naming :D
  • [10:46] Tess Linden: what should (will_you_take_v) and (give_up_v_to_pipe) be called?
  • [10:47] Zha Ewry: Ahh
  • [10:47] Zha Ewry: So, REST
  • [10:47] Teravus Ousley: nods and hrmms
  • [10:47] Zha Ewry: what's the state we're trying to trasnger
  • [10:47] Rex Cronon: if u could provide what is sent during each step, type of each data, and an example. that will be very helpfull:)
  • [10:47] Zha Ewry: *transfer
  • [10:47] Zha Ewry: And what's the state its ending in
  • [10:48] Zha Ewry: The state is the avatar?
  • [10:48] Zha Ewry: The question is
  • [10:48] Zha Ewry: Can you let this avatat rez here?
  • [10:48] Teravus Ousley: should probably have something like authorize_token_for_agent.. or something for (will_you_take_v)
  • [10:49] Zha Ewry: Hmm
  • [10:49] Zha Ewry: We're gettigng back a seed cap?
  • [10:49] Teravus Ousley: presumably we're sending some kind of auth data.
  • [10:49] Teravus Ousley: .. or the agent domain is.
  • [10:49] Tess Linden: we're getting back a cap, not a seed cap
  • [10:49] Zha Ewry: Ah
  • [10:49] Zha Ewry: OK
  • [10:49] Tess Linden: the cap should be named (give_up_v_to_pipe)
  • [10:49] Zha Ewry: so we're rquestoing a cap to rez an avayar
  • [10:50] Tess Linden: its more a cap to derez
  • [10:50] Zha Ewry: Hmm
  • [10:50] Zha Ewry: Well that's a little odd
  • [10:50] Zha Ewry: because, the de-rez happens on A, but we're askgn for the new cap on B
  • [10:51] Zha Ewry: I understand the reasoning
  • [10:51] Zha Ewry: Just.. tryign to think abotu the implied messages
  • [10:51] Teravus Ousley: yep, :D box the agent up into a safe and happy boat so it can be transferred safely
  • [10:51] Tess Linden: hm you're right zha
  • [10:51] Tess Linden: the protocol is incorrect then. sb should give ad a cap to (teleport_agent)
  • [10:52] Zha Ewry: So
  • [10:52] Zha Ewry: COuple of interestign thoughts
  • [10:52] Zha Ewry: One is that we're righ on the edge of a call that wants to be acid
  • [10:52] Zha Ewry: or at least transqactional
  • [10:52] Teravus Ousley: The only issue potentially is you'd have to re-rez your avatar if you derezed without knowing if the new place will accept you.
  • [10:53] Zha Ewry: (eg. c-http_ space
  • [10:53] Zha Ewry: well,
  • [10:53] Zha Ewry: one nice thing is that you don't lose your connection to the AD
  • [10:53] Zha Ewry: so you can always have the viewer and the AD
  • [10:53] Zha Ewry: returty
  • [10:53] Zha Ewry: *retry
  • [10:54] Zha Ewry: But
  • [10:54] Zha Ewry: what we're doing is tryign to keep
  • [10:54] Zha Ewry: One and only one active connection to a simulator, yes?
  • [10:54] Tess Linden: yes
  • [10:54] Zha Ewry: So
  • [10:55] Zha Ewry: We're saying "Hey, new region, willy you take this avatar to be your lawfully rezed agent?"
  • [10:55] Tess Linden: you also have agent state that is transfered from one sim to another
  • [10:55] Zha Ewry: It says "Yeah, I can do that thing, here is a cap to use"
  • [10:55] Tess Linden: the agent domain isnt a part of that transaction
  • [10:55] Tess Linden: yep!
  • [10:55] Zha Ewry: No, but, in the worst case, we're still connect to the ad
  • [10:55] Tess Linden: yes
  • [10:55] Zha Ewry: so. we can regroup
  • [10:56] Tess Linden: we would have to make sure that sa definitely gives you up
  • [10:56] Zha Ewry: right
  • [10:57] Zha Ewry: That's a general issue in TP
  • [10:58] Teravus Ousley: ghost agents, yay
  • [10:58] Teravus Ousley: :D
  • [10:59] Zha Ewry: So..
  • [10:59] Zha Ewry: We get the new caop
  • [10:59] Zha Ewry: *cap
  • [10:59] Zha Ewry: We then close out the old one, and then, and only then try to use the new one?
  • [10:59] Tess Linden: Hm, sa will have to remove v before finalizing the transfer
  • [11:00] Zha Ewry: So
  • [11:00] Zha Ewry: SA tells the viewer "you're gone"
  • [11:01] Zha Ewry: and when it sees that happen, then it tells SB
  • [11:01] Zha Ewry: "Now go and connect to v?"
  • [11:01] Zha Ewry: Presumably
  • [11:01] Zha Ewry: If things break along the way
  • [11:01] Zha Ewry: we have several hanging bits of possible state
  • [11:01] Teravus Ousley: .. well, more likely it tells v to connect to sb, and then issues the packet to shut down the connection
  • [11:01] Tess Linden: I think sa doesnt have to tell v anything until the very end
  • [11:02] Zha Ewry: Its the one and only one, that is nervous makinbg
  • [11:02] Rex Cronon: i am trying to get to the havok4 office hour. sadly map tp doesn't work:(
  • [11:02] Tess Linden: just the responsibility is switched to the new sim
  • [11:02] Rex Cronon: bye everybody
  • [11:02] Zha Ewry: Bye Rex
  • [11:02] Tess Linden: sorry, I have to head off to another meeting
  • [11:02] Teravus Ousley: tc, Rex
  • [11:02] Zha Ewry: OK
  • [11:02] Zha Ewry: I think we can pick up with this tho
  • [11:02] Rex Cronon: have fun:)
  • [11:02] Enus Linden: smae here, offski
  • [11:02] Zha Ewry: And.. I think it was really good
  • [11:02] Enus Linden: ty zha
  • [11:02] Teravus Ousley: tc :D
  • [11:02] Tess Linden: thanks zha
  • [11:02] Zha Ewry: Gettig the names lined up
  • [11:02] Zha Ewry: and the messaging about the semantcis clear
  • [11:02] Zha Ewry: will really help
  • [11:03] Zha Ewry: Thanks all