User:Which Linden/Office Hours/2008 May 22

From Second Life Wiki
Jump to: navigation, search
  • [2008/05/22 11:03] Saijanai Kuhn: Hey which
  • [2008/05/22 11:03] Ruud Lathrop: A plant from the ground
  • [2008/05/22 11:03] Ruud Lathrop: Hi Which
  • [2008/05/22 11:03] Which Linden: good morning all
  • [2008/05/22 11:04] Which Linden: What is new?
  • [2008/05/22 11:04] Saijanai Kuhn: not too much. Enus went ahead and did a page for the Python test harnes
  • [2008/05/22 11:04] Which Linden: nice
  • [2008/05/22 11:04] Which Linden: Enus is awesome
  • [2008/05/22 11:04] Saijanai Kuhn: https://wiki.secondlife.com/wiki/AWG_Test_Harness
  • [2008/05/22 11:05] Saijanai Kuhn: so now I'm trying to figure out how to plob my UDP handler and my command parser into the eventlet strategy
  • [2008/05/22 11:05] Saijanai Kuhn: plop*
  • [2008/05/22 11:06] Which Linden: Have you seen the expanded eventlet documentation?
  • [2008/05/22 11:06] Saijanai Kuhn: don't think so
  • [2008/05/22 11:06] Which Linden: [1]
  • [2008/05/22 11:06] Gareth Ellison: eventlet has some nice generic replacements for the standard python sockets lib
  • [2008/05/22 11:07] Gareth Ellison: saijanai - that what you were talking about?
  • [2008/05/22 11:07] Which Linden: Yeah donovan has put out some nicer documentation so you don't have to be an code-reader to be able to use it
  • [2008/05/22 11:07] Which Linden: (well, less so anyhow)
  • [2008/05/22 11:07] Tao Takashi: Hi there
  • [2008/05/22 11:07] Gareth Ellison: so it's like VB now?
  • [2008/05/22 11:07] Gareth Ellison: lol
  • [2008/05/22 11:08] Saijanai Kuhn: yeah. Hoping to create a generic avatar thigie that would be a... meta-coprocess? COuld have dozens of test avatars running in teh same script
  • [2008/05/22 11:08] Which Linden: You mean a coroutine?
  • [2008/05/22 11:09] Saijanai Kuhn: each avatar would have UDP processing, EventQUeueGet processing, and reverse http processing for a command parser
  • [2008/05/22 11:09] Saijanai Kuhn: right, coroutine.
  • [2008/05/22 11:09] Which Linden: Yeah that is what eventlet and greenlets are all about
  • [2008/05/22 11:09] Tao Takashi: are we talking about test harness?
  • [2008/05/22 11:09] Gareth Ellison: seems like right now we're talking about test harness coding with eventlet
  • [2008/05/22 11:10] Saijanai Kuhn: the examples you give are of single socket setups per task. A test-harness needs several sockets per avie
  • [2008/05/22 11:10] Which Linden: We're talking about creating a bunch of bots, essentially, to log in to the agent domain, right?
  • [2008/05/22 11:10] Saijanai Kuhn: UDP/CAP/commands
  • [2008/05/22 11:10] Saijanai Kuhn: right, initially
  • [2008/05/22 11:10] Tao Takashi: we maybe should write down the goals of the test harness
  • [2008/05/22 11:10] Saijanai Kuhn: But I think the same harness could test group IM/inventory and any other non-graphics intensive set of protocols
  • [2008/05/22 11:11] Tao Takashi: is it for load tests or is it in general just to check if a component is OGP compliant?
  • [2008/05/22 11:11] Tao Takashi: or both
  • [2008/05/22 11:11] Which Linden: Hm, I understand that an av would need to accept udp packets but I don't understand why it would need more than one tcp socket
  • [2008/05/22 11:11] Gareth Ellison: i can - one per region and 1 per domain at least
  • [2008/05/22 11:11] Tao Takashi: but I guess also for non load tests you might need at least 2 avatars at some point
  • [2008/05/22 11:11] Saijanai Kuhn: well, was thinking that more than one comp might be involved. You'd add a command processor to the mix
  • [2008/05/22 11:12] Saijanai Kuhn: for a fully functional avie, you'd need 11 connections: 2 per sim, and 1 for commands
  • [2008/05/22 11:12] Gareth Ellison: where is the 1 for commands terminated?
  • [2008/05/22 11:14] Saijanai Kuhn: sorry SL froze
  • [2008/05/22 11:15] Which Linden: d'oh, client crash
  • [2008/05/22 11:15] Which Linden: did everyone else get hosed?
  • [2008/05/22 11:15] Saijanai Kuhn: if you want to handle more than a single test harnes, you need a command and control script separate from the test harness
  • [2008/05/22 11:16] Saijanai Kuhn: we all crash?
  • [2008/05/22 11:16] Which Linden: So, I saw the bit about the 11 connections -- I forgot about multiple sims
  • [2008/05/22 11:16] Which Linden: Yeah it appears that everyone crashed
  • [2008/05/22 11:16] Angelo Biondi: actually not
  • [2008/05/22 11:16] Angelo Biondi: i'm a lucky one alive
  • [2008/05/22 11:16] Which Linden: Hm, I wonder if this is the result of a bad texture or something
  • [2008/05/22 11:16] Which Linden: Sole survivor
  • [2008/05/22 11:16] Saijanai Kuhn: UDP and CAP per sim and one connection for external control if you want to have more than one harnes controlled by the same tester
  • [2008/05/22 11:16] Gareth Ellison: what happened?
  • [2008/05/22 11:17] Saijanai Kuhn: if the avie doesn't move, you only need 2 + 1 instead of 10 + 1
  • [2008/05/22 11:18] Which Linden: Though the control connection doesn't have to be continuous
  • [2008/05/22 11:18] Saijanai Kuhn: but eventually, I'd expect you to need to test TP and sim-crossing
  • [2008/05/22 11:18] Which Linden: It could just pop up whenever you wanted
  • [2008/05/22 11:18] Saijanai Kuhn: well, was thinking of a reverse http server connection to a control script. Start the test harness and it registers itself with the control script by posting a message to a known URL
  • [2008/05/22 11:19] Which Linden: Yeah ok, that's true
  • [2008/05/22 11:19] Which Linden: IF it's gotta be punching through a firewall
  • [2008/05/22 11:19] Saijanai Kuhn: that way you could have any arbitary number of harnesses controlled by one tester
  • [2008/05/22 11:19] Unable to: load gesture New Gesture.
  • Please try again.
  • [2008/05/22 11:22] Gesture New: Gesture is missing from database.
  • [2008/05/22 11:22] Gareth Ellison: seems to be a nice crashy day today
  • [2008/05/22 11:22] JayR Cela: hello everyone
  • [2008/05/22 11:22] Angelo Biondi: maybe my chair is lucky
  • [2008/05/22 11:22] Saijanai Kuhn: wondering if it IS a texture. Everyone is grey. By the time everyone loads, I crash again
  • [2008/05/22 11:22] Gareth Ellison: heh
  • [2008/05/22 11:23] Which Linden: christ on a cracker it's back to the release viewer for me
  • [2008/05/22 11:23] Gareth Ellison: well, don't look at me - my avatar is nothing but "looks like a lesbian"
  • [2008/05/22 11:23] Gareth Ellison: is the RC buggy?
  • [2008/05/22 11:23] Which Linden: I guess we'll find out
  • [2008/05/22 11:23] Saijanai Kuhn: well, for Mac PPC users its crash on startup
  • [2008/05/22 11:23] Saijanai Kuhn: hey enus
  • [2008/05/22 11:23] Gareth Ellison: if i crash again it's back to 1.19.1.4
  • [2008/05/22 11:23] JayR Cela: yeah camera control seems a bit flaky too
  • [2008/05/22 11:23] Which Linden: Hey there enus
  • [2008/05/22 11:23] Enus Linden: hi which
  • [2008/05/22 11:23] Kooky Jetaime: Gareth - what are you on?
  • [2008/05/22 11:24] Gareth Ellison: latest RC on linux
  • [2008/05/22 11:24] Enus Linden: RC7 has been pulled from the version manager...
  • [2008/05/22 11:24] Saijanai Kuhn: anyway, wsa just thinking that GUI might be easier to do with a single control point for all test harnesses, but obviously it could be optional or not used at all
  • [2008/05/22 11:24] Gareth Ellison: x86_64 on FC6
  • [2008/05/22 11:24] Kooky Jetaime: Oh... don't get me started.. RC's should be feature froze.. otherwise you will never stop rolling new RCs
  • [2008/05/22 11:24] JayR Cela: @Enus / RC7 was pulled ? / must of just happened
  • [2008/05/22 11:25] Saijanai Kuhn: I'm on the main trunk viewer and still get the same crash as everyone else
  • [2008/05/22 11:25] Which Linden: Oh, hm
  • [2008/05/22 11:25] Enus Linden: it happened recently. well, was supposed to've
  • [2008/05/22 11:25] Gareth Ellison: nobody grey here - and i haven't crashed
  • [2008/05/22 11:25] Saijanai Kuhn: I did a dump of the crash report to Apple. As usual, the SL crash reporter didn't start
  • [2008/05/22 11:25] JayR Cela: I just downloaded and set up the linux version 1/2 hour ago
  • [2008/05/22 11:25] Gareth Ellison: so if it is a abd texture then it's not affecting me
  • [2008/05/22 11:25] Gareth Ellison: *bad
  • [2008/05/22 11:26] Gareth Ellison: VERY laggy tough
  • [2008/05/22 11:26] Gareth Ellison: hence the lag typos
  • [2008/05/22 11:26] Saijanai Kuhn: yeah no grey here either. So much for tha ttheory
  • [2008/05/22 11:26] Enus Linden: we have plenty of crash reports from rc7 to work from :(
  • [2008/05/22 11:26] Which Linden: Ironically one of the things in the RCs is an improved crash reporter
  • [2008/05/22 11:26] JayR Cela: beleave it or not the windows client actually runs faster
  • [2008/05/22 11:26] Gareth Ellison: LIES
  • [2008/05/22 11:26] Kooky Jetaime: well the thing is
  • [2008/05/22 11:26] JayR Cela: usually the Linux client trounces the Windows version
  • [2008/05/22 11:27] Yohan Pintens: Which parts of the grid are chttp likely to be implemented in?
  • [2008/05/22 11:27] Saijanai Kuhn: anyway, Enus was explaining to WHich why I think a long-distance controller for the test harness might be a good option
  • [2008/05/22 11:27] Kooky Jetaime: LL needs to realize what RC symbolizes to almost anyone (outside LL) with regards to software release.
  • [2008/05/22 11:27] JayR Cela: @Kooky / your right about that
  • [2008/05/22 11:27] Which Linden: Yohan: probably the first place will be L$ transactions
  • [2008/05/22 11:27] Gareth Ellison: what kind of GUI?
  • [2008/05/22 11:28] Gareth Ellison: web interface or traditional GUI?
  • [2008/05/22 11:28] Saijanai Kuhn: command line, GUI, web page, whatever.
  • [2008/05/22 11:28] Kooky Jetaime: CQ said they were going to do something different, but apparently different = more of the same.
  • [2008/05/22 11:28] Saijanai Kuhn: just, if we want to test scaling to scary number levels, we gotta think BIG
  • [2008/05/22 11:28] Kooky Jetaime: and how does failure to buy L$ from the client pass QA?
  • [2008/05/22 11:28] Enus Linden: sure sai, i see it's value, but the libs and test framework are more important to me than their extended imnplementations
  • [2008/05/22 11:28] Yohan Pintens: so that will be all internal to the existing grid? (trying to guage the priority to add the support to libsl)
  • [2008/05/22 11:28] Which Linden: This is not the right office hours to talk about the release process, I'm afraid
  • [2008/05/22 11:28] JayR Cela: a meeting with Aric & Birdied may 5 this issue was brought up numerous times / and the official reply was that the RC Beta / Production client labeling was under intense internal LL disscussion
  • [2008/05/22 11:29] Enus Linden: but yes, a distributed test framework would be awesome
  • [2008/05/22 11:29] Which Linden: Yohan: yeah, for the moment. But it could crop up at any time during AWG (for example I proposed it for some step of the login)
  • [2008/05/22 11:29] Saijanai Kuhn: I think the eventlet strategy can handle it no problem
  • [2008/05/22 11:29] Saijanai Kuhn: and it would be a great test of reverse http
  • [2008/05/22 11:29] Saijanai Kuhn: but wouldn't be hard to add at a later date
  • [2008/05/22 11:30] Which Linden: Yeah, so I think that what you want to do is have the long polls/reverse http be handled by callbacks on an Avatar object
  • [2008/05/22 11:30] Saijanai Kuhn: If we work out the structure for handlign multiple sockets per test-avie, then adding a new socket is no biggie
  • [2008/05/22 11:30] Which Linden: I guess reverse http is a callback interface of a sort
  • [2008/05/22 11:30] Saijanai Kuhn: and whiteout. BLah
  • [2008/05/22 11:31] Saijanai Kuhn: if its not crashes, its the pure white scenary
  • [2008/05/22 11:32] Which Linden: I would expect that the pseudocode for that would be something like this: __init__(self): self.sim_ptth.append(api.spawn( reverse_http.do(self)))
  • [2008/05/22 11:33] Which Linden: and then you'd define a bunch of handle_foo methods on the Avatar object that would handle the reversified requests
  • [2008/05/22 11:33] Saijanai Kuhn: right. ANd the same thing for an external command if we get that far
  • [2008/05/22 11:34] Which Linden: Yeah exactly
  • [2008/05/22 11:34] Angelo Biondi: i gotta go, nice to meet you all
  • [2008/05/22 11:34] Which Linden: Seeya angelo
  • [2008/05/22 11:34] JayR Cela: bye Angelo
  • [2008/05/22 11:34] Dahlia Trimble: bye :)
  • [2008/05/22 11:35] Angelo Biondi: bye all
  • [2008/05/22 11:35] Which Linden: me looks up the ptth startup command
  • [2008/05/22 11:35] Which Linden: d'oh that irc-ism didn't survive the journy
  • [2008/05/22 11:35] Gareth Ellison: you missed the / :)
  • [2008/05/22 11:35] Saijanai Kuhn: the alternative, more complicated but uses less sockets, would be 1 command socket per test-harness and just have some dispatch mechanism internally
  • [2008/05/22 11:36] Gareth Ellison: i'd favour that
  • [2008/05/22 11:36] Gareth Ellison: with a sexy mulib setup
  • [2008/05/22 11:36] Gareth Ellison: root = {'bot1':bot1_handler,...}
  • [2008/05/22 11:36] Which Linden: Hm that's not bad
  • [2008/05/22 11:36] Saijanai Kuhn: harder to program, but definitely easier on the CPU, especialy if we have hundreds or thousands of test-avies per machine
  • [2008/05/22 11:37] Gareth Ellison: these handle_foo calls - would they be HTTP methods or what?
  • [2008/05/22 11:37] Which Linden: I think Gareth demonstrated that it wouldn't be that hard to cod
  • [2008/05/22 11:37] Which Linden: e
  • [2008/05/22 11:37] Gareth Ellison: i'd be inclined towards dumping an LLSD map
  • [2008/05/22 11:37] Saijanai Kuhn: some xml thingie like a predefined LLSD map...
  • [2008/05/22 11:37] Gareth Ellison: with one field specificying the method name
  • [2008/05/22 11:37] Gareth Ellison: oh yeah saijanai
  • [2008/05/22 11:38] Which Linden: You lost me with the llsd map
  • [2008/05/22 11:38] Gareth Ellison: we're getting teh sexy RPC setup now
  • [2008/05/22 11:38] Gareth Ellison: which - an LLSD map with method_name field
  • [2008/05/22 11:38] Gareth Ellison: and then some params
  • [2008/05/22 11:38] Gareth Ellison: or just 2 fields, method and parmas
  • [2008/05/22 11:38] Gareth Ellison: *params
  • [2008/05/22 11:38] JayR Cela: LSD ???? / :_) geeze I gave up using that stuff in the 1970's with the rest of the hippies / LOL :_)))
  • [2008/05/22 11:38] Gareth Ellison: ugh, lag typos
  • [2008/05/22 11:38] Which Linden: Um, isn't that what reverse http is for? Specifying the http method?
  • [2008/05/22 11:39] Gareth Ellison: i meant the method as in the function we're calling, not raw HTTP GET/POST/whatever
  • [2008/05/22 11:39] Gareth Ellison: like tp_to_sim()
  • [2008/05/22 11:39] Saijanai Kuhn: well, you wouldn't use a different method for each command or each incoming EQG packet
  • [2008/05/22 11:39] Gareth Ellison: i would
  • [2008/05/22 11:39] Which Linden: Oh, yeah, I've taken to calling those 'actions'
  • [2008/05/22 11:39] Gareth Ellison: you wouldn't, i would
  • [2008/05/22 11:39] Gareth Ellison: ok, actions
  • [2008/05/22 11:39] Gareth Ellison: (they're methods of the python object of course)
  • [2008/05/22 11:39] Which Linden: Sure, yeah
  • [2008/05/22 11:39] Saijanai Kuhn: ok, overladed term. Thought you meant POST vs something else
  • [2008/05/22 11:40] Gareth Ellison: perhaps call it do_foo instead of handle_foo
  • [2008/05/22 11:40] Which Linden: Though, if it's possible to do it via a REST method, that's preferable
  • [2008/05/22 11:41] Saijanai Kuhn: thought POST would be REST
  • [2008/05/22 11:41] Which Linden: Like, instead of POST /actions {'action':'tp_to_sim'} you do PUT /location/sim {'new_sim':<url>}
  • [2008/05/22 11:41] Which Linden: Yeah you're right Sai.
  • [2008/05/22 11:42] Saijanai Kuhn: Whichever, I'm not aware of the distinctions enough to worry.
  • [2008/05/22 11:42] Which Linden: But the benefit of rest is that it has very few verbs that have specific meaning
  • [2008/05/22 11:42] Gareth Ellison: that scheme which said works
  • [2008/05/22 11:42] Saijanai Kuhn: so PUT /do_thing/parameters
  • [2008/05/22 11:43] Which Linden: But I agree that it's very hard, when designing something like this that's very imperative in nature, to make it map well to REST verbs
  • [2008/05/22 11:43] Gareth Ellison: PUT /location works
  • [2008/05/22 11:43] Gareth Ellison: we're using state machines people - edit state over REST
  • [2008/05/22 11:43] Saijanai Kuhn: but would other comman activities of the test harness work that way?
  • [2008/05/22 11:43] Lalinda Lovell: are graphics very bad here for anyone else?
  • [2008/05/22 11:43] Which Linden: Dunno, what other things might it be doing?
  • [2008/05/22 11:44] Gareth Ellison: all these RPC functions are just editing state
  • [2008/05/22 11:44] Which Linden: Lalinda: yeah my bamboo is not friends with nvidia
  • [2008/05/22 11:44] Saijanai Kuhn: not sure,e xactply. We're talking TP and login/logout. Inventory control and group IM eventually, I'm sure
  • [2008/05/22 11:44] Lalinda Lovell: yes i think thats the problem
  • [2008/05/22 11:44] Saijanai Kuhn: because those are heading to Agent Domain
  • [2008/05/22 11:45] Which Linden: So, Sai, it might be helpful to think of things in terms of the end results. I.e. it's not that you want to tp somewhere, it's that you want to be at the new location
  • [2008/05/22 11:45] Saijanai Kuhn: would eventually want to look at sim-crossing
  • [2008/05/22 11:45] Which Linden: And it's not that you want to be logged out it's that you want your status to be offline
  • [2008/05/22 11:45] Lalinda Lovell: pulls off a leaf and chews on it
  • [2008/05/22 11:45] Which Linden: That's kind of the mental state you have to be in to develop REST apis. :-)
  • [2008/05/22 11:46] Which Linden: I am only a novice at it though
  • [2008/05/22 11:46] Gareth Ellison: as i said - we're using state machines
  • [2008/05/22 11:46] Gareth Ellison: so manipulate state via REST
  • [2008/05/22 11:46] Saijanai Kuhn: movement, communication, inventry (not sure what that would be .. inventory?)
  • [2008/05/22 11:47] Which Linden: Hm, yeah, what can you do with inventory -- you can list the contents, and you can rez and derez things
  • [2008/05/22 11:47] Saijanai Kuhn: hand to other people
  • [2008/05/22 11:47] Saijanai Kuhn: delete
  • [2008/05/22 11:47] Lalinda Lovell: move to trash
  • [2008/05/22 11:47] Saijanai Kuhn: rename
  • [2008/05/22 11:47] Gareth Ellison: how do we implement a MOVE over normal REST stuff?
  • [2008/05/22 11:47] Tao Takashi: btw, Enus, that login is still not working for me
  • [2008/05/22 11:47] Lalinda Lovell: make a copy within inventory
  • [2008/05/22 11:47] Gareth Ellison: seems like standard HTTP not do that
  • [2008/05/22 11:48] Gareth Ellison: DAV does
  • [2008/05/22 11:48] Gareth Ellison: the future is DAV
  • [2008/05/22 11:48] Lalinda Lovell: on a totally random note, has anyone seen kaylee lindens avi?
  • [2008/05/22 11:48] Which Linden: Well you could replicate the hierarchy of the inventory via rest
  • [2008/05/22 11:48] Lalinda Lovell: she turned me lesbian in 00.3 seconds
  • [2008/05/22 11:48] Which Linden: Yeah, DAV is a solution
  • [2008/05/22 11:48] Gareth Ellison: which - but how do you do MOVE over normal HTTP?
  • [2008/05/22 11:49] Gareth Ellison: DAV is messier
  • [2008/05/22 11:49] Gareth Ellison: sexy but messy
  • [2008/05/22 11:49] Which Linden: Um, hm, yeah, you kinda gotta wedge MOVE in there dontcha
  • [2008/05/22 11:49] Gareth Ellison: put MOVE into normal HTTP?
  • [2008/05/22 11:49] Which Linden: No
  • [2008/05/22 11:49] Gareth Ellison: the IETF will slaughter you with a chainsaw if you do that
  • [2008/05/22 11:49] Which Linden: Hm
  • [2008/05/22 11:49] Gareth Ellison: en masse
  • [2008/05/22 11:50] Gareth Ellison: thetexas linden massacre
  • [2008/05/22 11:50] Which Linden: The chainsaw will already be running at the mention of reverse http. :-)
  • [2008/05/22 11:50] Enus Linden: Tao, i hope to be free to look at that login tomorrow
  • [2008/05/22 11:50] Gareth Ellison: or is that the IETF linden massacre
  • [2008/05/22 11:50] Saijanai Kuhn: what, nonstop keyproesses shouldn't be POSTed
  • [2008/05/22 11:50] Gareth Ellison: heh - good point
  • [2008/05/22 11:50] Tao Takashi: you could say you are not using HTTP but DAV ;-)
  • [2008/05/22 11:50] Gareth Ellison: reverse HTTP is kinda crazy
  • [2008/05/22 11:50] Which Linden: A DAV subset
  • [2008/05/22 11:50] Tao Takashi: LLDAV
  • [2008/05/22 11:50] Saijanai Kuhn: Tao, login STILL not working?
  • [2008/05/22 11:50] Gareth Ellison: hey, we have LLString
  • [2008/05/22 11:51] Gareth Ellison: complete with memory leaks
  • [2008/05/22 11:51] Gareth Ellison: and my fave LLCrashAndLoop
  • [2008/05/22 11:51] Which Linden: It's crashing and looping for your protection, don't you know?
  • [2008/05/22 11:51] Lalinda Lovell: guns dont kill people, people kill people
  • [2008/05/22 11:51] Gareth Ellison: llcommon - cos std namespace isn't enough
  • [2008/05/22 11:52] Which Linden: Cause std namespace didn't exist and/or was hella buggy at the time of creation
  • [2008/05/22 11:52] Gareth Ellison: wtf
  • [2008/05/22 11:53] Which Linden: ftw :-)
  • [2008/05/22 11:53] Gareth Ellison: truely LLSD is Linden LSD
  • [2008/05/22 11:53] Tao Takashi: Sai: <?xml version="1.0" ?><llsd><map><key>reason</key><string>bad identity and authentication information</string><key>authenticated</k
  • [2008/05/22 11:53] Gareth Ellison: do we have new non-first/last name credentials now on beta grid?
  • [2008/05/22 11:53] Gareth Ellison: cos that'd be sexy
  • [2008/05/22 11:54] Tao Takashi: Enus: thanks, no hurry though. I personally can also wait for the next update of that script
  • [2008/05/22 11:54] Which Linden: I'm a little sad at the proliferation of authentication mechanisms
  • [2008/05/22 11:55] Which Linden: But it seems that there is no one-size-fits-all solution for that sort of thing
  • [2008/05/22 11:55] Which Linden: Except KERBEROS.
  • [2008/05/22 11:55] Which Linden: Fits. Everything.
  • [2008/05/22 11:55] Saijanai Kuhn: well, this particular one we asked for specifically: no excess info given or sent. Make sit more generic with foreign grids
  • [2008/05/22 11:56] Saijanai Kuhn: unlike the current xml-rpc login
  • [2008/05/22 11:56] Which Linden: Heh, yeah.
  • [2008/05/22 11:56] Which Linden: I think pruning that junk is Very Good(tm)
  • [2008/05/22 11:56] Gareth Ellison: heh, i hacked up LLSD login months and months ago
  • [2008/05/22 11:56] Gareth Ellison: and nobody praised me
  • [2008/05/22 11:56] Gareth Ellison: i just didn't have the sim connection
  • [2008/05/22 11:56] Enus Linden: praise be to Gareth
  • [2008/05/22 11:56] Enus Linden: and Sai
  • [2008/05/22 11:56] Enus Linden: and Tao
  • [2008/05/22 11:56] Enus Linden: and
  • [2008/05/22 11:56] Enus Linden: and
  • [2008/05/22 11:56] Which Linden: hallelujah amen
  • [2008/05/22 11:56] Gareth Ellison: but i had a viewer with LLSD login and python scripting clientside
  • [2008/05/22 11:57] Which Linden: How did you hack that, the llsd login turned around and did xml-rpc on the backend?
  • [2008/05/22 11:57] Gareth Ellison: no, with my own mulib-based services
  • [2008/05/22 11:57] Gareth Ellison: including agent domain and region domain
  • [2008/05/22 11:57] Which Linden: Oh, nice
  • [2008/05/22 11:57] Gareth Ellison: had a special API defined for opensim i tried to convince mr zaius to implement
  • [2008/05/22 11:57] Gareth Ellison: may still have the code somewhere
  • [2008/05/22 11:59] Which Linden: Cool beans
  • [2008/05/22 12:00] Enus Linden: tao, what are getting back from auth.cgi now?
  • [2008/05/22 12:01] Tao Takashi: <?xml version="1.0" ?><llsd><map><key>reason</key><string>bad identity and authentication information</string><key>authenticated</k...
  • [2008/05/22 12:01] Tao Takashi: maybe my avatar is simply not known ;-)
  • [2008/05/22 12:01] Tao Takashi: but I think that gives a different error then
  • [2008/05/22 12:02] Enus Linden: well, you're logging into aditi... if you can login with a client, should be able to login via ad
  • [2008/05/22 12:02] Tao Takashi: I use '$1$'+md5.new('some4451').hexdigest() as password
  • [2008/05/22 12:02] Tao Takashi: now I have to change my pw again ;-)
  • [2008/05/22 12:03] Enus Linden: afaik, password is unecrypted password
  • [2008/05/22 12:03] Tao Takashi: really?
  • [2008/05/22 12:03] Enus Linden: md5-password can be substituted for password
  • [2008/05/22 12:03] Tao Takashi: ah, I tried this aswell
  • [2008/05/22 12:03] Which Linden: Shoot, I'm totally distracted today. I have to depart now, but thank you all for coming.
  • [2008/05/22 12:04] Which Linden: It is always nice to talk. :-)
  • [2008/05/22 12:04] Saijanai Kuhn: is there a public code repository for the client that can login to the agent domain?
  • [2008/05/22 12:04] Enus Linden: bye which
  • [2008/05/22 12:04] Saijanai Kuhn: later which
  • Saijanai