AW Groupies/Chat Logs/AWGroupies-2008-01-08

From Second Life Wiki
Jump to: navigation, search
[9:33] Zha Ewry is Online
[9:34] Teravus Ousley: mild indeed
[9:34] Zha Ewry: Ok
[9:34] Zha Ewry: Five minutes past call time
[9:34] Zha Ewry: So.. I posted three, items
[9:34] Zha Ewry: next f2f
[9:34] Dr Scofield: hi all
[9:35] Zha Ewry: Protocol docoumentatoin mess
[9:35] JayR Cela: i have a question
[9:35] Zha Ewry: and what, the OpenSim community needs to get from Linden if there isgoing to be any progress
[9:35] Adam Xinpeng: Zha: anything.
[9:35] Zha Ewry: Sure JayR
[9:35] Adam Xinpeng: We've made so far pretty good with no documentation - but we are going to need some if we want to tackle grid work.
[9:35] Dr Scofield thinks whoever did the web based login stuff deserves to be slapped with an old fish...HTML page would not load until the third attempt
[9:35] Zha Ewry: Heh. Adam you'd like to see a pulse?
[9:35] JayR Cela: why do we allways save textures in .bmp instad of .jpg
[9:35] Gigs Taggart: JayR I have a patch for that, I filed it months ago
[9:36] Adam Xinpeng: Zha: Client work, we can mostly do ourselves without too much hassle, we've got good at that (by nessecity)
[9:36] Gigs Taggart: LL hasn't applied it
[9:36] Adam Xinpeng: But, server-side work, and probably some of the CAPS stuff too is what needs documenting.
[9:36] Adam Xinpeng: What the sim needs to do to conform to LL's grandoise standard plans.
[9:36] Zha Ewry: Agreed
[9:36] Teravus Ousley: I might point out that there's also bug: VWR-4021 that kind of needs to be resolved
[9:36] Tao Takashi: Hi everybody
[9:36] Adam Xinpeng: CAPS is probably the first area, since it doesnt need too much help from LL.
[9:36] Zha Ewry: And.. its worse than that
[9:36] JenzZa Misfit: Alanagh Recreant would like a group invite , Sai ..if possible
[9:36] JayR Cela: ok thats good but it needs to be implemented into the official client
[9:36] Zha Ewry: As far as I can tell, Niether OpenSim, nor LibSL have fully caught up to the caps stuff
[9:37] Dr Scofield: main roblem is that we (the group) are wasting our time reverse engineering LL grid
[9:37] Adam Xinpeng: OpenSim is partly by choice
[9:37] Adam Xinpeng: A lot of the OpenSim devs take objection to some of the CAPS
[9:37] Adam Xinpeng: notably, our favourite EventQueueGet
[9:37] Zha Ewry: The totaly kichen sink approach?
[9:37] Dr Scofield: instead of LL providing the meat so that we can then build on that
[9:37] Adam Xinpeng: That too.
[9:38] Tao Takashi: what parts of CAPS needs to be done for opensim? implementing a caps server properly?
[9:38] Zha Ewry: After watching Siaj chew through the process, i'd have to say that we need to ask Linden to a) document some of this
[9:38] Adam Xinpeng: We've got a CAPS server already
[9:38] Zha Ewry: and.. b) clean up some of the most egrigious crap
[9:38] Saijanai Kuhn loves EventGetQueue mostly because I can't get it to work
[9:38] Adam Xinpeng: and we've got a decent LLSD parser, etc.
[9:38] JayR Cela: if not for my wife Cathy I would leave this place and goto Open Sim
[9:38] Adam Xinpeng: What we dont have is documentation for the individual caps, their parameters, etc.
[9:39] Zha Ewry: What concerns me about things like not doing cpas.. is
[9:39] Zha Ewry: *caps
[9:39] Zha Ewry: that we'll end up haviing to fork the client soonner or later
[9:39] JayR Cela: ok i am confused / what is caps
[9:39] Zha Ewry: Of course, if Linden can't allow authentication to be fixed
[9:39] Adam Xinpeng: Oh,you do know there's a project starting up to build a new client from the ground up as a reimplementation along the lines of OpenSim/libsl.
[9:39] Teravus Ousley: indeed, not having the documentation on the individual caps is, of course, shooting in the dark.
[9:39] Saijanai Kuhn: the special web address used to communicate with the server
[9:39] Adam Xinpeng: OpenViewer[.org]
[9:40] Zha Ewry: that will happenwhen they force the update
[9:40] Saijanai Kuhn: it has a special format
[9:40] Zha Ewry: to the new auth
[9:40] Zha Ewry: And.. Yes. I saw that
[9:40] Saijanai Kuhn: and is called a cap as short for "capability"
[9:40] Day Oh: I found slproxy can proxy caps now, that should be really useful
[9:40] Saijanai Kuhn: can't get it to work with MacOS X though
[9:40] Zha Ewry: Couldn't be worse than the Linden on sturcture
[9:40] Dr Scofield: we pretty much don't have the docs on any grid internal stuff...which is going to make the LL--OpenSim grid interop a tad difficult
[9:40] Tao Takashi: I described CAPS here: http://mrtopf.de/blog/secondlife/slga-capabilities-explained-technical/
[9:41] Dr Scofield: libsl seems to do CAPs sufficiently well
[9:41] Adam Xinpeng: Dr: yeah, that's what would be the ultimate documentation
[9:41] Zha Ewry: But.. I expect that the rendeing isn't going to be any fun. (OpenGL, mlutiple shader models, multiple graphics card bugs, is a nasty thing)
[9:41] Burhop Piccard: yes
[9:41] JayR Cela: i about ready to dump everything and just use Ubuntu Linux
[9:41] Tao Takashi: I wonder if the documentation should not be done anyway along the SLGA process.. I'd think that we need different caps for a start anyway
[9:41] Tao Takashi: this of course raises the question of who is doing that documentation
[9:42] Saijanai Kuhn reasises his hand
[9:42] Zha Ewry: Well.. Part of the problem, is that as far as I can tell, Linden is about two months behind thier intended schedule
[9:42] Saijanai Kuhn: as far as I know, I'm the main documenter of SL right now
[9:42] Dr Scofield: you need some inside knowledge to do that
[9:42] Zha Ewry: and.. has not moved very much on the agent/sim split
[9:42] Saijanai Kuhn: and I'm stuck
[9:42] Dr Scofield: otherwise it's quite a lot of guesswork
[9:43] Media Hax: im a good guesser...
[9:43] Dr Scofield applauds saijanai for his effort!
[9:43] Zha Ewry: I also am willing to argue, that if they can't get us a stable, working version of the new-auth we can sue for OpenSim login, that they won't be able to evbolve the client to anything the AWG comes up with
[9:43] Saijanai Kuhn: on a bright note, I might be getting insider info soon.
[9:43] Dr Scofield: right
[9:43] Zha Ewry: Yeah, but Saij, they will take away your access to doocumenting
[9:43] Dr Scofield: i assume you meant use not sue? ;-)
[9:44] Saijanai Kuhn: gawd talk about shooting themselves in th foot if that happens
[9:44] Dr Scofield: saijanai linden?
[9:44] JayR Cela: oh they cant sue any of us / is OS
[9:44] Saijanai Kuhn: maybe (hopeful)
[9:44] Zha Ewry: So...
[9:44] Ina Centaur: lol
[9:44] Dr Scofield: hey, hey! cool!
[9:44] Zha Ewry: In theory, Still on Zero's page
[9:44] Zha Ewry: there is a f2f comming up
[9:45] Tao Takashi: was their any response to our complaints about the AWG process and collaboration/communication issues?
[9:45] Zha Ewry: Oooh. Inbound Lindens
[9:45] Tao Takashi: I think the caps problem again comes down to that problem
[9:45] Media Hax: I hope he didnt mean sue, cause Jennza is not an attorney....
[9:45] Ina Centaur: :-O
[9:45] Adam Xinpeng: Shiny
[9:45] Adam Xinpeng: which ones?
[9:45] Dr Scofield: the question is what sense does a f2f make if they don't participate in the SL AWG process ?
[9:45] Saijanai Kuhn: Zero told us to talk to people on IRC
[9:45] Saijanai Kuhn: adn they've tried to be helpful but it hasn't always helped
[9:45] JenzZa Misfit: media ! *snicker*
[9:46] Zha Ewry: Periapse, and Tess
[9:46] Saijanai Kuhn: hey guys. Welcomde
[9:46] Saijanai Kuhn: First LIndens
[9:46] Adam Xinpeng: ack, shame I have a meeting in 15 mins.
[9:46] Periapse Linden: thx
[9:46] Tao Takashi: Hi Periapse, Tess!
[9:46] Zha Ewry: Hello Lindens
[9:46] Teravus Ousley: greetings
[9:46] Zha Ewry: Welcome to snowy AWG land
[9:46] Lom Hax: Speaking of "they." :)
[9:46] JayR Cela: well ya go a whole truckload full of X86 servers / and they are screwed up
[9:46] Adam Xinpeng: Hey, we should host one of these AWG meetings on OpenSim some time. ;)
[9:46] Zha Ewry: Well, They.. can't be the right model
[9:46] Zha Ewry: And.. Yes, we should
[9:46] Adam Xinpeng: Cant let llSim hog all the meetings. :P
[9:47] Day Oh: Great idea
[9:47] Zha Ewry: Tho.. We won't be quite so well dressed, most of us
[9:47] Adam Xinpeng: True
[9:47] Adam Xinpeng: although that's been mostly fixed, inventory is running now
[9:47] Zha Ewry: yeah, content needs to stagger in
[9:47] JayR Cela: we need to move the platform to a main Frame
[9:47] Tess Linden: hey guys!
[9:47] Zha Ewry: So... The biggest impediment, is clearly, the documentation side
[9:47] Dr Scofield: hi tess
[9:47] Adam Xinpeng: Hey Tess
[9:47] Day Oh: Howdy
[9:47] Teravus Ousley: Greetings Tess
[9:48] Zha Ewry: And. yes, it has to be us, not we vs. them
[9:48] Adam Xinpeng: Right, I think the things that would be good re: documentation would be:
[9:48] Adam Xinpeng: 1. Serverside connectivity documentation & region-domain implementation specs. Even if seriously incomplete.
[9:48] Adam Xinpeng: 2. CAPS documentation for client connections (AWGroupies & users can do this one)
[9:48] Zha Ewry Nods
[9:48] Adam Xinpeng: #1 is somewhat important, even if incomplete so we know which direction to head in
[9:49] Saijanai Kuhn justwnats help implementing Group IM and maybe a pointer on getting his Python proxy working
[9:49] Zha Ewry: Tho.. Some coherence on the core evolution of the caps/queue infrastructure would be awful nice
[9:49] JayR Cela: i think is best if Sun Micro or IBM just buy LL
[9:49] Adam Xinpeng: Not sun plz.
[9:49] FWord Utorid: lag
[9:49] Ina Centaur: lol
[9:49] Zha Ewry: The tangle between the caps, and udp, at the moment is painful to sort out
[9:49] Dr Scofield: jayr, no, don't think you want that
[9:49] Zha Ewry: (send X on one pipe, see responses on other pipes)
[9:49] Dr Scofield: eclipse rich client as a viewer base? shudder
[9:49] Burhop Piccard: JayR - I've thought the same only 1/2 in a joking way.
[9:49] JayR Cela: well they are both big OS supporters
[9:50] Ina Centaur: eclipse for lsl editing
[9:50] Tao Takashi: but one is too Java centric ;-)
[9:50] Dr Scofield: lol
[9:50] Tao Takashi: at least for me
[9:50] Tao Takashi: anyway
[9:50] Adam Xinpeng: Sun is very GPL-centric. More than LL.
[9:50] Zha Ewry: Which one Tao?
[9:50] Lom Hax uses TextMate for LSL
[9:50] Dr Scofield: tao, just one?
[9:50] Adam Xinpeng: (although LL is pretty bad.)
[9:50] Ina Centaur: :-P
[9:50] Catherine Pfeffer: Sun already has its own frid initiatives
[9:50] Catherine Pfeffer: grid
[9:50] Tao Takashi: Yes, I interviewed some SUN guys about Project Wonderland
[9:50] FWord Utorid: catchphrase is too corpocentric
[9:50] Zha Ewry: so.. Adam, I agree
[9:50] Media Hax: yes Sun has some great 3D tools..
[9:51] Tao Takashi: actually quite nice
[9:51] Zha Ewry: Some sort of a roadmap on the sim/service front would be very nice
[9:51] Burhop Piccard: Tao... me too.
[9:51] Saijanai Kuhn: OK as the Doucmenter in the Field, I'd like help getting started with EventQueueGet and the overall CAPs docs
[9:51] Adam Xinpeng: Actually, I'd probably add a "3." -- a list of things where OpenSim and other reimplementers can help out. Either feasibility testing or otherwise.
[9:51] Saijanai Kuhn: Kelly tried to help me the other day, but still cant get an arbitrary event out of EQG
[9:51] JayR Cela: well we all know this LL is not going to last forever
[9:52] FWord Utorid: as the daring one in the group, i'd like help discarding the entire thing and making something new, better, and functional.
[9:52] Burhop Piccard: Adam, I think that is hard.
[9:52] FWord Utorid: instead of writing a free manual for LL ;)
[9:52] Adam Xinpeng: FWord: I agree on lots of parts of the protocol as it is - I'm hoping that part of the AWG process will annihilate those bits.
[9:52] JayR Cela: i guess we will all end up in Open Simm
[9:52] FWord Utorid: adam: i didn't ask you ;)
[9:53] Adam Xinpeng: EQG is a good example, there are lots of others (the UDP resend system, asset download system, etc.)
[9:53] Saijanai Kuhn: After the Fast FOod Wars, all restaurants are Taco Bell
[9:53] Zha Ewry is Offline
[9:53] Saijanai Kuhn: er, OpenSim
[9:53] Adam Xinpeng: Haha
[9:53] Dr Scofield: zha just crashed, will be back in a tick, she said
[9:53] Teravus Ousley: as I said before, I don't have an issue suring up documentation as I find it lacking and I'm working on that particular portion's re-implementation... (and I've done so in the past)
[9:53] Rex Cronon: hi everybody
[9:54] JayR Cela: oh god / UDP / why the heck we even use UDP is an embarassement to everyones intelligence
[9:54] Adam Xinpeng: Yeah, agree with Teravus - what we need is documentation where it's not possible to R.E. it.
[9:54] Adam Xinpeng: UDP makes a lot of sense for lots of SL's traffic
[9:54] JayR Cela: screw UDP
[9:54] Zha Ewry is Online
[9:54] Saijanai Kuhn: well, we official docs everywhere
[9:54] Adam Xinpeng: particularly agent & object updates where the data goes stale quickly.
[9:54] Rex Cronon: something wrong with the main grid? a few minutes ago i crashed, and had to reset pc:(
[9:54] Zha Ewry: Sigh
[9:54] Burhop Piccard: FWord - I think there is room for both. I think AWG is on the evolve track. I haven't seen much on the "Revolution" track (open Virtual worlds based on standards)
[9:54] Saijanai Kuhn: because otherwise they can change without notice and no-one can say "but the docs show differently"
[9:54] FWord Utorid: given how slow everything in sl is i totally think the existing architecture is defunct.
[9:54] Adam Xinpeng: But, when it needs to be reliable - the UDP resend system should be replaced with raw TCP.
[9:54] Zha Ewry: Onwn fault that one
[9:54] JayR Cela: UDP creates way tooooooooo much IP traffic
[9:54] Dr Scofield: the problem i see is that non-LL folks are in the worst possible position to doc
[9:55] Zha Ewry has to remember not to drag Windlight windows between X sessions
[9:55] FWord Utorid: burhop, i have begun the revolution, and I AM REVOLTING
[9:55] Saijanai Kuhn nods vigorously
[9:55] Adam Xinpeng: UDP is smaller than TCP, especially for things where your streaming updates and you can throw packets away safely.
[9:55] Adam Xinpeng: which makes it great for things like streaming video, streaming object updates, etc.
[9:55] Adam Xinpeng: that are temporally bound
[9:55] Burhop Piccard: THen I shall join you, FWord!
[9:55] Tao Takashi: Zero nevertheless said that he wanted to test at least a grid without UDP or at least without most of it
[9:55] Lom Hax: Non-LL folks may be in a poor position to document, but every working programmer knows it is hard to work on documentation when the bullets are flying.
[9:55] Zha Ewry: UDP makes perfect sense for a set of things, which are basically transtory
[9:56] Adam Xinpeng: I dont think it will be a good idea to throw it away completely, but for the majority of SL packet types, like IM, etc - definetely.
[9:56] Tao Takashi: but I see my area mostly in the agent domain anyway so hopefully I won't have to deal with UDP
[9:56] JayR Cela: But UDP is no message correection / so we got a ll these bits bouncining back and forth
[9:56] Teravus Ousley: add textures to that
[9:56] Zha Ewry: But.. the currrent tangleof UDP/TCP and such is hard to untangle
[9:56] Lom Hax: Meaning that LL folks will have a hard time doing that work when theic bosses expect them to be delivering actual code.
[9:56] Saijanai Kuhn: I've been told its not the UDP that is the problem with IM
[9:56] Adam Xinpeng: Jayr: it doesnt need resend for temporally bound stuff
[9:56] Adam Xinpeng: IE, let's say I send two frames of video
[9:56] Adam Xinpeng: If we've already got the second one, why re-send the first?
[9:57] Zha Ewry nods at Adam
[9:57] Adam Xinpeng: The same theory applies to object updates, if there's two updates - one later than the other, then we dont care about the old one, it's stale data already
[9:57] Dahlia Trimble: UDP saves server memory so the TCP /IP stack doesnt need to store undeliverable packets
[9:57] Lom Hax: So, to whatever extent we can receive docs from LL, great, but we need to be prepared to do some outselves. Hopefully we can show to LL and get good feedback.
[9:57] Zha Ewry: When that's what we're doing, UDP is appropriate
[9:57] FWord Utorid: THIS CUBE SHOWS SCREENSHOTS OF THE VIEWER I HAVE BEEN WORKING ON WITH CAPSLOCK ON FOR ALMOST A YEAR
[9:57] Tess Linden: Zha, we have some internal documents/diagrams that we think may help you understand how the message system works, I'll work with peri to get it into the public wiki
[9:57] Catherine Pfeffer: Good point Adam
[9:57] Day Oh: Do we need to come up with another channel for asking for help while documenting things, or is SLDev enough?
[9:57] Zha Ewry: Tess, that woudl be super
[9:57] Lom Hax: Tess, tha's great
[9:57] Lom Hax: Thanks so much.
[9:57] Saijanai Kuhn: wonderful
[9:57] Adam Xinpeng: Tess: yeah, definetely appreciated.
[9:57] Tao Takashi: great, Tess :)
[9:57] Tess Linden: it may not solve all problems, but will get you something to look at and I will be available on IRC for questions
[9:58] FWord Utorid: after writing it i decided that all technology is crap
[9:58] Zha Ewry: One of the challnages is that as far as I can tell, Nobody, iniside, or outside Linden could actually write down what it all does
[9:58] Zha Ewry: Three years of evolving code, and clearly some odd things here and there
[9:58] Dr Scofield: tess, excellent!
[9:58] Adam Xinpeng: OK, I need to fly. Good seeing you all again, hopefully will have some time to attend next week. :)
[9:58] Zha Ewry: But.. if we can get it closer to understood, it would really help
[9:58] Zha Ewry: Good to see you adam
[9:58] Tess Linden: this stuff is hard to understand, we had to give several Linden University talks to get internal developers to understand it at the time, so getting the diagrams out and communicating will be key
[9:59] Teravus Ousley: tc Adam
[9:59] Zha Ewry: You likely to make the end of th month?
[9:59] Dr Scofield: take care, adam
[9:59] Tess Linden: yep!
[9:59] Adam Xinpeng: Zha: if I'm invited, yes.
[9:59] Zha Ewry nods
[9:59] Zha Ewry: Well, that n eeds to go on someone's list
[9:59] Adam Xinpeng: all the opensim devs will be in Finland to the 25th/26th, so hopefully it wont be the 27th.
[9:59] Zha Ewry: Its getting to be time to buy tickets
[9:59] Zha Ewry: They said the 31st
[9:59] FWord Utorid: i would like to thank LL for a year of crashes and stressful problems, btw
[9:59] Adam Xinpeng: OK plenty of time to fly back.
[9:59] Adam Xinpeng: I'll probulate Rob.
[9:59] Dr Scofield: how about having it in finland then?
[10:00] JayR Cela: ok / bye~byee everyone / and Sai / thank you for sending me a Taxi :_)
[10:00] Zha Ewry: Get your annual dose of polar Radiation Adam?
[10:00] Saijanai Kuhn: ;-)
[10:00] Adam Xinpeng: Yeah, we've got a small opensim-dev-meet thing going on.
[10:00] Burhop Piccard: bad day, Fword ;-)
[10:00] FWord Utorid: bad year
[10:00] Adam Xinpeng: Anyway!
[10:00] Zha Ewry: So, we'll look for dome doc in the wiki
[10:00] Adam Xinpeng: I have to run for a meeting with another Fortune-50 company, so back later.
[10:00] Zha Ewry: We'll also be utterly glad to help flesh stuff out
[10:01] Zha Ewry: But.. we do need some hints, from time to time
[10:01] Braulio 0.3: All Go
[10:01] FWord Utorid: i will be happy to see the next version of non-functional technology based on the old one.
[10:01] Alanagh Recreant slips out politely.... fascinating stuff :)
[10:01] FWord Utorid: and the wonderful licensing headaches and lies and deceit that go along with it will be a nice show.
[10:02] Zha Ewry: The biggest tangle to unknot, client/sim, is the connections between the various type of communication
[10:02] FWord Utorid: the communication isn't going to be the issue, it will be the lawyers.
[10:02] Zha Ewry: The places where X has to happen, before Y, and the response then is triggered on seperate channels
[10:02] Dr Scofield: protocols
[10:02] Saijanai Kuhn: and how to USE said channels
[10:02] Zha Ewry: Well, in some cases, patterns between the protocols
[10:03] Zha Ewry: Because we've got some odd couplings
[10:03] FWord Utorid: if my prim hair is transferred to ibm's grid i will have to set ibm on fire.
[10:03] Ina Centaur: hmm but you never even talked to ET though
[10:03] FWord Utorid: you do not have a license to transfer my commodities to your multigrid technology
[10:03] FWord Utorid: i forbid it.
[10:03] Lom Hax riffles through Tao's pockets while he his afk.
[10:04] Zha Ewry nods at Fword
[10:04] Ina Centaur: personally, i like the idea that the asset servers are immortal above SL
[10:04] Dr Scofield: fword, ok, you'll get an ugly wig on the multigrid
[10:04] Ina Centaur: but i missed the a chunk of the convo here...
[10:04] Zha Ewry: One of the obvious things that needs to be done right in multi-grid, is protecting the rights of content creators
[10:04] Zero Linden is Online
[10:04] Ina Centaur: would the multigrid support commerce or will we all turn into communists? o.O
[10:04] FWord Utorid: that's the entanglement
[10:05] Teravus Ousley: I've started work also.. on the web-login in OpenSim, however ran into an issue.. I think I've mentioned the VWR issue on Jira though.
[10:05] Dr Scofield: there's not much that can be done -> DRM
[10:05] Day Oh: http://jira.secondlife.com/browse/VWR-4021
[10:05] Zha Ewry: Well, we can respect intent, but not much more
[10:05] FWord Utorid: communications protocols underneath it aren't nearly as bad as the lawyers and the people who don't understand the implications
[10:05] Dr Scofield: teravus, i was looking at that as well, have you checked in stuff to the opensim svn yet?
[10:05] Dahlia Trimble runs out quickly to file a copyright on the ruth avatar
[10:05] Teravus Ousley: I have
[10:05] Zha Ewry: Prior Art, Dahlia
[10:05] Dr Scofield: zha, excactly
[10:05] Teravus Ousley: it's all in the code for the 'user server'
[10:06] Ina Centaur: i think Poser/Metacreations sold off all rights to ruth...
[10:06] FWord Utorid cheers for Dahlia
[10:06] Dr Scofield: ah, ok
[10:06] FWord Utorid: if IBM uses my textures I will sue them and take over.
[10:06] Ina Centaur: ... and with current 3d sculpting tools it wouldn't take that long to make a ruth-twin
[10:06] Ina Centaur: from scratch!
[10:06] Tess Linden: Icehouse will look into VWR-4021
[10:06] Teravus Ousley: the user server displays messages about what the requests contain and what not for easier debugging
[10:06] Zha Ewry: So... The other thing which would be really nice, from Linden, would be some sense of the plans to migrate the client/sim connections towards more UDP
[10:06] Ina Centaur: i dunno. but why would anyone go with ruth nowadays..
[10:07] Zha Ewry: Sorry, away from DUP to caps
[10:07] Zha Ewry: *UDP
[10:07] FWord Utorid: no UDP!
[10:07] Saijanai Kuhn: towards TCP/Caps...
[10:07] Zha Ewry cannot type this afternoon
[10:07] Zha Ewry: From UDP->Caps
[10:07] Teravus Ousley: Thanks, that would be appreciated.
[10:07] Saijanai Kuhn agress with FWord whats up with that
[10:07] Dr Scofield: caps, caps, caps...zha menat caps!
[10:07] Zha Ewry: CAPS
[10:07] Zha Ewry puts on her CAP
[10:07] FWord Utorid: because I am always right and you are slow, Sai.
[10:07] Saijanai Kuhn: and my protocols useless
[10:08] FWord Utorid: I told you that from day one
[10:08] Teravus Ousley is Online
[10:08] Zha Ewry: I assume there is a plan to get some stuff of UDP in the near term, Lindens?
[10:09] Saijanai Kuhn: the REAL question is HOW will this get "off of UDP?"
[10:09] Zha Ewry nods
[10:09] Saijanai Kuhn: the EventQueueGet is a rather odd way of doing it, IMHO
[10:09] Dr Scofield: what and how
[10:09] FWord Utorid: ok these precedings are a sham to make sai work all the time for no money. I will not stand for it.
[10:09] Zha Ewry: Well, towards the evenQueue, one assumes
[10:09] Ina Centaur: lol
[10:09] Saijanai Kuhn: It feels (like my son commented) like an interum solution
[10:09] FWord Utorid sits
[10:10] Lom Hax: saij, is that question premature? Do we know enough about the protocol to start figuring out how to evolve it?
[10:10] Rex Cronon: r u getting paid fword>
[10:10] Dr Scofield: also, for interop with OpenSim we need to know the boundary line, where it runs?
[10:10] Teravus Ousley: well, better documentation on that particular cap and what the near term possibilities and parameters are would be of benefit.
[10:10] Zha Ewry: Between Agent and Region?
[10:10] FWord Utorid: rex, i am smart, i do not work for nothing.
[10:10] Dr Scofield: agent and region and also region to region
[10:11] Dr Scofield: asset server?
[10:11] Zha Ewry nods
[10:11] Lom Hax: If I understand correctly, the boundary between client and server was not terribly clear historically. Cleaning that up is part of hte challenge. Yes?
[10:11] Saijanai Kuhn: all of my volunteer work has been designed to make me more attractive, job-wise, either for LL or for some other company
[10:11] Zha Ewry: Region/Region is a pretty small set of things
[10:11] Tess Linden: I think that migration will begin with login
[10:11] Saijanai Kuhn: I'd do the work anyway, but not necessarly the way I've been doing it
[10:11] Tess Linden: when we begin to split off agent and region domains
[10:11] Tess Linden: Donovan just recently released login.cgi support for llsd
[10:12] Multi Gadget v2.0.3b by Timeless Prototype, '/44 info'
[10:12] Zha Ewry: Well, login is pretty simple
[10:12] Lom Hax: cgi, eh? going Old School!
[10:12] Dr Scofield: just a name
[10:12] FWord Utorid: yes, the technology in use is dated and hacked together.
[10:12] Zha Ewry: But.. even there, what exactly it means isn't obvious
[10:12] Teravus Ousley notes that jhurliman added working LLSD support for login on OpenSim, though it's incomplete.
[10:12] Saijanai Kuhn: though care has to be taken not to log things into a LL-viewer-centric login procedure
[10:13] Day Oh: That's a point, it seems like there can't be much plan until the domain split thingy
[10:13] FWord Utorid: what is needed is to ignore the old way of thinking and embrace a new one
[10:13] Zha Ewry: Login sort of happens in three bits, at the moment
[10:13] Saijanai Kuhn: not all clients want to know about inventory. Some only want group IM and some might only want to check yoru linden account balance
[10:13] Teravus Ousley: it's missing friends, and inventory definitions atm.
[10:13] Zha Ewry: The first tiny bit, which gets you in the door
[10:13] Zha Ewry: The connectiont to the sim, which is in two parts, and several phases
[10:14] Tess Linden: Linden has been doing lots of server side work to put web services in front of old school "messages"
[10:14] Zha Ewry: and.. how that changes, if we have an Agent domain is very unclear
[10:14] FWord Utorid: web services are definately the way to go
[10:14] Tess Linden: as we test them gradually, we can begin to migrate to use wbe services, and then add caps in front for clients
[10:14] Zha Ewry: How much of that is on the sim/serice path?
[10:14] FWord Utorid: none of the informational content should come from the simulators if it's possible to offload the data
[10:14] Zha Ewry: Is there a migration plan sort of laid out Tess? That would also be really useful
[10:15] Saijanai Kuhn: FW0rd none should come from the simulators, but its all server-driven unless its static
[10:15] Zha Ewry: and. Yes, in an ideal world, the simulators, eventually, offload everything that isn't about physical state melding
[10:15] Tess Linden: zha: the migration plan at the moment is mainly server side, but the client side stuff will stem out from login
[10:15] FWord Utorid: sai I am nearly omnipotent, you don't have to tell me how things should work ;)(
[10:15] X-Flight Device - Wear or Attach to Fly ANYWHERE: All Go
[10:16] Saijanai Kuhn: well, then I should know that already, FWOrd
[10:16] Tess Linden: Rdw and Phoenix are working on inventory services
[10:16] Zha Ewry nods
[10:17] FWord Utorid: even the fastest horsedrawn carriage is no match for a v8.
[10:17] Zha Ewry: It might make terrififc sense to have a set of office hours, where each group that's working on a set of function like that, took people through what they are doing, and where they are
[10:17] Zha Ewry: Smaller, more focused than Zero's hours
[10:17] Tao Takashi: I actually wonder how we go on from login or how we get the login working now with a testgrid and probably testclient
[10:17] FWord Utorid: yes, we should corporatize this process so we accomplish less.
[10:17] Saijanai Kuhn: even ifthey're only once or twice a month, would be a great help
[10:18] Zha Ewry: Whichh has done a terrific job with his hours on http-c, and escrow
[10:18] Tao Takashi: because then we would have some sort of a start of a protocol and implementation
[10:18] Tao Takashi: yes, Which is quite great in what he does and how he communicates it
[10:18] Tess Linden: Sai: I believe Which/Rdw has his own office hours
[10:18] Tao Takashi: he has
[10:18] Saijanai Kuhn: Been there several times. Not able to follow a lot of it though
[10:19] Zha Ewry: You can go to the wiki, and find a lot of what he's doing written up as well
[10:19] FWord Utorid: the dream of a metaverse becomes a nightmare when people all start as ruth and their flesh blurs three times a minute.
[10:19] Zha Ewry: What's currently missing is the bridge between that, and the current caps, and the queue style pipes
[10:19] Tao Takashi: the problem with his office hours are maybe that he is quite a lot into escrow and stuff while we try to grok the simple things at the beginning ;-)
[10:19] Rex Cronon: wiki, the great hall of knowledge, where everything is so easy to find
[10:19] Dr Scofield: lol
[10:20] Teravus Ousley: wiki + google :D
[10:20] Saijanai Kuhn: Wiki: Ward's Impossible Knowledge Index
[10:20] Tao Takashi: he also writes good updates on the sldev mailinglist but then again you still need to know what he's talking about ;-)
[10:20] Rex Cronon: is more like babel tower
[10:20] Zha Ewry nods
[10:20] Ina Centaur: / i have a somewhat off-tangent question... is there a significance that these meetings always conflict in time with Dr Dobb's Tuesday Gridtalks?
[10:20] Dr Scofield: tess, is there a timetable for switching to completely web auth based login?
[10:21] Tess Linden: Tao: you can always ask him questions about general things too
[10:21] Zha Ewry: No, there isn't. If we want to move them, we can
[10:21] Tao Takashi: Tess: of course ,that's what I do :)
[10:21] Zha Ewry: I aimed for a slot before Zero's afternoon hours
[10:21] Lom Hax: Ina has a point. I've been struggling with the conflict myself.
[10:21] FWord Utorid: every day for a year has been wednesday offline for maintenance day.
[10:21] Burhop Piccard: me tooo (about conflict)
[10:21] Tess Linden: Dr Scofield: yes, sometime in March, look out for an sldev email soon with schedules
[10:21] Lom Hax: Though mornings are good so I'm not splitting up my work day.
[10:22] Burhop Piccard: I like the time, just not the day (but I'm flexible)
[10:22] Dr Scofield: ok, thx for the info
[10:22] Lom Hax concurs with Burhop.
[10:22] Zha Ewry: The OpenSim community, is looking for some solid progress on the client being OpenSim friendly, before then. We've sort of been wrestling with that on and off for a month now
[10:23] Dr Scofield: yes, it would be extremely good to have VWR-4021 resolved for all platforms
[10:23] Tess Linden: Zha: the main issue facing OpenSim login is VWR-4021, is that correct?
[10:23] Dr Scofield: yes
[10:23] Zha Ewry: I think that's the last one Tess
[10:23] Dahlia Trimble: is that the windlight login issue?
[10:24] Tess Linden: I will mark it as triaged so you know Icehouse will look into it
[10:24] Dr Scofield: the question is how do we get from
[10:24] Dr Scofield: secondlife:///app/login?first_name=SomeDude&last_name=SomeDude&location=home&grid=Other&web_login_key=bce8dasd-2ddd-4278-b06c-ccfd2a04c7d7
[10:24] FWord Utorid: I am amazed that I haven't crashed yet.
[10:24] Dr Scofield: to the client logging in to some non-LL grid
[10:25] Teravus Ousley: Yes that key actually worked for a small period of time on the Bounced Offcourse account on Agni
[10:25] Dr Scofield: perhaps, grid=XXX should be a hostname:port value
[10:25] Rex Cronon: why does my body want to bend over like this?
[10:25] Tess Linden: scofield: we can talk more about that bug after the meeeting
[10:25] Zha Ewry: I think, if 4021 is fixed, the client actually can be pointed to other web pages for re-direct
[10:25] Day Oh: I want in, Scofield & Tess (:
[10:25] Saijanai Kuhn: Theese meetings tend to go on until time For Zeros OH tess
[10:25] Zha Ewry: Heh
[10:25] FWord Utorid: rex, welcome to second life. you are all ruths with muscular control problems.
[10:25] Dr Scofield: i think teravus should be in as well
[10:26] Zha Ewry: Well, I plan for 1 to 1.5 hours.. but they end time, is rather vague, as people keep talking (just like we have peeps hang out for hours after Zero vanishes at his hours)
[10:26] Dr Scofield: let's just do it "afterwards" here
[10:27] Zha Ewry: Joshua jost posted the 1.19.0 roadmap
[10:27] Zha Ewry: (on SL dev)
[10:27] Dr Scofield: or, if we are through with the agenda, do it now?
[10:27] Zha Ewry: Go for it
[10:27] Dr Scofield: tess?
[10:27] Tess Linden: yep!
[10:27] Dr Scofield: teravus? day oh?
[10:27] Day Oh: Hehe
[10:28] Zha Ewry: All yours 4021 people :-)
[10:28] Day Oh: I like Dr Scofield's idea of having the loginuri part of the slurl
[10:28] Saijanai Kuhn is all ears (looking at the Ross Perot pretuberances)
[10:28] Day Oh: Can't think of another way to be able to launch the whole thing from a web page
[10:28] Dr Scofield: is that a good idea, tess?
[10:28] Tess Linden: hm, i see, because theres no way to set the login uri
[10:28] Dr Scofield: yep
[10:28] Dr Scofield: currently everything seems to be LL geared
[10:29] Saijanai Kuhn: would make proxies easier (at least for some :-()
[10:29] Teravus Ousley: well, one benefit of having it in the SLURL is, in the future, it can change and clients won't have to be updated.
[10:30] Dr Scofield: the goal would be that we can point the client to a non-LL login page, and have that return a secondlife://... SLUR pointing to our OpenSim grid
[10:30] Jessica Qin is Online
[10:30] Tess Linden: are there any security concerns about being able to arbitrarily set the server to connect to in URL that you click?
[10:30] Saijanai Kuhn: as long as you can clearly read which URL it is pointing to at some point...
[10:30] Neas Bade is Online
[10:31] Tess Linden: but you cant do that on the client, it doesnt display the login page URL
[10:31] Dr Scofield: hmm...if you include the avatar name, yes
[10:31] Dr Scofield: i think...
[10:31] Dr Scofield: you could then do a man-in-the-middle attack
[10:31] Dr Scofield: couldn't you?
[10:31] Teravus Ousley: well, at this moment, you'd need to change the loginpage using the shortcut
[10:31] Dr Scofield: true
[10:32] Dr Scofield: how about leaving out the first/last name and just supply the web_login_key?
[10:32] Teravus Ousley: .. I agree however, in the future, if we do choose to go the 'web page only' route, there may be some important notice or difference noted to the client.
[10:32] Day Oh: True or false: If you're generating the url that points to the particular grid, you must've also generated the web_login_key?
[10:32] Dr Scofield: the login server could have the association (avatar name, web_login_key) stored
[10:32] Teravus Ousley: .. such as the menu bar turning 'red', like it does when you use the -user parameter
[10:33] Zha Ewry: Hey Neas
[10:33] Saijanai Kuhn: keep in mind that not all clients will be the official LL client
[10:33] Teravus Ousley: major companies such as ebay, and paypal use specific things on the interface to let the user know that the site isn't owned by LL
[10:33] Saijanai Kuhn: should be a pattern for security, rather than a specific "thing" like menu color or whatever
[10:33] Dr Scofield: then on agent/region login the grid could use the web_login_key to retrieve avatar name, etc
[10:33] Dr Scofield: how does that sound?
[10:33] Teravus Ousley: (well, I'm relating it back to LL)
[10:34] FWord Utorid: I am keeping the hope that there will be someone with the resources and desire to do it *right* and that i will no longer have to endure the pain that is caused by this platform and it's flawed architecture.
[10:34] Tess Linden: hm, so if -loginpage option was given, the client would give some visual notification of the change
[10:34] Dr Scofield: that would work
[10:34] Zha Ewry: I actually quite like the notion that the mainlien client should make it clear when you're going someplace else
[10:35] Zha Ewry: It can be hard to be sure you've spelled loginuri right today ;-(
[10:35] Tess Linden: but it doesnt stop anyone from making a link pointing to "secondlife:......" that automatically logs you in
[10:35] Tao Takashi: I was thinking about maybe doing some proxy for my initial AWG login server so I might be able to somehow use that.. but then I started to wonder how this might work with the new login without actually changign the client
[10:35] Teravus Ousley: alternatively, using the web-only style, if the Login URI wasn't in the linden grid array as Bounced put it
[10:35] Dr Scofield: except that you'd need a short lived web_login_key, wouldn't you?
[10:35] Day Oh: I'm no expert, but it seems like it should be safe, someone can't make a url to log you in without a web_login_key
[10:35] FWord Utorid: the problem with secondlife:// urls is that they run a new viewer which dies and transfers to the existing viewer
[10:36] Tess Linden: day: true. by then you would have already authenticated, and the steps after that would be to have you log into the grid
[10:36] Teravus Ousley: in theory, the web_login_key could be re-generated often.. like a security token
[10:36] Tao Takashi: so what happens when the secondlife://app link is triggered?
[10:36] Tao Takashi: the client connects to the XML-RPC service with webkey etc.?
[10:36] Day Oh: That's the only part we have trouble with (:
[10:36] Tess Linden: teravus: it is re-generated
[10:37] Dr Scofield: teravus, i thought that's what happened?
[10:37] Tess Linden: tao: yes
[10:37] Dr Scofield: see :-)
[10:37] Zha Ewry: Right. You want it to be very short term
[10:37] Zha Ewry: And. . Even so, with alts, and multiple clients up
[10:37] Tao Takashi: and what about simply giving that XML-RPC server address on the command line?
[10:37] Zha Ewry: Its goign to be entertaining
[10:37] Dr Scofield: how about secondlife://grid:port/...
[10:37] FWord Utorid: the notion that needs to be considered is called a REFACTOR
[10:37] Tao Takashi: and //app/ will then use that?
[10:37] Tess Linden: we can trust the authentication server which provides the web login key to also provide the login url for the client to connect to?
[10:38] Jessica Qin is Offline
[10:38] Day Oh: If it doesn't, the key won't work
[10:38] Tess Linden: tao: the method for which that information is passed on is dependent on the operating system
[10:38] Zha Ewry: I think the OpenSim community would be OK doing that step if they can point to a web page they control
[10:38] Zha Ewry: Right. I can linke secondlife:// to anything I want
[10:38] Tao Takashi: ok, but at some point the client is triggered to initiate the XML-RPc request
[10:38] Tess Linden: tao: its only configurable based on the secondlife://... part of it
[10:39] Tao Takashi: so the client cannot override it?
[10:39] FWord Utorid: oh that gives me some good ideas for sploits
[10:39] Tao Takashi needs to read client sourcecode again it seems :)
[10:39] Tess Linden: tao: the operating system adds a -url <info> where <info> is exactly the secondlife:// url
[10:39] Dr Scofield: secondlife://gridX.secondlife.com:XYZ/app/login/....
[10:39] FWord Utorid: bah
[10:40] Neas Bade: right, one of the key issues today is that secondlife:/// has an implicit hostname in how it is handled
[10:40] Dr Scofield: is the -url option implemented on all platforms?
[10:40] Tao Takashi: and this can't be overridden?
[10:40] Tess Linden: scofield: /app/login is the correct path for the service that the client is handling
[10:40] Neas Bade: which should be purged
[10:40] Tess Linden: scofield: we can add additional query arguments to it
[10:40] FWord Utorid: I will not submit to a new form of internet solely controlled by one company
[10:40] Neas Bade: and require an explicit hostname in all urls
[10:40] FWord Utorid: metaverse://ip:port/login:pass/sim:x:y:z
[10:40] Dr Scofield: tess, secondlife://grid:port/app/login is not acceptable?
[10:40] Day Oh: If someone wants to generate a url to direct you to another server, they have to tack the web_login_key onto it too... if they have the web_login_key already, there's no need to send you to another server anyway... they won't have the web_login_key (:
[10:40] Tao Takashi: so is the problem that the thing handling secondlife:// is a new instance and might not know about cmdline parameters?
[10:41] FWord Utorid: the name secondlife should not be applied to everything
[10:41] Lom Hax nods
[10:41] Saijanai Kuhn: its a good point, long-term, FWord
[10:41] Tess Linden: scofield: it should remain secondlife://app/login?....
[10:41] Lom Hax: Yes, virtual worlds are bigger than SL.
[10:41] FWord Utorid: metaverse://
[10:41] Saijanai Kuhn: but the pattern is what is important right now, not the specific term
[10:41] Dr Scofield: tess, currently you have secondlife:///...
[10:41] Lom Hax: si
[10:41] FWord Utorid: sai, i will decide what is important.
[10:41] Dr Scofield: that's without the host part
[10:41] FWord Utorid: you are not important, sai.
[10:42] Saijanai Kuhn: OK, for ME, the pattern is important
[10:42] FWord Utorid: ok better ;)
[10:42] Dr Scofield: i was wondering whether it wouldn't make sense to actually use the host part there for the grid
[10:42] Neas Bade: the lack of a host in the url is part of what is causing the mess right now, as it means there is implicit hostnames built into the client to handle that fact
[10:42] Tess Linden: hm you're right scofield
[10:42] FWord Utorid: actually the traditional format for uri based logins is
[10:43] FWord Utorid: proto://userpass...serverport/subinfo
[10:43] Tess Linden: scofield: yes that does make sense for the place to put it
[10:43] FWord Utorid: one might observe standards
[10:43] Dr Scofield: great! :-)
[10:43] FWord Utorid: you know, things we've overlooked everywhere here
[10:43] Dr Scofield: teravus, neas, day oh: what do you thing?
[10:44] Teravus Ousley: I'd be excited if that were the case
[10:44] FWord Utorid: the truth is though
[10:44] FWord Utorid: url based logins are ridiculous
[10:44] Teravus Ousley: it would open up easy management and load balancing opportunities
[10:44] Neas Bade: yep, secondlife://osgrid.org:8002/app/login would be great
[10:44] FWord Utorid: it creates a seriously annoying loop
[10:44] FWord Utorid: click an slurl to open a browser
[10:44] Tess Linden: scofield: how does it convert the host/port into a login uri?
[10:44] FWord Utorid: the browser shows a map
[10:44] FWord Utorid: then launch another instance of sl?
[10:45] FWord Utorid: it's unusabe.
[10:45] Day Oh: I like the idea
[10:45] FWord Utorid: *able.
[10:45] Neas Bade: this actually gets into a more interesting point of the fact that there are hostnames hardcoded in the xml files for the client in a bunch of places
[10:45] Neas Bade: I think I found 50ish occurances
[10:45] Teravus Ousley: presumably the address in the SLURL would be the address to the login.cgi application
[10:45] FWord Utorid: also, peer to peer is the way to go
[10:45] FWord Utorid: with directory services
[10:45] Neas Bade: it would be good if things like search used the same base url as provided to loginpage
[10:46] Saijanai Kuhn: there used to be host names encoded into the client source.
[10:46] Dr Scofield: if we have secondlife://opensim.ibm.com:9000/app/login then it would go to http://opensim.ibm.com:9000/app/login/
[10:46] FWord Utorid: requiring osgrid or any other intermediary is lame
[10:46] Neas Bade: Saijanai: there may still be, I was just doing a quick grep of the xml
[10:47] Tess Linden: scofield: the secondlife uri would be cgi-bin/login.cgi
[10:47] Day Oh: You were saying secondlife://app/login?host=opensim.irb.com:9000&etc=etc, right?
[10:47] Neas Bade: Day Oh: huh?
[10:47] Dr Scofield: day oh: no
[10:47] Day Oh: Oh (:
[10:47] Zha Ewry: it's fine to hold some of those inside the XML, as long as they don't have to be used
[10:48] Neas Bade: secondlife://host:port/path
[10:48] Tess Linden: the only thing is that host port part of the secondlife url is actually referring to the hostport of the *client* itself isnt it though?
[10:48] Neas Bade: Zha: I disagree, they should be returned as part of the login packet or relative to the loginpage
[10:48] Tess Linden: its not really referring to the host port of *login*
[10:49] Dr Scofield: hmm...is it? i thought it would point to the grid server
[10:49] Neas Bade: if they are there in the xml as hostnames, the use of alternative urls will get forgotten, much like the loginpage issues now
[10:49] Dr Scofield: login
[10:49] Zha Ewry: Thats a good point, Neas
[10:50] Dr Scofield: basically, the client would just substitute http for secondlife in the SLURL to get the login call
[10:50] FWord Utorid: this message is not on topic
[10:50] Dr Scofield: login URL
[10:50] Tess Linden: scofield: the secondlife scheme means that the client will respond to the URL
[10:51] Tess Linden: app/login means that that is how the client will respond
[10:51] Dr Scofield: yes, the client gets the SLURL
[10:51] Tess Linden: the query arguments are configuring how the client will login
[10:51] Jason Ayakashi is Online
[10:51] Dr Scofield: hmm
[10:51] Neas Bade: ok, but why are you thinking of the slurl as anything other than a network resource like an http url?
[10:51] Day Oh: I see, y'all got a weird idea (:
[10:52] Tess Linden: the missing host:port part is actually supposed to refer to *which* client, so its kinda weird
[10:52] Neas Bade: I understand that is the way it is implemented today, but I think it's a bit of a funky way of doing urls
[10:52] Tess Linden: its the way to get the client to launch from a web browser
[10:52] Dr Scofield: i was thinking that the client would just take the SLURL to login in with
[10:53] Dr Scofield: the secondlife: part is,
[10:53] Neas Bade: yeh, I'm with scofield on this one
[10:53] Teravus Ousley: right, following that logic, you might have something like secondlife://app/login?first_name=somedude&last_name=SomeDude&LoginURI=http://someServer:someport/login.somescript
[10:53] Neas Bade: ug
[10:53] Teravus Ousley: &web_login_key=xxxxxxx-xxxx-xxxx-xxxxxxxxxx
[10:53] Tess Linden: scofield: what if the override the grid argument?
[10:53] Day Oh: perhaps without the http://
[10:54] Dr Scofield: the rest i thought would just point to the login server
[10:54] Tess Linden: like instead of grid=Agni, its grid=agni.lindenlab.com
[10:54] Zero Linden is Offline
[10:54] Dr Scofield: fine with me
[10:54] Day Oh: Anything similar to that sounds good to me
[10:54] Zha Ewry: As much as possible it would be good if we can keep the grids pretty well divorced from machine names and yes/.
[10:54] Dahlia Trimble: how about a case where there may be a stand alone sim that manages it's own logins without a separate server?
[10:55] Jessica Qin is Online
[10:55] Teravus Ousley: well, also be aware that things like port and resource also matter
[10:55] Zha Ewry: A stand along sime can host a single page to do the redirect, can't it?
[10:55] Dr Scofield: didn't realize that you have the client interpret the SLURL
[10:55] Teravus Ousley: agni.lindenlab.com:9025/resource
[10:55] Tess Linden: we hate having to add additional grids to some client list for hardcoding
[10:55] Zha Ewry nods
[10:55] Saijanai Kuhn: The final form of that would be a client/server integrated pakcage
[10:55] Dr Scofield: teravus, and we'd URL escape that
[10:55] Zha Ewry: No, don't want that. Might want a way to register onto a service, tho.
[10:56] Tess Linden: ok, i'll check with the rest of the team to see if anyone has any other objections
[10:56] Dr Scofield: so, the proposal is
[10:56] Zha Ewry: Tess, btw, this has been insanely furitful
[10:56] Neas Bade: yeh, I think the issue is that currently secondlife:// is a hack to get a web browser to launch the SL client, and isn't being thought of as an actual network resource being loaded
[10:56] Rex Cronon: i think andrew linden office hours should start in a few minutes, so i am going there, bye everybody
[10:56] FWord Utorid: yes! we accomplished much! we rearranged the letters in secondlife://this.that.is.important!
[10:56] FWord Utorid: thanks for using my hour!
[10:56] Dr Scofield: secondlife:///app/login?...grid=opensim.bla.com:9000/resource...
[10:56] FWord Utorid: you'll be billed!
[10:56] Neas Bade: it should be a real network resource url that mime configuration ensures it hands off to the right applications
[10:57] Zha Ewry nods at Neas
[10:57] Dr Scofield: with the opensim.bla.com:9000/resource bit URL escaped
[10:57] FWord Utorid: WE ARE IMPORTANT
[10:57] Neas Bade: no no, you don't want to put in the grid hostname as a parameter
[10:57] Saijanai Kuhn: one question: would this pattern work with a situation where the client and server were teh same app living on a single computer?
[10:57] Neas Bade: it defeates the purpose of having a uri in the first place
[10:57] Tess Linden: scofield: can you send an email to sldev describing what we discussed here?
[10:58] Dahlia Trimble: so if you are logged in to one grid, and you click this slurl and it redirects you to outside that grid, would you go with the same client and just log in, or would it open a new client?
[10:58] Dr Scofield: yes, i can
[10:58] Tess Linden: sai: the uri is the location of the client though
[10:58] Tess Linden: its not the location of the login server
[10:58] FWord Utorid: yes
[10:58] Tess Linden: dahlia: it is ignored if you're already logged in
[10:59] Tess Linden: thanks scofield
[10:59] Neas Bade: tess, that's very backwards of the way that it should be
[10:59] Dr Scofield: neas, the secondlife:/// URI is intended for the client, as tess explained
[10:59] Saijanai Kuhn: ok. So it really shouldn't matter. Was just checking, because its an issue for some possible scenarios
[10:59] FWord Utorid: ok. the sl blog says that they are outlawing chat because the government doesn't want people to talk!
[10:59] Neas Bade: Dr Scofield: then it isn't a uri, so shouldn't look like one
[11:00] Dr Scofield: but you need to make it look like one so that the browser will hand it over to the client
[11:00] Tao Takashi: I am off for now.. cya at Zero's!
[11:00] Neas Bade: secondilfe://osgrid.org/region/wright%20plaza/20/40/20
[11:00] FWord Utorid: i will sit here and wait for IBM to atone for it's crimes
[11:01] Tess Linden: ok ill see you guys later
[11:01] Teravus Ousley: ok, tc
[11:01] Dr Scofield: LL is using the SLURL to also tell the client how to login...i imagine the idea is to have different schemes later
[11:01] Day Oh: Thanks for everything Tess
[11:01] Neas Bade: I think that creating uri like things because of ways to trick current browsers without actually making them real uris just creates a mess
[11:01] Zha Ewry: gee
[11:01] Dr Scofield: ok, tess! thanks for your time
[11:01] Zha Ewry: Someone seems to have teleport Fword home
[11:01] Saijanai Kuhn: thanks for dropping by Tess, will need to pick your brain (or someone's) later about various CAP/EventQueueGet/etc issues
[11:01] Day Oh: Can't wait for whatever that documentation you mentioned was XD
[11:01] Zha Ewry: tess, this was super
[11:02] Day Oh: >_>
[11:02] Day Oh: She poofed quick
[11:02] Zha Ewry: And. Hello Donovan, when did you drop in?
[11:02] Saijanai Kuhn: Hey Donovan
[11:02] Dr Scofield: neas, ok...but it seems like tess and her team like the /app/login idea, so having grid=.... is the next best thing we can get
[11:03] Dahlia Trimble: thanks for the info... bye :)
[11:03] Neas Bade: it's a crappy design point
[11:03] Neas Bade: and it wouldn't be hard to make it the correct design point by having real uris
[11:03] Saijanai Kuhn: it may fit in with all the work they've already done though
[11:03] Zha Ewry: It would be nice if they believe in the web
[11:03] Neas Bade: the more this differs from the way the www works, the worse it is
[11:03] Saijanai Kuhn: so they're selling you on their work, rather than on a design
[11:04] Neas Bade: and the harder it will be to get it to work with other things
[11:04] Donovan Linden: hey, I got here about 20 minutes ago, just lurking
[11:04] Saijanai Kuhn: 3 LInden in one day? Whats up with that?
[11:05] Dr Scofield: ok. for the time being we seem to have a solution that will help us resolve the problem we currently have
[11:05] Lom Hax: Must be our new collogne
[11:05] Dr Scofield: what i'll do is summarize and post to sldev
[11:05] Dr Scofield: we can then think of improvements and suggest those to tess
[11:05] Saijanai Kuhn: I think, in the long run, we'll need to develop a way of accomdiating the current LL design and come up with our own at the same time
[11:05] Dr Scofield: we need to be able to communicate to the client which grid to use
[11:06] Neas Bade: dr scofield, yes we do
[11:06] Zha Ewry: Sigh
[11:06] Day Oh: It prolly needs more discussion in the future, like there's a slurl for viewing group info, secondlife:///app/group/groupUUID/about but it doesn't point to a particular grid
[11:06] Saijanai Kuhn: that way, whichever design Linden Lab is using, it doesn't break OPenSim and vice versa
[11:06] Dr Scofield: LL apparently wants to also communicate to the client how to login in
[11:06] Zha Ewry: which they really shouldn't.
[11:06] Dr Scofield: agree
[11:06] Neas Bade: but the problem is that if it is done as such an after thought, it would break in a future client because no one really groked how fundamentally these need to be network uris
[11:07] Neas Bade: which means we'll be back here doing this all again in 3 months with the 1.19 client
[11:07] Zha Ewry: Well, hopefully, with the 1.20
[11:07] Dr Scofield: no, we can do the discussion on sldev
[11:07] Neas Bade: because it was forgotten that other grids needed to be accomodated
[11:07] Zha Ewry: Lets reply to the outline on SLdev, and see if we can get it fixed before it gets coded
[11:07] Saijanai Kuhn: Zero's expressed a need for OpenSim and SL grids to talk to each other ASAP. Should raise the point with him also
[11:07] Zha Ewry: Its still helpeful that they actually paid attention
[11:08] Dr Scofield: right...what kind of tag would i use? [VIEWER]?
[11:08] Dr Scofield: and we at least got to a solution that would work for us, now we can try and improve on that
[11:09] Neas Bade: we still won't be able to localize search, or any of the other network facing functions of the client
[11:09] Neas Bade: because hostname is an after thought
[11:09] Zha Ewry: Oh. We *NEED* to raise those
[11:09] Neas Bade: that's why I think this approach isn't workable
[11:09] Dr Scofield: that needs to be part of the seed caps we get as part of login, i'd think
[11:09] Dr Scofield: good point
[11:09] Zha Ewry: Those ought to come back from the sed-cap, yes
[11:10] Zha Ewry: Not in the client config
[11:10] Saijanai Kuhn: I'd rather the seed cap be ALL that is returend during login
[11:10] Zha Ewry: Well, one step in
[11:10] Zha Ewry: "get_search_url"
[11:10] Saijanai Kuhn: right, returned from the seed-cap itself
[11:10] Zha Ewry: is a perfectly goo thing to get
[11:10] Neas Bade: because if we don't fundamentally get people thinking of these things as real network uris, then we loose the mind set to make host name something that is a relative construct instead of hard coded
[11:10] Saijanai Kuhn: the current login process returns about 1 megabyte of stuff.
[11:10] Saijanai Kuhn: I'd rather hold the line at one thing and let THAT kitchen sink if we can't get anything more structured
[11:11] Dr Scofield: ok, i'm off...need to get dinner and post that summary :-)
[11:11] Lom Hax: OK, folks. I need to go do some work for a client. See y'all later.
[11:11] Dr Scofield: good night all!
[11:11] Saijanai Kuhn: later all
[11:11] Zha Ewry: Dr.:
[11:11] Day Oh: Goodnight
[11:11] Zha Ewry: Please post the comment Neas made about search and such
[11:11] Zha Ewry: So that it gets into people's thinking
[11:11] Dr Scofield: ok, will do!
[11:11] Neas Bade: thanks dr scofield
[11:12] Dr Scofield: np :-)
[11:12] Dr Scofield: cu
[11:12] Zha Ewry: Wow.
[11:12] Donovan Linden: I'll participate more in the next meeting... I'm still gridmonkey today :)
[11:12] Donovan Linden: bye
[11:12] Zha Ewry: Almost productive
[11:12] Dr Scofield: :-)