User:Which Linden/Office Hours/2008 Nov 20

• [11:04] Which Linden: morning sai
• [11:04] Which Linden: I'm going to have to depart at 11:30 PST
• [11:04] Saijanai Kuhn: hey semi here anyway
• [11:05] Which Linden: HEh, k
• [11:05] Saijanai Kuhn: one thing you might find amusing. I think I can sorta get an agent domain working in LSL using the new http-in calls
• [11:05] Which Linden: Hah!
• [11:05] Which Linden: That is hilarious
• [11:06] Which Linden: Makes sense though
• [11:06] Saijanai Kuhn: is easier to experiment with them than with the python equivalents
• [11:06] Which Linden: Really? I'd have thought that the abysmal data structures and text parsing support in lsl would make it a poorer choice
• [11:07] Saijanai Kuhn: well, for a REAL AD, yeah, but just to experiment with sending, receiving messages of the right structure...
• [11:07] Which Linden: I guess you can just do simple text interpolation rather than full llsd parsing
• [11:08] Saijanai Kuhn: right. For now there's only a tiny handful of possible commands to send/receive anyway
• [11:08] Which Linden: Yeah
• [11:08] Which Linden: Simple is beautiful
• [11:08] Which Linden: I do really appreciate the effort given to keep the surface area low
• [11:08] Saijanai Kuhn: yep
• [11:09] Saijanai Kuhn: Hoping that when Infinity gets back, we can work on service discoveryin and the gorup IM thing a bit more
• [11:09] Saijanai Kuhn: prolly have to go to python at that point ;-)
• [11:09] Which Linden: If you're having difficulty writing http handlers in python, perhaps you're using the wrong http framework?
• [11:10] Saijanai Kuhn: but, my hope is that we can start experimenting with peer 2 peer between ADs and between clients
• [11:10] Saijanai Kuhn: ah, well, that's the thing, not using ANY framework, just raw http classes. And LSL leverages all of SL as its "framework"
• [11:11] Saijanai Kuhn: so its just set up the HTTPRequest handler and wait for a get/post
• [11:12] Jagen Sideshow: lovely evening
• [11:12] Saijanai Kuhn: with python, I'd have to actually set up a lot more stuff
• [11:13] Saijanai Kuhn: including opening ports on my comp and whatnot.
• [11:13] Which Linden: hm, sure, go with what works
• [11:14] Saijanai Kuhn: the only drawbacks are 1) can't dierctly access LL sims
• [11:14] Saijanai Kuhn: and forgot 2
• [11:14] Which Linden: "... and I'm all out of bubblegum"
• [11:15] Saijanai Kuhn: ah, and regular GPL client may not be able to handle plain-texdt responses. Tested it with pyogp and loggedin to a LL sim by having the prim script return the generic seedcap URL for an LL sim
• [11:16] Saijanai Kuhn: so the next step is to try to access a OPneSim directly instead of by cut and paste
• [11:16] Saijanai Kuhn: and then I'll try with the GPL client and if it doesn't work, go back and set up a "real"AD with python
• [11:17] Saijanai Kuhn: but its still funny
• [11:17] Which Linden: Oh yeah the content-type restrictions
• [11:17] Saijanai Kuhn: yeah xml-llsd vs plain-text;utf-8
• [11:18] Saijanai Kuhn: but its no biggie was a proof of concept thing anyway
• [11:18] Saijanai Kuhn: though it would be fun to resurrect Day's LSL client login code and combine that with an AD
• [11:19] Saijanai Kuhn: can't maintain sim presence without UDP, but still fun cncept
• [11:19] Saijanai Kuhn: add a faux simulator and you'd have a metaverse on a prim
• [11:19] Saijanai Kuhn: client, AD, sim
• [11:20] Which Linden: The recursion would kill a horse
• [11:20] Saijanai Kuhn: lsl emulator in lsl...
• [11:21] Saijanai Kuhn: but anyway, it is just the springboard to the REAL stuff, which is P2P connections.
• [11:21] Saijanai Kuhn: lots of croquet-like stuff lurking if you can establish direct connections between clients
• [11:22] Which Linden: Yeah definitely
• [11:22] Saijanai Kuhn: and when we get the service discovery and outgoing CAP protocol defined, we can start experimenting
• [11:23] Saijanai Kuhn: got ideas for direct musician collaboration for example
• [11:23] Which Linden: Like what?
• [11:23] Saijanai Kuhn: let the clients do the mixing and then have one of them send the final audio to the server
• [11:24] Saijanai Kuhn: could share audio and MIDI P2P
• [11:24] Which Linden: Hm, how's that different than the server doing the mixing though?
• [11:24] Which Linden: Better control of the timing?
• [11:24] Saijanai Kuhn: server doesn't handle MIDI and less latency
• [11:25] Saijanai Kuhn: I'm pretty sure there are already collaborative software out there for MIDI over hte internet, so this would just use somethign along those lines and then merge it for SL
• [11:25] Saijanai Kuhn: likewise, better whiteboarding
• [11:26] Saijanai Kuhn: and mechanima, etc
• [11:26] Saijanai Kuhn: sky's the limit
• [11:27] Saijanai Kuhn: anyway, time is up for both of us I think. Son has doctor's appt and you had to leave
• [11:27] Saijanai Kuhn: so good chatting. Catch you later
• [11:28] Which Linden: Yeah -- sorry this was so short
• [11:28] Which Linden: Nice to hear about your progress!
• [11:28] Which Linden: Always a pleasure
• [11:28] Saijanai Kuhn: worked out for me I have to leave now anyway
• [11:28] Saijanai Kuhn: likewise I'm sure
• [11:28] Saijanai Kuhn:  ;-)
• [11:28] Which Linden:  :-)