Simulator User Group/Transcripts/2011.04.01

From Second Life Wiki
Jump to: navigation, search


Prev 2011.03.29 Next 2011.04.05

List of Speakers

Andrew Linden Arawn Spitteler Ardy Lay
Ashiri Sands Draconis Neurocam Hermione Ranger
Jonathan Yap Joshua Linden Moy Loon
Opensource Obscure Pauline Darkfury PLDude1 Tester
Psi Merlin Qie Niangao Sebastean Steamweaver
Siana Gearz Sigma Avro Simon Linden
Stickman TankMaster Finesmith Techwolf Lupindo


[16:03] Siana Gearz: hello kelly

[16:03] Draconis Neurocam: hello andrew

[16:03] Andrew Linden: hi there

[16:03] Siana Gearz: hello andrew.

[16:03] TankMaster Finesmith: oh, that reminds me tech, LL is pawing over a patch for more things to gzip before sending...

[16:03] Simon Linden: So the goals of this test system are to be able to re-create a crowded region on our development systems so we can measure what's slow and test how our code changes

[16:03] Siana Gearz: hello there obscure.

[16:04] Simon Linden: One of the problems of doing that is getting a large number of accounts logged in, (mis)behaving like regular people

[16:04] Techwolf Lupindo: The messaging system can be gzip before going over the wire. If the messages packets can be keep under 1500 bytes, things improve a lot.

[16:04] Simon Linden: So we've been working on a system to drive avatars using python scripts

[16:05] Simon Linden: I'm logging one in right now

[16:05] Techwolf Lupindo: Simon, just advertize free lindens, they will come. lol

[16:05] Simon Linden: Right, but they won't do the exact same thing every time

[16:05] Siana Gearz: ahahah free lindens on beta grid, very useful :D

[16:05] PLDude1 Tester: Hey, I'm ready

[16:05] Techwolf Lupindo: Sorry. Its april first, can't help the jokes.

[16:05] Simon Linden: So this Tester account behind Andrew is a bot

[16:05] Sigma Avro: is this on beta grid Simon, sorry I missed begining ?

[16:05] Simon Linden: when I tell her to start, she will cycle through all the gestures

[16:06] Simon Linden: Start

[16:06] PLDude1 Tester: These are all the active gestures

[16:06] PLDude1 Tester: Gesture Female - Chuckle

[16:06] PLDude1 Tester: Gesture Female - Embarrassed

[16:06] Pauline Darkfury: Will you try to include some real 20k+ traffic regions, perhaps, Simon? I can see the value in having test regions that are strictly lab conditions, but by also looking at numbers from real 20k and upwards locations, you might get some useful data

[16:06] PLDude1 Tester: Gesture Female - Blow kiss

[16:06] PLDude1 Tester: Gesture Female - Looking good

[16:06] PLDude1 Tester: Gesture Female - Hey

[16:06] PLDude1 Tester: Gesture Female - Over here

[16:06] Moy Loon: Hey!

[16:06] PLDude1 Tester: Gesture Female - Laugh

[16:06] PLDude1 Tester: Gesture Female - Boo

[16:06] PLDude1 Tester: Gesture Female - Stick tougue out

[16:06] PLDude1 Tester: Gesture Female - Excuse me

[16:06] PLDude1 Tester: Gesture Female - Shrug

[16:06] PLDude1 Tester: Gesture Female - Wow

[16:06] Simon Linden: this is just a test bed, not something we'll use directly

[16:06] PLDude1 Tester: Gesture Female - Repulsed

[16:06] PLDude1 Tester: Gesture Female - Hey baby

[16:06] Siana Gearz: OMG how rude of you!!!

[16:06] PLDude1 Tester: Gesture Female - Cry

[16:06] Sigma Avro: please stop

[16:06] Simon Linden: Don't hurt her feelings

[16:06] PLDude1 Tester: Gesture Female - Bored

[16:06] Siana Gearz: you made her fart!

[16:06] PLDude1 Tester: Gesture Female - Please

[16:06] Andrew Linden: lol

[16:06] PLDude1 Tester: Gesture Female - Get lost

[16:06] PLDude1 Tester: That's it

[16:07] Simon Linden: I've been working on adding other behaviors -moving, sitting, chatting, etc

[16:07] Simon Linden: Quit

[16:07] Simon Linden: and quitting

[16:07] Techwolf Lupindo: Need to add some of those freebies gestures that gesterbaters use.

[16:07] Arawn Spitteler: Unscented Fahrts are socially excusable

[16:07] Techwolf Lupindo: lol

[16:07] Simon Linden: Right - that's obviously a default AV, and a very lightly loaded account

[16:07] Simon Linden: We'll be building more realistic accounts for our test

[16:07] Siana Gearz: the flashy 10 000 furry gesture box :D

[16:08] Arawn Spitteler rehearses a silent but deadly, for his next ride on an SL Elevator

[16:08] Sigma Avro: it works good, I have this stuf

[16:08] Simon Linden: We also have taken snapshots of all the messages going in and out of some busy regions, and will compare our tests and tweak them until we get a similar profile

[16:08] Techwolf Lupindo: Simon, jokes aside, that is a good thing. Being able to measure what residents been bitching about for years.

[16:08] Simon Linden: Once that's done, we'll be in a much better position to test and work on performance

[16:09] Sigma Avro: what performances ?

[16:09] TankMaster Finesmith: you need to load a sim with 50 techwolf avatars... thatll give good performance hit ;)

[16:09] Siana Gearz: byte performances over wire!

[16:09] Simon Linden: We're also hoping this can be use to automate testing ... for example, I have another one that walks in a square path, going around between 4 regions. That will stress crossings

[16:09] Siana Gearz: teaching them wiredance you know.

[16:09] Sigma Avro: lol

[16:09] Techwolf Lupindo: hahahah, let me paste something...

[16:09] Arawn Spitteler: Ever try a four corner crossing?

[16:10] Techwolf Lupindo: Scripts Counted: 0 [0K]

[16:10] Simon Linden: Exactly at the corner?

[16:10] Sigma Avro: spin prim at corner, 6 pi radians per second, haha

[16:10] Techwolf Lupindo: Scripts Counted: 7 [208K]

[16:10] Arawn Spitteler: Actually, I've only gone near the corner, but it really blows up a good vehicle

[16:10] Pauline Darkfury: yeah, the double crossings at corners can be a real killer

[16:10] Sigma Avro: (with avat sited on it)

[16:11] Techwolf Lupindo: Simon, that automated walk I wanted couple years ago and managed to make a test script to do that. Very primative.

[16:11] Pauline Darkfury: There's a few roads and sections of the SLRR have corner crossings, so a fairly real issue in places

[16:11] Draconis Neurocam: yeah my flight assist has pushed me into regions ive not seen any terrain on, i have to fly back into the other region before the emptiness loads

[16:11] Siana Gearz: that's good. primates walk just like silliest of humans.

[16:12] Techwolf Lupindo: There is no llwalkto, so I had to work up something around that.

[16:12] Simon Linden: Those 4 corner crossings also need to be built better ... it's best to put some big obstacle right at the intersection to keep people away

[16:12] Siana Gearz: the smarter the human is, the less it walks.

[16:12] Simon Linden: Anyway, that gives you an idea of some of the pieces coming together for testing

[16:12] Sigma Avro: but in anycase, it does no fly

[16:13] Arawn Spitteler: Oslar (2,256,33) Fun place for a party wagon

[16:13] Andrew Linden: That's going to come in handy Simon.

[16:13] Simon Linden: We hope to also use it to test the new group chat system

[16:14] Andrew Linden: When we have the bots configured for shoot-em-up rollplay we can mock up some of the combat regions.

[16:15] Ardy Lay: Sounds like you are setting up for machinima

[16:15] Pauline Darkfury: A suggestion for a different part of load testing, buy a copy of Novatech HORIZONS, it's fully script-controllable, so could do some very realistic and repeatable tests for large scale rezzing

[16:15] Andrew Linden: Sometime after the new chat system goes online maybe we could make a bot wrapper for these User Groups

[16:15] Andrew Linden: I could send my bot here and then just type into an IRC channel.

[16:15] Sigma Avro: how long will be the testing on chat system ?

[16:16] Hermione Ranger: what's up

[16:16] Andrew Linden: I haven't heard any news about the chat project.

[16:16] Arawn Spitteler: Open seats in the southeast, next to Andrew.

[16:16] TankMaster Finesmith: its on the beta grid

[16:16] Simon Linden: I'm not sure of that, Sigma. We're actually just starting ... the system can't do group chat yet, only local chat

[16:16] Simon Linden: .. our test sytem, that is

[16:16] Andrew Linden: Is it really? Cool.

[16:17] Draconis Neurocam: yeah theres an xmpp viewer and everything

[16:17] Sigma Avro: yes, beta grid is not good place to test it, would you do a new channel just for it ?

[16:17] Andrew Linden: Sigma, where is a better place to test it?

[16:17] Siana Gearz: i'm thinking, fur the purpose of testing groups, you also want to simulate people logging in, connecting to group, and logging out but not talking.

[16:18] Sigma Avro: well, where there is people chating into groups,

[16:18] Siana Gearz: i assume group has to keep track of who is connected to it.

[16:18] Sigma Avro: agni

[16:18] Joshua Linden sneaks in late and waves

[16:18] TankMaster Finesmith waves back

[16:18] Siana Gearz: and i also kind of assume the join/quit traffic can be the order of magnitude - or larger - that chat itself.

[16:18] Simon Linden: well, we sure have to test it elsewhere before putting it on the main grid

[16:18] Arawn Spitteler marks Joshua down as Tardy

[16:19] Andrew Linden: I'm guessing we'll test it on aditi (beta) then upgrade to a test deploy (RC?) to agni

[16:19] Techwolf Lupindo: I often log in and only hours later find out a group chat was active, but no traffic showed up on my end and can't join the group chat.

[16:19] Andrew Linden: I don't have any interesting news. I've just been doing code cleanup all week.

[16:20] Siana Gearz: squeaky code \o/

[16:20] Ardy Lay: I have been missing group notices.

[16:20] Andrew Linden: But other people have been fixing misc bugs and committing them to one of the maint-server repositories

[16:20] Arawn Spitteler wishes Viewer Devs could spend a few weeks so

[16:20] Andrew Linden: so next week we should have something to test on aditi

[16:20] Andrew Linden: and submit for RC the week after that

[16:21] Andrew Linden: Huseby Linden is working on fixing some crash modes and Josh Linden has done a few fixes.

[16:21] Draconis Neurocam: are the group chat servers themselves sharing traffic from agni and aditi with just an arbitrary division, or are they completely different servers, i would wonder if bandwidth from larger groups is part of the problem, and if you would even get the same issues on aditi

[16:22] Andrew Linden: I believe the group-chat servers are an indipendent stack of servers devoted to processing XMPP stuff

[16:23] Andrew Linden: so there wouldn't be an agni stack yet, just an aditi stack

[16:23] Andrew Linden: but it wouldn't be hard to bring up a new stack once the system is working well

[16:23] Pauline Darkfury: The new XMPP stack of servers. Is that setup in such a way that if it turns out to be too small when the big switchover happens that it could easily be doubled quickly?

[16:23] Andrew Linden: I was talking to Seth Linden about the group chat services several weeks ago

[16:24] Andrew Linden: he says it is built on ejabberd

[16:24] Andrew Linden: and should scale better than anything we've written in-house ;-)

[16:24] Draconis Neurocam: erlang is perfect for what they will be doing

[16:24] Draconis Neurocam: threads out of thin air

[16:25] Andrew Linden: but that claim remains untested

[16:25] Pauline Darkfury: Jabber/XMPP certainly scales large.

[16:25] Andrew Linden: and there is some amount of custom glue code we've written

[16:25] Sigma Avro: I wander how big groups will be moderated if there is no more chat lag at all !

[16:25] Andrew Linden: for auth and handling voice stuff

[16:26] Arawn Spitteler: I don't se how lag would benefit moderation

[16:27] Arawn Spitteler: Big Groups have no moderation of lag

[16:27] Sigma Avro: if text scrolls too fast without lag, large group chats will be a pain

[16:27] Pauline Darkfury: Doesn't XMPP have a standardised voice addon? Google Talk has voice, as does Apple iChat

[16:27] Draconis Neurocam: google talk only has 1 to 1 voice i thought, and it doesnt matter anyway, i think vivox would be annoyed with the loss of business

[16:27] Andrew Linden does not know how the chat+voice work was done

[16:28] Sigma Avro: text to voice is not that nice and fast

[16:28] Andrew Linden: however the group chat system would have been finished months ago if it wasn't for the voice work

[16:29] Draconis Neurocam: thats interesting to know andrew

[16:29] Draconis Neurocam: heh

[16:29] Andrew Linden: one more little item of news...

[16:29] Andrew Linden: Falcon wants me to mention that he's going to have some llCastRay() challenge soon

[16:30] Pauline Darkfury: not certain whether GTalk is 1-to-1 only on voice or not, Draconis. Part of my thought for mentioning GTalk and iChat is that if the client-server side of the new SL chat could be kept compatible with the way those put voice on top of Jabber, it could be a nice win

[16:30] Sigma Avro: hah ! like ?

[16:30] Andrew Linden: where he'll be supplying some L$ awards for people who can lag the simulator using llCastRay

[16:30] Andrew Linden: he says you have to be on the beta emial list to get the announcement

[16:30] Sigma Avro: oh, but this depends on miltations of cast ray !

[16:30] Sigma Avro: *limitations

[16:30] Andrew Linden: but I'm sure we'll re-mention it here and in other User Groups

[16:30] Draconis Neurocam: he said 5k at oskars meeting, but it has to be purely raycast related, and he will test it

[16:31] TankMaster Finesmith: has to crash the region by recast

[16:31] TankMaster Finesmith: if it also crashes w/o the raycast, you dont get anything

[16:31] Andrew Linden: hrm... if he has truely protected the system it may be an anti-climatic contest ;-)

[16:31] Ashiri Sands: clever recruitment =^_^=

[16:31] Draconis Neurocam: he also said something about crashing a sim with the new physics properties, like friction and so on

[16:32] Andrew Linden: maybe he should insert some little easter eggs for the hunters

[16:32] Andrew Linden: Ok that is all the news I had.

[16:32] Andrew Linden: The table is open

[16:33] Draconis Neurocam: is it true the region windlight stuff has been done for a while, and is just waiting on the viewer, an if so any idea how it will be set up?

[16:34] Siana Gearz: i've already got meta7 LightShare in my viewer... wonder how similar that is.

[16:34] Andrew Linden: Draconis, I think I've rumors to that effect: that the work was largely done already, but I don't know if they are true.

[16:34] Andrew Linden: So I don't know when the ETA would be.

[16:34] Draconis Neurocam nods

[16:34] Sigma Avro: about improved and independent hands animations, equivalent to face emotes ?

[16:35] Andrew Linden: Sigma, was that a question about "hands animations"?

[16:35] Andrew Linden: could you rephrase that?

[16:35] Arawn Spitteler: I'd only do those viewer side, and viewer development is focused on the one people don't like

[16:35] Siana Gearz: Arawn, grrrrr.

[16:35] Sigma Avro: lol, hand postures are now govenred by other animations, could they be independent, so priority 5 and acting only on hands ?

[16:36] Andrew Linden: Arawn, the people who like Viewer2 don't complain, so we never hear from them ;-)

[16:36] Arawn Spitteler: I don't think they know how to tyype

[16:36] Andrew Linden: Ah, feature work on the animation priorities.

[16:36] Siana Gearz: also if you don't like viewer 2, you can blame the dead. Q primarily :)

[16:36] Pauline Darkfury: Half of them probably can't figure out how the chat works in V2 ;)

[16:37] Draconis Neurocam: i only complain about things i know could be fixed and most of them have, ive gotten comfortable with v2

[16:37] Andrew Linden: Stickman asked a similar question about whether we were thinking about adding more customizable animation priority overrides

[16:37] Sigma Avro: yes, same idea, indeed

[16:37] Andrew Linden: The answer I got from Richard was: yeah, if only were were working on the animatioin system (I paraphrase)

[16:38] Sigma Avro: haha, ok. But it is making avatars less ridicoulous, so would be a huge benefit for second life image outside, to my opinion

[16:38] Andrew Linden: We've been "on the fence" about whether we would hire an animation engineer to roll our own overhaul, or try to buy some third party animation solution, or just punt the idea entirely into the future.

[16:39] Simon Linden: Agreed, it would be nice to make animations work better

[16:39] Andrew Linden: It could go any way, but I don't think we've got off the fence for that one.

[16:39] Arawn Spitteler: Separating hand and body gestures would run into timing problems. Default Anims should be updated, for systems less than half a decade old.

[16:39] Sigma Avro: no, I can time perfectly face and body animations

[16:40] Draconis Neurocam: they did update the walks and runs in v2 a little bit

[16:40] Sigma Avro: well, not asking to play piano !

[16:40] Ashiri Sands: The default walk in V2 is not ugly =^_^=

[16:40] Andrew Linden: I think Richard fixed some walk/run coordination issues when placing feet

[16:40] Andrew Linden: and that was done in viewer2 I believe

[16:40] Siana Gearz: Arawn, why so? default animations are shipped with the viewer in the VFS. older viewers would just not recognize the new ones.

[16:40] Andrew Linden: but was done months ago

[16:41] Siana Gearz: (could be made not to by deliberately not having assets on grid)

[16:41] Stickman: Sorry, I'm lagged behind in the chat history, catching up.

[16:41] Draconis Neurocam: simon, do those python scripts only work because they are run linden side, or could they work for residents as well, im sure people would be interested in them if that was the case

[16:42] Arawn Spitteler: How are default anims loaded? I know the content anims are downloaded like notecards.

[16:42] Siana Gearz: furthermore with walk etc being very implicit, there are great possibilities for upgrade.

[16:42] Stickman: Hand animations are done with morphs at the moment. It's possible to upload an "empty" animation that only triggers a morph. I don't know if morphs respond properly to priorities. There has been a lot of problems with morphs that don't trigger as expected, both face and hands.

[16:42] Sigma Avro: yes, but working on animations would be much less laggy

[16:42] Andrew Linden: I belive the python scripts talk to the "headless viewer" that was recently developed (viewer without rendering)

[16:43] Ashiri Sands: Would new hand and face animations require a new skeleton?

[16:43] Andrew Linden: I'll bet by design the bots could be outside the company network.

[16:43] Draconis Neurocam: i dont think ive seen an hg repo for the headless one, is it linden only?

[16:43] Simon Linden: There's an interface between the python code and our 'headless' viewer

[16:43] Andrew Linden: The real question is... are we going to be open sourcing the python bots and headless viewer?

[16:44] Jonathan Yap: I saw a pile of commits a while back for the headless viewer

[16:44] Siana Gearz: Ashiri, if they are made as morphs, no. if they are made as hierarchical animation, yes, or rather, the skeleton could be extended and compatibility left intact.

[16:44] Draconis Neurocam: yeah, will it be open

[16:44] Stickman: Since the original avatar files were lost, it's likely that any new morphs would require a new avatar. And morphs are dead. According to the mesh team, creation of morphs will not be supported, and bones with hardware support are the future.

[16:44] Andrew Linden: ok good, so that work is going into the viewer-development project already

[16:44] Simon Linden: That part of glue code is not open source yet ... I'd like to see it released someday, but it's changing pretty fast right now

[16:44] Andrew Linden: that is, the open-source codebase

[16:44] Pauline Darkfury: That would be a really cool thing to do, Andrew, open sourcing a basic bot framework! Loads of legitimate uses for bots where we can take some load off the LL servers and/or do clever things which are just beyond LSL's ability to do either at all or cleanly

[16:45] Siana Gearz: Stickman, avatar dead? it can be rather trivially be gained from the viewer...

[16:45] Draconis Neurocam: jonathan, in which repository, ive been watching viewer-dev like a hawk?

[16:46] Jonathan Yap: Draconis, I saw them in whatever mailing list sends out commit messages

[16:46] Draconis Neurocam: ah

[16:46] Stickman: I don't follow, Siana.

[16:46] Draconis Neurocam: ill look in backlogs then i guess

[16:46] Stickman: "Avatar dead" lost me.

[16:46] Simon Linden: FWIW the headless viewer work is in

[16:47] Siana Gearz: Stickman: well morphs for many tiny bones are itchily bad idea when you are already drawcount limited like SL is, but it's not a problem to auto generate morphs in viewer...

[16:47] Jonathan Yap: Dracons, hmmm, I could be wrong, it was a few weeks ago

[16:47] Stickman: Future creations of morphs is not on the roadmap.

[16:47] Simon Linden: The recent commits should be interesting :)

[16:47] Ardy Lay: llwindowheadless.cpp

[16:47] Simon Linden: Like the one I pushed a few minutes ago

[16:47] Stickman: Siana, convince the mesh team. I'm not advocating, I'm stating what they said.

[16:47] Ardy Lay: I see that in v-d

[16:47] Siana Gearz: Stickman: that might be true, but i don't see "original avatar files being lost" being the major limitation there.

[16:47] Draconis Neurocam: oh wow vs2010 changes just got merged

[16:48] Jonathan Yap: yeah, 400+ changesets went in

[16:48] Siana Gearz: actually i'd be both hands for replacing the avatar mesh with whatever :)

[16:48] TankMaster Finesmith: noticed that to...

[16:48] Siana Gearz: cause it's really bad.

[16:48] TankMaster Finesmith: XMPP got updated with it also

[16:48] Pauline Darkfury: Ooooh, llRegionSayTo() sounds interesting

[16:49] Sigma Avro: is that from this week ?

[16:49] Sigma Avro: yes, good idea the regionsays to

[16:49] Simon Linden: That's still being worked on ... basic functionality is OK, but I have a few tweaks for attachments

[16:50] Simon Linden: LIke if you send a message to an AV, the attachments hear it

[16:50] Arawn Spitteler: How is llRegionSayTo different from IMming an agent?

[16:50] Simon Linden: Normally, you have to send it directly to the prim that will hear the message

[16:50] Pauline Darkfury: Is it going to work for object to object comms, so that we don't have stuff firing events in many objects when 1 would do?

[16:50] Stickman: Andrew, I've been told to remind you about llSetObjectScale.

[16:50] Simon Linden: These are like chat messages, not IMs

[16:51] Simon Linden: So they respect channels, etc

[16:51] Sigma Avro: it will spare a lot of script lines

[16:51] Andrew Linden: Ah yes, llSetObjectScale(). That would be nice.

[16:51] Andrew Linden: Thanks for the reminder. I wonder when I'll get to it.

[16:51] Simon Linden: yes, and you can send messages directly between 2 objects, and no others will be able to listen() and hear them

[16:51] Sigma Avro: trough,

[16:52] Pauline Darkfury: Fantastic, that's a huge win for RLVRS devices, as they currently broadcast to everyone with a target UUID in the message

[16:52] Sigma Avro: llSetObjectScale would be grate also

[16:53] Sigma Avro: (it is now only manual or tremendous amount of script lines)

[16:53] Simon Linden: I'm also thinking of putting in a hack for messages sent down to the viewer ... perhaps ones on DEBUG_CHANNEL that start with a speical tag will never be displayed. They could feed into our python-driven programs, or perhaps be used by other systems like RLV

[16:54] Sebastean Steamweaver: Thanks Stick

[16:54] Draconis Neurocam: regionsaysto and viewer only messages would be amazing, neck deep in reading changelogs

[16:54] Sebastean Steamweaver: Someone said we're talking about object-to-object communication?

[16:54] Sebastean Steamweaver: Well, Stick said

[16:55] Stickman: Hey, don't blame me.

[16:55] Sigma Avro: in region talks

[16:55] Pauline Darkfury: Looks like we have llRegionSayTo(key target, integer channel, string message) coming soon

[16:55] Sebastean Steamweaver: Wait, coming soon?

[16:55] Draconis Neurocam: sweet

[16:55] Sebastean Steamweaver: Not just being discussed?

[16:55] Andrew Linden: there is a new LSL call in the works? llRegionSayTo() (this is the first I've heard of it)

[16:55] Simon Linden: I just pointed out an obscure change pushed to one of our viewer repos ... I've been working on llRegionSayTo()

[16:55] Sigma Avro: why not name alos ?

[16:55] Sebastean Steamweaver: Simon, I am going to send you a fruit basket.

[16:56] Draconis Neurocam: every scripter on the grid will become your best friend

[16:56] Pauline Darkfury: I'm just going from spotting it in a diff for a branch of the viewer

[16:56] Sigma Avro: well, name is risky indeed

[16:56] Sebastean Steamweaver: It's true that it can communicate directly between objects?

[16:56] Simon Linden: yes, works like chat but the message is only visible by scripts on the targeted prim

[16:56] Sigma Avro: target = object or avat I suppose ?

[16:57] Simon Linden: Normally object (and not linked ones)

[16:57] Sebastean Steamweaver: Haha! You've made my month.

[16:57] Stickman: Targeted prim, not even object?

[16:57] Opensource Obscure: yeah..thanks!

[16:57] Simon Linden: If an AV, it includes all attachments

[16:57] Simon Linden: yes, targeted prim

[16:57] Stickman: Nice.

[16:57] Sebastean Steamweaver: Simon, simon, simon

[16:57] Sigma Avro: well, beter also, to avoid griefs

[16:57] Draconis Neurocam: ive seriously had nights just pondering workarounds because of the lack of that

[16:57] Pauline Darkfury: We'd upgrade it to a deluxe fruit basket if we could do some form of simple object->object message grid-wide, Simon ;)

[16:57] Qie Niangao: even super-deluxe, for that!

[16:58] Andrew Linden: Simon, there must be some restrictions right? Can anyone "say" messages to anyone else's attachments?

[16:58] Simon Linden: grid-wide gets really complicated ... trying to find an object on the grid is tough

[16:58] Sebastean Steamweaver: Pauline, I suspect that wouldn't work for the same reason you can't deliver inventory from one prim to another over regions.

[16:58] Stickman: Anyone can do that now.

[16:58] Stickman: if (llGetOwnerKey(id) != llGetOwner()) return;

[16:58] Sebastean Steamweaver: I suspect it would work like normal listen events. The scripter would define the restrictions.

[16:59] Andrew Linden: oh, these appear to be regular listen() events on the receiving end?

[16:59] Sebastean Steamweaver: Yes

[16:59] Pauline Darkfury: Yeah, listen handlers have to filter by checking the sender as it is, this just cuts out a bunch of unnecessary script events and filtering on other objects that didn't need to hear it

[16:59] Andrew Linden: there is no way for the receiver to tell which way they were sent?

[16:59] Simon Linden: Andrew - yes, this is set up via listen calls on the reciever, so they have to filter accordingly

[16:59] Andrew Linden: ok I get it.

[16:59] Andrew Linden: excellent

[16:59] Sigma Avro: yes, but name is missing

[16:59] Sebastean Steamweaver: Name is missing?

[16:59] Pauline Darkfury: Why does it need name?

[17:00] Sebastean Steamweaver: Name is always included in listen event handlers.

[17:00] Sigma Avro: well, no, sorry

[17:00] Joshua Linden: for grid-wide object communication, what do you see as the limitations of HTTP?

[17:00] Pauline Darkfury: That's on the receiving end, the call being discussed is the sending end

[17:00] Sigma Avro: missing in the regionsayto, but for obvious reasons

[17:00] Draconis Neurocam: the urls timing out joshua

[17:00] Sebastean Steamweaver: Joshua: using HTTP is extremely unstable, unless you'[re using an outside server to orchestrate it all.

[17:00] Draconis Neurocam: you need a dns to keep them up

[17:01] Sebastean Steamweaver: If the URL or the object key changes (such as during a region reset or something else) your network is broken.

[17:01] Pauline Darkfury: Yeah, the win for some sort of grid-wide simple object comms would be the elimination of 3-5 llEmail slave scripts per object for things which don't want the complexity or cost of an off-world rendevous server

[17:02] Sebastean Steamweaver: Even llEmail is unstable, for the same reason. If the object key changes, so does the email.

[17:02] Draconis Neurocam: right, because of the 20 second email delay

[17:02] Sebastean Steamweaver: So you can't really hard-code it.

[17:02] Pauline Darkfury: An alternative would be llEmailFast() with no delay but an overall per minute spam throttle which caused it to fail if you spam

[17:02] Andrew Linden: Interesting... yeah, would need a DNS system to coordinate communication across the grid.

[17:02] Sebastean Steamweaver: Err, so does the email address.

[17:02] Stickman: PrimDNS

[17:02] Qie Niangao: well, you can fake the off-world DNS with llEmail-mediated in-world service, but that's fragile too (and it's back to llEmail)

[17:02] Sigma Avro: llInstantMessage(key of object)

[17:02] Sigma Avro: lol

[17:02] Joshua Linden: go for it, stickman! :)

[17:03] Pauline Darkfury: Yes, llIM(object_id, message) would be the ideal case from a scripter's point of view

[17:03] Stickman: Naw, I'll let someone else with more time.

[17:03] Sebastean Steamweaver: Actually, a friend of mine has been working on just that (DNS for LSL)

[17:03] Stickman: I'd love to, but can't handle it.

[17:03] Arawn Spitteler: If we could contact individual objects, by Key, could we also contact objects that are children of an original asset?

[17:03] Stickman: I think there are already some "DNS" services set up for SL for that purpose. I don't know if any are still alive. I remember hearing about them years ago.

[17:04] Opensource Obscure: stickman: some of them still work

[17:04] Sigma Avro: not possible to orchestrate llIMsfrom - to any object in sl

[17:04] Simon Linden: Well, that all comes down to the same problem -- how do you locate these objects around the grid? That service or system is going to be interesting

[17:04] Sebastean Steamweaver: Also, when is this expected to come out Simon? A month, or two?

[17:05] Sebastean Steamweaver: If Simon were able to make grid-wide communication viable, I'd like to also enable grid-wide llGiveInventory/List - I think the same problem is at the root of both.

[17:05] Sebastean Steamweaver: From what I understand, it's not easily solved.

[17:05] Joshua Linden: What simon said. And Linden's current systems which attempt to do that - email, xmlrpc, http - don't keep names stable indefinitely

[17:05] Simon Linden: I'm definitely not doing anything outside of the region right now

[17:05] Draconis Neurocam: xmlrpc is sort of a joke to scripters, everyone i even mention it to says avoid it

[17:06] Joshua Linden: yeah, just mentioned it for completeness

[17:06] Sebastean Steamweaver: Yeah, I admit, I avoid XMLrpc.

[17:06] Sigma Avro: could be that a small number of region object would accept IMs, they could tghen be used as relays

[17:06] Qie Niangao: Sebastean: I hate for llGiveInventoryList to agents to get bogged down by grid-wide *object* addressability

[17:07] Stickman: Hah, I remember playing with XMLRPC, just to do it. It was so bad.

[17:07] Sebastean Steamweaver: Qie: there's no guarantee it would. It would depend on how it was done - as it is though, I don't expect it will be, because the current system doesn't really "work" that way. So the point is a bit moot.

[17:07] Pauline Darkfury: Hmm, Simon, what about llObjectMessage(string region, key target, string message), remove the need for LL to keep track of where the target is, but give is a light weight in-world object comms system?

[17:08] Andrew Linden wonders how we would handle a DDOS attack of llObjectMessage()'s

[17:08] Stickman: We're over time and I'm late.

[17:08] Stickman: Later!

[17:09] Sigma Avro: yes, that is why I sugeste a few numbezr of object /per region allowed to get IMs

[17:09] Pauline Darkfury: Similar to the anti-DDOS on llHTTPRequest, maximum rate per minute?

[17:09] Simon Linden: Yeah, we've thought about some of this, and as soon as you go outside the region it gets very complicated

[17:09] Simon Linden: Thus we're just sticking with llRegionSayTo() for now

[17:09] Joshua Linden needs to head out, and waves

[17:09] Andrew Linden: well, you could throttle the outgoing message from one region

[17:09] Draconis Neurocam: regionsayto is still a wonderful addition

[17:09] Ardy Lay: ALmost need the region director to update a name server.

[17:09] Pauline Darkfury: Yeah, llRegionSayTo is a lovely win anyway

[17:09] Opensource Obscure: bye joshua, thanks

[17:09] Sigma Avro: well, yes, it is already great indeed , Simon

[17:09] Sebastean Steamweaver: llRegionSayTo has really made me happy

[17:10] Qie Niangao: agreed: Yay for llRegionSayTo() !

[17:10] Andrew Linden: but suppose you made 1000 bots to scatter the world and send llObjectMessage() to a particular region all at once?

[17:10] Sebastean Steamweaver: I'm just wondering when we'll be getting it - I have several projects that will use it.

[17:10] Simon Linden: I'm not sure, Sebastean

[17:10] Pauline Darkfury: Is that any worse than 1000 bots spamming llEmail, llHTTPRequest, or llRegionSay, Andrew?

[17:10] Andrew Linden: Hrm... maybe not.

[17:11] Andrew Linden: well, I think email goes through one chokepoint

[17:11] Andrew Linden: so maybe we could supply such a single point for llObjectMessage()

[17:11] TankMaster Finesmith: have a good weekend, simon

[17:11] Draconis Neurocam: thanks simon andrew and kelly

[17:11] Andrew Linden: yeah, time to go

[17:11] Ashiri Sands: thanks Simon

[17:11] Simon Linden: I'm going to get going ... thanks everyone for coming

[17:11] Opensource Obscure: thanks ... bye

[17:11] Andrew Linden: Thanks for coming everyone.

[17:12] Sebastean Steamweaver: Take care simon and Andrew, and Joshua - and everyone.

[17:12] Pauline Darkfury: Thanks, Simon, Andrew, everyone, have a great weekend :)

[17:12] TankMaster Finesmith: tc andrew, thx for all you do

[17:12] Sigma Avro: sure, thanks for extras Simon and Andrew

[17:12] Hermione Ranger: ty

[17:12] Qie Niangao: Thanks guys. :)

[17:12] Psi Merlin: Thanks /Lindens

[17:12] Andrew Linden: And imagine that... we did a whole hour of meeting without one April Fools joke... except for that llRegionSay() thing

[17:12] Pauline Darkfury: lol

[17:12] Qie Niangao: lol

[17:12] Sebastean Steamweaver: :O

[17:12] Sigma Avro: haha

[17:12] Siana Gearz: :D

[17:12] Sebastean Steamweaver: Andrew, your sense of humor is horrible

[17:12] Andrew Linden: hehe

[17:12] Sebastean Steamweaver: No fruit basket for you!

[17:13] Andrew Linden: Have a good weekend.

[17:13] Sebastean Steamweaver: You too

[17:13] Draconis Neurocam: hahaha

[17:13] Ashiri Sands: cya Andrew

[17:13] Siana Gearz: o/ byebyes lindens

[17:13] Simon Linden: Bye everyone

[17:13] Arawn Spitteler: /teletubby_bye_bye

[17:13] Pauline Darkfury: Take care call, have fun :)


Prev 2011.03.29 Next 2011.04.05