Simulator User Group/Transcripts/2011.03.01

From Second Life Wiki
< Simulator User Group
Revision as of 09:28, 10 March 2011 by Andrew Linden (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search


Prev 2011.02.25 Next 2011.03.04

List of Speakers

Andrew Linden Ardy Lay Cummere Mayo
Desdemona Enfield Draconis Neurocam Falcon Linden
Fancy Greeter Flip Idlemind Jill Yedmore
Jonathan Yap Kaluura Boa Kelly Linden
Liisa Runo Luisa Bourgoin Methias Kira
NullBit String Opensource Obscure Oskar Linden
Pauline Darkfury Qie Niangao Rex Cronon
Script Counter Shepard Vyper Simon Linden
Special Tail Stickman Syn Enyo
Techwolf Lupindo Vincent Nacon Void Singer


[12:00] Jill Yedmore: thank you guys :)

[12:00] Cummere Mayo: thanks for the usergroup, but i do wish yall would lsiten a bit more to the jiras out there

[12:01] Stickman: Hi Andrew, Simon!

[12:01] Stickman: Andrew, did you get my horribly long and complex email?

[12:01] Fancy Greeter: Oskar Linden has arrived!

[12:02] X-Carbon: All Go

[12:02] Stickman: Hi Kelly!

[12:02] Stickman: We were just talking about llName2Key in the last usergroup.

[12:02] Flip Idlemind: I changed the name of my object from "Fancy Linden Greeter"...I may have accidentally started a rumor that there's a "Fancy Linden"

[12:02] Andrew Linden: Hello everyone.

[12:02] Draconis Neurocam: hello, andrew, oskar, kelly, and simon

[12:02] NullBit String: greetings andrew

[12:02] Oskar Linden: good day ladies and prims

[12:03] Liisa Runo: Happy BD Oskar

[12:03] Andrew Linden: I'll start with my news...

[12:03] Vincent Nacon: a good one I hope

[12:03] Andrew Linden: Hrm... things are running slow in Denby today.

[12:03] Flip Idlemind: 29 people

[12:04] NullBit String: andrew, do you remember that im to email problem lonetorus habilis had?

[12:04] Oskar Linden: denby must be running maint-server ;-)

[12:04] Stickman: IM to email randomly turning off? I've had that before. Not for a while. It seems rare.

[12:04] Andrew Linden: Yes NullBit, I remember lonetorus's email problem.

[12:04] Oskar Linden: nope. mono2-aware

[12:04] NullBit String: so yeah, its fixed, thank you very much :)

[12:04] Andrew Linden: I think for lonetorus it just never worked.

[12:04] Flip Idlemind: Its the first Tuesday of the month and there's a social user group finishing up over in the middle of the sim (That was news to me)

[12:04] NullBit String: its working now

[12:04] Andrew Linden: Huh? lonetorus says it is fixed?

[12:05] Pauline Darkfury: Hi folks

[12:05] NullBit String: yes, i am his alt

[12:05] Andrew Linden: Ok, well I didn't fix anything.

[12:05] NullBit String: well, something made it work aagin

[12:05] Simon Linden considers re-aiming the bear gun for the middle of the region

[12:05] Andrew Linden: I never had time to dig into it. I suspect it was an email config thing on our servers.

[12:05] Andrew Linden: I might ask around to see if anyone has ideas as to why it was fixed.

[12:06] Andrew Linden: NullBit, what was the approximate time of the fix?

[12:06] NullBit String: yeah im just happy its working

[12:06] NullBit String: means i can stay offfline for longer periods at a time

[12:06] NullBit String: and still do customer service

[12:06] NullBit String: :D

[12:06] Andrew Linden: Yeah, I rely on the IM to email feature a lot.

[12:06] Stickman: Same.

[12:06] Shepard Vyper: :o

[12:07] Kelly Linden: Is there a jira for that bug Tournesol/Andrew?

[12:07] Rex Cronon: greetings everybody

[12:07] Stickman: Welcome, Rex.

[12:07] Shepard Vyper: Didnt know you could send an IM through email

[12:07] Andrew Linden: I think Kelly's mono2-aware project was promoted to the main grid channel today.

[12:07] Rex Cronon: hi stickman

[12:07] Kelly Linden: It was.

[12:07] Shepard Vyper: Yep

[12:07] Rex Cronon: mono gets a new life? how good:)

[12:07] NullBit String: kelly, i never opened a jira for it actually, just a few supprt tickets that amounted to nothing, then i went baawwing here at andrews ;)

[12:07] Methias Kira: I've gotten them before, it's pretty interesting. A friend used to talk to me at work that way. It was a little slower cause it's email, but it was nice, didnt know it was possible till that day.

[12:08] Pauline Darkfury: yeah, if you reply to the magic email address on the email you get from IMs, it gets converted to IM for replies, just don't send any Word docs ;)

[12:08] Shepard Vyper: Oh, nice

[12:08] Andrew Linden: I think the new project taking over the mono2-aware slot is the next maint-server, the main piece of which is Simon's "faster UDP message system" work.

[12:08] NullBit String: also it seems to collect im bursts

[12:08] Shepard Vyper: I'll try that out myself

[12:08] NullBit String: its about 5-10 min delayed

[12:08] NullBit String: and will send several im lines in one email

[12:08] Oskar Linden: so we're going to have a server-maint and a maint-server branch at the same time?

[12:08] Oskar Linden: heh

[12:08] Methias Kira: Yeah, though sometimes if you send too many lines, it splits it up.

[12:09] Methias Kira: Or wait too long between lines

[12:09] Shepard Vyper: Oh, hello, Oskar

[12:09] Techwolf Lupindo: Ooo...does this mean UDP can get faster then 1Mbs?

[12:09] Draconis Neurocam: faster udp sounds wonderful, considering how much use it gets in sl

[12:09] NullBit String: anyhow, i dont wish to divert any more time for this, since its fixed, so what news do you bring today andrew

[12:09] Andrew Linden: Heh, I didn't even know it would split the email response up. I would always try to keep my email responses short.

[12:09] Methias Kira: What Drac said.

[12:09] Andrew Linden: I was afraid it would just truncate.

[12:09] Void Singer thought the same thing

[12:09] Stickman: It splits it up? :o I don't think it used to do that.

[12:10] Simon Linden: Techwolf - it won't change bandwidth settings. I don't think there's a 1 MB cap on UDP, but the total bandwidth is divided up to different types

[12:10] Kelly Linden: I don't think it splits it up.

[12:10] Stickman: I remember getting a reply back once that said "message too long" or something. I could be remembering incorrectly.

[12:10] Andrew Linden: I've been working on a script to convert chatlogs to colorized wiki text. Halfway done.

[12:10] Methias Kira: Well i remember having IMed a few lines, then I waited a bit, sent another, and on my friends response it only had the first few IMs i sent.

[12:10] Andrew Linden: When I'm done I'll actually publish these transcripts.

[12:10] Methias Kira: Which is no biggie really.

[12:10] Pauline Darkfury: UDP can do a lot more than 1Mbit/s. Beyond 100Mbit/s, the typical 1500 byte frame size starts to become a limit, need jumbo frames to get very high bandwidth over UDP (or TCP)

[12:10] Vincent Nacon: have Lindens stands out more than residents :)

[12:10] Jonathan Yap: Andrew, check with Oz, he has a tool you might want

[12:10] Methias Kira: Just a warning, it has a timer or something to how many it sends first.

[12:11] Andrew Linden: That's all the interesting news I've got. Kelly, Simon, and Oskar... if you have anything go ahead.

[12:11] Methias Kira: Hi Vincent :3 Nice to see you.

[12:11] Vincent Nacon: heya

[12:11] NullBit String: yeah andrew, there are tools like that out there already

[12:11] Shepard Vyper: I would ask several things

[12:11] Techwolf Lupindo: The most i've ever got out of the server is 1.2Mbs. Now with http fetch on, 1.2Mbs is now allocated to !texture while http with unlimited speed handles the textures. Imporoving thing overal. Espelly tps. :-)

[12:11] Pauline Darkfury: One thought on your UDP work, Simon. Does the current server hardware allow UDP checksums to be computed in the NIC hardware, and is that enabled?

[12:12] Simon Linden: Pauline - I don't know. My work was at a higher code level

[12:12] Andrew Linden: I don't know the answer to that Pauline, I'll have to ask around in #ops.

[12:12] Pauline Darkfury: Might be worth looking at, if the goal is to squeeze as much out for as little CPU

[12:12] Simon Linden: What I discovered was the sim (and viewer, actually) spends a bunch of time verifying the packet format

[12:12] Techwolf Lupindo: (note that bits per second, not Bytes per second.)

[12:12] Shepard Vyper: Andrew, is there still that limit on the maximum users in a region?

[12:12] Shepard Vyper: default iirc is 100

[12:13] Pauline Darkfury: 40 AVs for all mainland full regions. EMs can choose any value between 1 and 100 AVs on private

[12:13] Pauline Darkfury: homesteads are 20 max, openspaces less

[12:13] Andrew Linden: yes Shepard. 100 is the max limit I think.

[12:13] Luisa Bourgoin: colorising chat for wiki rings a bell ... could do well with something that parses LSL for web pages (html)

[12:13] Andrew Linden: Some regions have a max that is lower than that.

[12:13] Shepard Vyper: There was some linden lexture a long time ago (I think 2007) which had 100 users

[12:13] Shepard Vyper: and it was mainland

[12:14] Simon Linden: The new messaging code skips verification for the message types that we know are common and work fine

[12:14] Andrew Linden: But if one were to change it, I think 100 is the max max.

[12:14] Pauline Darkfury: I have my private region set to 40, personally, more than that would just be pointless (we're not even interested in getting 40 there, just 20, maybe 25)

[12:14] NullBit String: didnt ibm put 1000 av's in a opensim?

[12:14] Oskar Linden: in a certain special case yes

[12:14] Luisa Bourgoin: who needs more than 640, ever?

[12:15] Andrew Linden: They did, with some technical tricks.

[12:15] Shepard Vyper: Were they bots or real avatars?

[12:15] Luisa Bourgoin: no prims, no walking arround

[12:15] Oskar Linden: they all were Andrew Linden avatars

[12:15] Void Singer is supposed to quietly remind andrew of

[12:15] JIRA-helper: [#SVC-255] Solid, invisible ghost prims

[12:15] Techwolf Lupindo: :-)

[12:15] Vincent Nacon: no phsic, no textures, no prim, no cowbells I'm sure

[12:15] Vincent Nacon: physic*

[12:15] Andrew Linden: I don't think they had 1000 live people driving the avatars, but they were testing the server code which doesn't care.

[12:15] Desdemona Enfield: Intel has a region driven by a 64 core experiemenal chip.. each core running a portion of the region.

[12:15] NullBit String: i suppose ibm switched off physics and spread the load across several machines

[12:16] Void Singer: oooo I like the script simon

[12:16] Andrew Linden: No, I thought IBM used a proxy to distribute the object updates. However I'm not familiar with the details.

[12:16] Shepard Vyper: Why isn't the LindenLab simulator software opensource?

[12:17] NullBit String: useful for small vw's but not practical for a whole grid i guess

[12:17] Shepard Vyper: Exploits?

[12:17] Andrew Linden: The viewer certainly did not render 1000 avatars in full glory. Dunno what they did for render proxies for members of the crowd.

[12:17] Void Singer: saleability, Shepard =)

[12:17] Andrew Linden: Yes Shepard, mostly for exploits reasons actually.

[12:17] Shepard Vyper: I remember reading a linden blog (think in 2007 or 2008) where they said the 'server' software is going opensource

[12:17] Simon Linden: Foxy - that would be a really long discussion which wouldn't get us anywhere

[12:18] Andrew Linden: Years ago we talked about open-source strategies for the server.

[12:18] Simon Linden: yeah, plans and management changed

[12:18] Vincent Nacon: and it was a bad idea

[12:18] Shepard Vyper: You could say the same about the viewer being opensource

[12:18] Shepard Vyper: :p

[12:18] Methias Kira: Yeah that sounds like a pretty terrible idea, business wise atleast.

[12:18] Andrew Linden: The main problem is that the simulator server is trusted to do a lot of stuff it doesn't really need to.

[12:18] Desdemona Enfield: actually, I would say te same about opensourcing the viewer.. not a great idea.

[12:19] Andrew Linden: Such as L$ transactions and inventory operations.

[12:19] Andrew Linden: We're in the process of moving stuff like that out of the simulator

[12:19] Andrew Linden: however it is going to take a while

[12:19] Shepard Vyper: Well...

[12:19] Void Singer: Des, that was rather a moot point at the time... proxies and libSL were already there

[12:19] Stickman: Hey Kelly, I was wondering what the official reason for a lack of llName2Key is, if there is one. Or if it's simply a low priority to be done at some point in the future.

[12:19] Vincent Nacon: and IP addresses but... of course I know about the other leak that has been happening anyway

[12:19] Syn Enyo: Would it be possible to get a llWideGiveInventory(string simulator, key destination, string item); ? To send a object to another object where allow inventory drop is true in another region? This funtion would really be useful for one of my product plans.

[12:20] Pauline Darkfury: Opensourcing the viewer has had some downside, but frankly the benefits have also been huge. There's a lot of innovation has come from outside LL which is making its way into the viewer under Snowstorm

[12:20] Shepard Vyper: Why worry about that, if you make the connection of a region to the server, require a 'key'

[12:20] Methias Kira: Im almost 100 percent certain that any exploits that have been done through the viewer side, would be able to done 100x worse through an opensourced server, that would be like saying "Here, do your worst." lol

[12:20] Opensource Obscure: JIRA about what Syn said >

[12:20] JIRA-helper: [#SVC-864] Allow llGiveInventory to work across region boundaries

[12:20] Techwolf Lupindo: Open the server sofware up with kill LL busness as that is there main assest. Opening up the viewer was good as the community has contrubtied quiet a bit of improvments. The server run on LL server, so why have the source code? The cleint run on the user computer, so there was a good case to open up the viewer.

[12:20] Kelly Linden: llName2Key is difficult to do globally

[12:21] Vincent Nacon: you mean llUsername2Key

[12:21] Kelly Linden: no, I don't. Though it is also true for that.

[12:21] Vincent Nacon: muhahahaha!

[12:21] Desdemona Enfield: I think the white hats have a revers avatar key lookup web siervice.

[12:21] Stickman: Good to know, thanks. I've been wondering about that for a while.

[12:21] Methias Kira: The viewer was a good idea to go open source, even with it's repercussions, because it allowed the users to feel more personally involved in SL in my opinion. You could use SL how you wanted to use it, or make it how you wanted to through the viewer, to a certain extent of course.

[12:21] Draconis Neurocam: just have a ll run key to name service for http request

[12:22] Shepard Vyper: Eh

[12:22] Flip Idlemind: What makes llName2Key harder than llKey2Name (or llRequestAgentData)

[12:22] Techwolf Lupindo: Case in point Dave. RLV viewer

[12:22] NullBit String: kelly, but its prety static data, why not have a llname2key function feed off a seperate database with just those key data in it?

[12:22] Void Singer: hey simon, is that jira title fetch script available for viewing? I rather like how that works

[12:22] Pauline Darkfury: There are a few different ways to do namr2key, just none particularly light and quick, all involve some sort of web stuff

[12:22] Methias Kira: Open sourcing the server would only be of any help to opensim people hahaha

[12:22] Vincent Nacon: because you have to run through an index of all database at once

[12:22] Shepard Vyper: The only reason I like OpenSim, is because you're able to host your own region

[12:22] Simon Linden: Void - I think Moon Metty wrote it ... it's in the blue triangle over behind Rex

[12:22] Andrew Linden: brb

[12:23] Shepard Vyper: I lol'd

[12:23] Stickman: Andrew, I send you an email a while back, February 18th, regarding animations. You were going to pass it along to Richard Linden. I was wondering if anything came of it.

[12:23] Shepard Vyper: the blue triangle

[12:23] Methias Kira: Which is why it's opensim, and not LL. LL would go out of business if they open sourced the server lol.

[12:23] Stickman: Op. He left.

[12:23] Kelly Linden: and as quick as that it has moved past trivial to implement and maintain service as I see the suggestions for its own db scroll by

[12:23] Rex Cronon: i wanted to make room for others to see it:)

[12:24] Shepard Vyper: Haha

[12:24] Stickman: LL still provides a service. They wouldn't "go out of business.' BUt they would be basically growing competition. Which isn't very smart unless you're very sure of your position and want to share.

[12:24] Shepard Vyper: You're looking at it wrong

[12:24] Andrew Linden: back

[12:24] Stickman: It's easy! Just make a second asset server that doesn't hold as much information as the other one.

[12:24] Shepard Vyper: OpenSim is pretty much the same thing, just different code

[12:24] Simon Linden: Stickman - I don't think anything has been planned for animation work yet, but it seems to be an ongoing topic. It's one of my favorite projects that I think we should do

[12:24] Desdemona Enfield: A review of Prok's comments on open source would be interesting.

[12:24] Shepard Vyper: And LL is still strong and growing

[12:24] Pauline Darkfury: I think the benefit from having an official name2key function from LL would be to save a bunch of script pain and load in a lot of places, plus would remove some load on the LL operated AV search

[12:25] Vincent Nacon: anyway... let's not get caught up in opensource sim topic

[12:25] Methias Kira: It would be counter productive in a business sense, they would be ASKING for competition, rather than Opensim just reverse engineering.

[12:25] Stickman: The email I sent to Andrew outlined my biggest thoughts and issues I've had with animations over the past couple years.

[12:25] Desdemona Enfield: How much are you willing to pay to LL in compensation in order to pursuade them to opensournce the simulator>

[12:25] Shepard Vyper: I would pay upto 5k USD for the binary to host my own region

[12:25] Stickman: Primarily, accuracy. Second, a superb lack of information about animations.

[12:25] Desdemona Enfield: 100 million? 200 million?

[12:25] Shepard Vyper: Binaries*

[12:26] Luisa Bourgoin: 1000 L$

[12:26] Shepard Vyper: And that's for a private use verion

[12:26] Vincent Nacon: Andrew, was wondering what was the recent news on Havok 7's vehicle library if any?

[12:26] Simon Linden: yeah, stuff like that ... syncronizing would be nice, plus eliminate the need for basic AO functionality

[12:26] Methias Kira: Thats why you go to opensim if you want to host your server.

[12:26] Shepard Vyper: version*

[12:26] Desdemona Enfield: You canbuy Dakota, yes? or was it called nebraska?

[12:26] Stickman: I love it when people offer L$500 for a week of my time.

[12:26] Ardy Lay: Nebraska seems to be cancelled.

[12:26] Stickman: Non-scripted AOs would be nice. But they were actually fairly low on my list. Synchronization was higher.

[12:26] Andrew Linden: Stickman, I think that email fell off my radar.

[12:26] Desdemona Enfield: (yes, noticed that too)

[12:27] Andrew Linden: I'll try to find it and forward to Richard somehow.

[12:27] Desdemona Enfield: Simon or Andrew, do you know when the per percel script memory limits are going to go into effect?

[12:27] Luisa Bourgoin: whaaat? 500 L$ ? ...totally overpaid...

[12:27] Stickman: Should I send it again? I could go straight to Richard if I knew his email. I would guess it's

[12:27] Void Singer: it'll make sense when if another business engineers something close enough to be a real competitor, then LL can market on strength of experience, or position for directional control and compatibility

[12:27] Simon Linden: I think the benefit of automatic AOs would be ease of use ... make it simpler for everyone to get custom animations

[12:27] Simon Linden: ... plus less server scripts

[12:27] Vincent Nacon: aye

[12:27] Kelly Linden: Des: "not yet"

[12:28] Stickman: Specifically for animations, a script-based AO gives you more power. So while it'd be nice on the server and load, it'd give you less options.

[12:28] Opensource Obscure: (( before adding new topics, Syn had a question I was interested in, may we talk about that ? ))

[12:28] Methias Kira: I think the big question is Mesh :3

[12:28] Andrew Linden: Vincent, no progress on Havok7 vehicles. Falcon has been working on the mesh project

[12:28] Syn Enyo: Would it be possible to get a llWideGiveInventory(string simulator, key destination, string item); ? To send a object to another object where allow inventory drop is true in another region? This funtion would really be useful for a few of my product plans. -

[12:28] JIRA-helper: [#SVC-864] Allow llGiveInventory to work across region boundaries

[12:28] Simon Linden: Very true

[12:28] Andrew Linden: and has a bunch of other projects he wants to get done now.

[12:28] Desdemona Enfield: AO's tend to run at timer event intervals of <0.1 sec.. It would be nice to have an changed() service called CHANGED_LOCOMOTION_STATE.

[12:28] Techwolf Lupindo: Being able to sync animations would make for some nice >1 avatar animations.

[12:28] Andrew Linden: I'm not sure where vehicles sit in Falcon's prioritized list these days.

[12:29] Vincent Nacon: does that means your position is no longer on physic department?

[12:29] Stickman: I'm more interested in implanting the ideas my ideas into his head and hoping they find root for later, than being immediately heard so they can be implemented immediately.

[12:29] Methias Kira: Atleast we have the beta grid though. Its not as bad as it was, when mesh was just a fleeting rumor. Now we can actually see it somewhere atleast

[12:29] Shepard Vyper: Vehicles should be lifted from the 32 (correct me if I'm wrong) max object ;o,ot

[12:29] Simon Linden: Also better pre-loading of animation

[12:29] Andrew Linden: Correct Vincent. Falcon is in charge of the physics engine now.

[12:29] Shepard Vyper: limit*

[12:29] Vincent Nacon: alrighty, such a shame I guess

[12:29] Shepard Vyper: physical objects

[12:29] Void Singer: syn, I dunno about using allow drop for it, but otherwise I like it

[12:29] Shepard Vyper: bleh

[12:30] Shepard Vyper: Max of linked physical objects should be put at 50

[12:30] Meeter: Timecheck : User Group is half over

[12:30] Methias Kira: Using mesh for vehicles might get tricky with the 32 prim limit.

[12:30] Vincent Nacon: no Shepard

[12:30] Andrew Linden: Shepard the mesh project should allow people to make vehicles that look more complicated than they collide.

[12:30] Vincent Nacon: aye

[12:30] Andrew Linden: Not just mesh shapes, but the ability to override the collision shape of a linked object.

[12:30] Shepard Vyper: I was messing the mesh vehicles, and they are indeed nice

[12:30] Methias Kira: Yes.

[12:30] Stickman: Custom physical shapes make more sense than just raising the prim limit.

[12:30] Shepard Vyper: with*

[12:30] Vincent Nacon: there are no need to have more than 32 physicl enabled prims anymore

[12:30] Shepard Vyper: I created myself a floating 'land speeder'

[12:31] Pauline Darkfury: Yeah, presumably the 32 limit will apply only to collision shape?

[12:31] Andrew Linden: Yes Pauline.

[12:31] Methias Kira: Will a mesh vehicle count 32 prims at 32 individual meshes as the limit, or 32 prims as in the calculated prim cost of x amount of meshes.

[12:31] Luisa Bourgoin: in a way, if you WEAR your vehicle ... you get similar effects. different collision object and > 32 prims

[12:31] Vincent Nacon: you can have up to 256 prims (in a linkset) with 32 prims enabled as physic while rest is phantom

[12:31] Shepard Vyper: Triangles

[12:31] Vincent Nacon: so... having more is no need

[12:32] Andrew Linden: We may re-evaluate the 32 prim limit also... 32 concave prims is not the same as 32 boxes, so the simple prim limit is not quite the right metric.

[12:32] Shepard Vyper: ^

[12:32] Luisa Bourgoin: 32 twisted prims could wreck it?

[12:32] Andrew Linden: Mesh objects will have a more complicated "physics cost" metric.

[12:33] Pauline Darkfury: Yeah, would make sense to translate the 32 limit into a more cost based limit under the new model

[12:33] Andrew Linden: Falcon knows the details. They are too complicated for me to remember off the top of my head.

[12:33] Void Singer: I tapped on it earlier, but was wondering, has there been any progress on "Ghost Prims"... seems like a a queueing issue, where the region is waiting for a response that it never gets due to poor connection to clients or extreme slowdowns... they seem to cripple a region as they accumulate

[12:34] Andrew Linden: Void, we don't know of a reliable way to reproduce "ghost prims"

[12:34] Oskar Linden: I've been looking for one

[12:34] Andrew Linden: btw, there are two types of problems often referred to as "ghosts"

[12:34] Jonathan Yap: I think these ghost prims are ones the physic engine still knows about

[12:35] Andrew Linden: there are the "non colliding and non-selectable ghosts" -- object is gone on the server but your viewer thinks it is still tehre

[12:35] Luisa Bourgoin: on my map, it's some prim that one can bounce into ... besides it no more beeing there.

[12:35] Void Singer: I have noticed, most are marked as selected/sat upon, and the owners are never around... maybe due to disconnects?

[12:35] Pauline Darkfury: I saw ghost prims where an object had been taken/deleted/returned, but not removed from physics, much more often a while ago. I don't see them very often at all these days

[12:35] Oskar Linden: if you guys can find a good solid repro let me know

[12:35] Andrew Linden: and then there are the "invisible collision ghosts" -- still in the physics engine but the server is not sending you the updates

[12:35] Pauline Darkfury: Can't remember the last time I actually saw one, tbh

[12:35] Luisa Bourgoin: Pauline, indeed, these seem to get rare

[12:35] Andrew Linden: I *think* that the first type of ghost has largely vanished. Is that true?

[12:35] Void Singer: definitely the colliding variety I think... at least that what detection seems to pick up

[12:36] Shepard Vyper: Yes

[12:36] Andrew Linden: I thought that the simulator now sends a "that object is dead" packet when you try to select a non-existent object, which cleans up the first type of ghosts.

[12:36] Liisa Runo: there used to be a way to make ghost prims but it dont work anymore. but i might find a new way, ill look in to it

[12:36] Andrew Linden: ok so that leaves the invisible colliders

[12:36] Luisa Bourgoin: even if they are gone, we can still tell each other's "ghost prim stories" :)

[12:36] Shepard Vyper: haha

[12:37] Oskar Linden: around the particle fire

[12:37] Stickman: Hah.

[12:37] Void Singer: now I've seen a few visible but disapear on selection... i just assumed those were viewer artifacts though, are theose the non-colliding type?

[12:37] Desdemona Enfield: : )

[12:37] Simon Linden: Andrew - yes, if you try to select something that's not there, you should get a kill message for it

[12:37] Shepard Vyper: xD

[12:37] Desdemona Enfield: I have seen that too Void.

[12:37] Andrew Linden: The Magnum channel currently has some code that cleans up the object deletion pipeline, which I hope will reduce the number of invisible colliders.

[12:37] Desdemona Enfield: IN my case, what creates them is repeated rezzing of temp on rez objects.

[12:38] Andrew Linden: Yes Void, I believe those are the first type of ghosts, if they disappear on selection.

[12:38] Andrew Linden: oh really Desdemona? You can reproduce the invisible colliders problem?

[12:39] Desdemona Enfield: I am seeing those 'diappaear on select things'

[12:39] Desdemona Enfield: I wonder why they are there.. I go to edit them and poof.

[12:39] Void Singer had 1500+ on home region a week or so ago.... but it was a weekend, so I no one to come check out the creation time

[12:39] Stickman: I've encountered the collision ones before. Can't say I know the cause. Resetting the region got rid of them.

[12:39] Desdemona Enfield: Very rare.. not a constant effect.

[12:39] Andrew Linden: Falcon says "Havok vehicles" are about item #17 out fo 20 these days.

[12:40] NullBit String: thats another feather in falcons hat, that ghost prims that are still resent in the phys simulator are mostly gone since he started messing around XD

[12:40] Vincent Nacon: muhahahaha! well you can tell him to relax

[12:40] Pauline Darkfury: I sometimes still get an object which is slow to vanish on take/delete/return, but just selecting with a right-click for the menu usually makes it poof from the world. I thought that was just due to an update packet getting lost/not sent to my viewer, tbh, but could be something else

[12:40] NullBit String: present

[12:40] Desdemona Enfield: Pauline.. that my hypothesis too.

Total Avatar Scripts - 745

Highest - Rex Cronon: 106 (14%)

Lowest - Vincent Nacon: 0 (0%)

[12:40] Simon Linden: - that scenario makes sense. Your temp objects are getting automatically deleted when they get too old, and at some point the message telling your viewer to delete one gets lost

[12:40] Stickman: Hi Rez.

[12:41] Stickman: Rex.

[12:41] Syn Enyo: What about a llWideGiveInventory(string simulator, key destination, string item); funtion To send a object to another object where allow inventory drop is true in another region? This funtion would really be useful for a few of my product plans. -

[12:41] JIRA-helper: [#SVC-864] Allow llGiveInventory to work across region boundaries

[12:41] Rex Cronon: hi stickman. did u type that at the begining of the OH?

[12:42] Stickman: Yes, I was commenting on the script counter, though.

[12:42] Andrew Linden: llWideGiveInventory()?

[12:42] Syn Enyo: To give a object to another object in another region.

[12:42] Stickman: I'm cruious how my dragon would compare to you, though. I didn't write down last time I counted his scripts.

[12:42] Qie Niangao: About a grid-wide llGiveInventory.... I assume that wouldn't do anything special with scripts... so still not a grid-wide llRemoteLoadScriptPin(), right.

[12:42] Syn Enyo: So you would have to put the region name, object key you send to and object to send

[12:42] Stickman: Cut out 80% of his scripts thanks to Kelly's new functions a while back, though.

[12:42] Rex Cronon: i have a scanner

[12:43] Rex Cronon: try now

[12:43] Desdemona Enfield: Yes, thank you for llGetLinkPrimParms()

[12:43] Pauline Darkfury: you'd presumably be able to give scripts with it and then have an existing script set them running, if you wanted that

[12:43] Void Singer still wants PRIM_POS_LOCAL though >=)

[12:43] Andrew Linden: Hrm... grid wide llGiveInventory()... sounds hard. I wonder what to do about all the various failure modes.

[12:43] Stickman: Yes, PRIM_POS_LOCAL would be very desirable.

[12:43] Fancy Greeter: Falcon Linden has arrived!

[12:43] Andrew Linden: Such as the recipient being offline at the moment of the send.

[12:43] Stickman: Hi Falcon!

[12:44] Falcon Linden: Ahoy

[12:44] Syn Enyo: So you would have to know the sim the object you wanted to send to was in. Becuase i guess if you was going to add it on to the llGiveInventory funtion, it guess it would take time to find the object.

[12:44] Pauline Darkfury: What is much more needed for commerce purposes is llGive* to give some sort of success/fail return, either at the time, or via a later dataserver event

[12:44] Vincent Nacon: oh now he decided to show up

[12:44] Shepard Vyper: Oh, Hello, Falcom

[12:44] Vincent Nacon: muhahahaha!

[12:44] Shepard Vyper: Loving the avatar

[12:44] Andrew Linden: Yes, and what about when the recipient calls llDie() after the sender sends, but before the object can be given.

[12:44] Simon Linden: I had to wake him up, Vincent

[12:44] Pauline Darkfury: SVC-595 is the big one that allows a form of fraud, but it's needed on llGiveInventory as well

[12:44] JIRA-helper:

[#SVC-595] llGiveMoney() method needs a money_given() event handler.

[12:44] Vincent Nacon: well I didn't say you should

[12:44] NullBit String: greetings falcon, too busy to join us on time?

[12:44] Draconis Neurocam: heya falcon

[12:44] Stickman: llGiveInventory only has a region limit for object giving. It's grid-wide for avatars.

[12:44] Andrew Linden: I guess if there were a RESTful API it could work.

[12:45] Vincent Nacon: but thanks?

[12:45] Vincent Nacon: :)

[12:45] Falcon Linden: Tournesol: I thought Tuesday meetings were at I never come :)

[12:45] Vincent Nacon: wrong meeting

[12:45] Andrew Linden: Avatars are special. They have some dedicated presence in the database.

[12:45] Stickman: llGiveInventoryList requires them to be in the same region. Or recently have been there.

[12:45] Rex Cronon: u didn't get the memo about the hour change:)

[12:45] Falcon Linden: I'm too busy breaking things to get here on time!

[12:45] Shepard Vyper: Oh look

[12:45] Syn Enyo: Well for my objects, i can already track them in a DB of mine since i track the http-in url.

[12:46] Flip Idlemind: llGiveInventoryList() on the other hand, is not grid wide for avatars

[12:46] Shepard Vyper: He's sitting next to me

[12:46] Shepard Vyper: Mr Grench

[12:46] Syn Enyo: So knowing the region name is not a big deal for me...

[12:46] Luisa Bourgoin: I really like the aspect that there are still office hours, that are not office hours of course, since they are called different now :)

[12:46] Falcon Linden: Foxy: It's spelled Gr/i/nch :)

[12:46] Shepard Vyper: Is it really?

[12:46] Falcon Linden: Yep

[12:46] Shepard Vyper: I had no clue :c

[12:46] Andrew Linden: "User Groups" Luisa

[12:46] Falcon Linden: Foxy: In that case, no worries!

[12:46] Shepard Vyper: :D

[12:46] Desdemona Enfield: Or user groupies, smiles.

[12:46] Vincent Nacon: anyway, I've sent you a reply about the llCastRay being used for vehicle shock system on the wiki

[12:47] Falcon Linden: Really?

[12:47] Falcon Linden: I haven't looked at that wiki in ages. I should. Let me do so now.

[12:47] Vincent Nacon: still the same anwser I gave you for the first time. :)

[12:47] Vincent Nacon: the "NO" I gave

[12:47] Vincent Nacon: muhahahaha!

[12:47] Stickman: What are your current top items, Falcon? What's on your todo list?

Total Avatar Scripts - 698

Highest - Pauline Darkfury: 106 (15%)

Lowest - Vincent Nacon: 0 (0%)

[12:48] Vincent Nacon: here's a link for you

[12:48] Falcon Linden: Oh, Vinjcent, you misunderstood

[12:48] Falcon Linden: You don't use it WITH wheels

[12:48] Vincent Nacon: oh I know

[12:48] Falcon Linden: you use it INSTEAD of wheels. And just apply impulses to the vehicle from below..assuming we can apply point impulses

[12:48] Falcon Linden: You can apply a point impulse in lsl, can't you?

[12:49] Falcon Linden: as in "llApplyImpulse(position, impulse)"

[12:49] Falcon Linden: or some such?

[12:49] Kaluura Boa: I wish!

[12:49] Rex Cronon: u mean push?

[12:49] Falcon Linden: no

[12:49] Vincent Nacon: um... that's not what I was going for, actually

[12:49] Draconis Neurocam: llApplyImpulse( vector force, integer local );

[12:49] Falcon Linden: Of course, you're right, it's not a real solution for vehicles anyway. I was just curious if it could be made to work

Total Avatar Scripts - 592

Highest - Bronson Blackadder: 100 (17%)

Lowest - Vincent Nacon: 0 (0%)

[12:49] Falcon Linden: draconis. That's unfortunate.

[12:49] Void Singer: no point impuls, only whole object COM

[12:49] Falcon Linden: As for my to do list, it's rather long. Right now I'm still focused on mesh

[12:49] Desdemona Enfield: On ApplyIMpulse, you have to couple it with a torque.

[12:50] Vincent Nacon: no worry, finish mesh first

[12:50] Falcon Linden: right. Unfortunate...

[12:50] Falcon Linden: After mesh? Who knows

[12:50] NullBit String: llApplyImpulse(vector force, integer local)

[12:50] Desdemona Enfield: It's an interesting physics problme.

[12:50] Falcon Linden: It'll be up to the execs and company direction I suppose

[12:50] Falcon Linden: I have some ideas I'm floating, though

[12:50] Draconis Neurocam: llSetForceAndTorque( vector force, vector torque, integer local );?

[12:51] Falcon Linden: I think we'll be seeing some great things in the coming months

[12:51] Shepard Vyper: Well I'm heading out. Cya!

[12:51] Desdemona Enfield: That might be better be llSetImpactParams( [ IMPLACT_TORQUE... vector.. and so on.. ]) extensible function.

[12:52] Andrew Linden: The various llApplyTorque/Impulse() methods don't work well because of the asnynchronicity between the script engine and the physics engine.

[12:52] NullBit String: llSetBuoyancy(1,falcons_ideas);

[12:52] Falcon Linden: nice, Tournesol :)

[12:52] Andrew Linden: If you want continuous force updates you need a "physics action" that is applied very physics step.

[12:52] Void Singer: besides a solid repro on ghosts, I assume there is nothing else that will help ( like creation time witin a few minutes)?

[12:52] Desdemona Enfield: llSetMass( "heavy, man.");

[12:52] Falcon Linden: I'm still deciding whether to enable llCastRay on agni as soon as it's possible. I may want to take into account some additional user feedback

[12:53] Pauline Darkfury: A question/observation which may interest you in particular, Falcon. I've observed (and a number of others have seen similar) that we seem to have an elevated "Physics Other" time being reported from sims that have a fairly high total prim count (not lots of physical objects, just lots of objects). Around 2.0ms showing for that on the sim stats, but very low "Physics Step". Top Colliders doesn't show any heavy activity. The theory is something has changed or been introduced sometime in the last month or 2 (maybe longer). The question part is: what does "Physics Other" actually represent?

[12:53] Falcon Linden: Andrew, weren't you just looking at that the other day?

[12:53] Andrew Linden: Pauline, that Physics Other time is a recent thing right? last several weeks?

[12:53] Pauline Darkfury: Yeah, apologies if it's already been raised, I've not made it to the last few OHs

[12:53] Pauline Darkfury: Yes, last month, maybe 2

[12:54] Andrew Linden: No, I haven't been able to look into it yet.

[12:54] Pauline Darkfury: Might be a newer thing than that, can't quite put a date on it

[12:54] Andrew Linden: Xan Linden brought it to our attention though.

[12:54] Pauline Darkfury: I see it on both Class 5 and 7 full regions

[12:54] Pauline Darkfury: Empty regions (very few prims rezzed) don't seem to show it

[12:54] Falcon Linden: I'm looking at the code right now

[12:55] Pauline Darkfury: Regions with say 13-15k prims rezzed show it the most

[12:55] Meeter: Timecheck : User Group is almost over

[12:55] Stickman: You have five minutes.

[12:55] Andrew Linden: Ok thanks for the tip. It is on my list of things to look into.

[12:55] Falcon Linden: It's not obvious what changed that might have caused that

[12:56] Falcon Linden: But we'll definitely be looking at it

[12:56] Pauline Darkfury: Thanks. As far as I can see, it's quite widespread. Obviously I only have Top Colliders info on a narrow set of cases, but the issue seems there. Could be a metrics bug, could be a physics bug, not sure

[12:56] NullBit String: simon, did i make the railing too small? (this is the first time i see it rezed here)

[12:56] Vincent Nacon: I have a gut feeling that it's related to avatar walking on things

[12:56] Falcon Linden: Good news, though: I'm fairly sure that well built mesh vehicles with well built mesh landscapes will perform far better than traditional vehicles

[12:57] Vincent Nacon: no it won't, Falcon.... maybe on physic score wise

[12:57] Simon Linden: No, it's fine ... we had to add another object to sit on, however

[12:57] Oskar Linden: "well built" means to Falcons high standards

[12:57] Falcon Linden: Very true, Oskar

[12:57] Flip Idlemind: What about terribly built mesh vehicles on crappy landscapes

[12:57] Falcon Linden: very very true

[12:57] Stickman: Are you going to release a standards guide for "well built?"

[12:57] Falcon Linden: FLIP: Those will hopefully have prohibitively high physics resource costs :)

[12:57] Desdemona Enfield: LOL flip.

[12:57] NullBit String: falcon do you really think resies can do meshes this good?

[12:57] Falcon Linden: Of course I do

[12:57] Vincent Nacon: I can

[12:58] NullBit String: falcon, can you provide guidelines at least

[12:58] Vincent Nacon: problem is not everyone understand physic law

[12:58] Rex Cronon: too bad though that it only takes a person to select a vehicle to stop it in its tracks:9

[12:58] Falcon Linden: Tournesol: that's the plan

[12:58] Rex Cronon: :(

[12:58] NullBit String: where on your todo list is the best practices wiki section?

[12:58] Falcon Linden: I'm also hoping that the physics resource costs on objects will help convey their relative performance quality

[12:58] Falcon Linden: tournesol: definitely on there.

[12:58] Stickman: One of the things I really want from Linden Lab is more information. Hardware that the users have, features that their clients support. I don't even know if I should still be supporting invisiprims or not.

[12:59] Vincent Nacon: it will... if the builder understood how to use the physic box in mesh

[12:59] NullBit String: yeah such information would be really nice to have

[12:59] Falcon Linden: vincent: well, if they don't, they'll probably just end up with a single convex hull and that'll perform fine ;)

[12:59] Stickman: I don't know if I should make an avatar and expect it to fit in a specific style of environment, or if it can take up more resources because things are generally sparse.

[12:59] Vincent Nacon: but I seriously doubt all the car/bike builder understood that

[12:59] Vincent Nacon: right

[12:59] Falcon Linden: I'm hoping we can build a museum of example content

[12:59] Vincent Nacon: so nothing would change much other than its appeal

[12:59] Falcon Linden: Good and bad

[12:59] Oskar Linden: Falcon needs to release a howto guide for perfect sl vehicles

[13:00] Stickman: I'd love that, Falcon.

[13:00] Pauline Darkfury: invisiprims should be mostly deprecated now, I'd guess, although still many products out there in need of update to use alpha masks

[13:00] Meeter: Thank you for coming to the Server User Group

[13:00] Falcon Linden: Falcon needs to release a howto guide for physics performance. Vehicles? Haha

[13:00] Vincent Nacon: Why oskar? to ruin my good chance at market?

[13:00] Falcon Linden: I've never written a vehicle script

[13:00] Vincent Nacon: silly you

[13:00] Flip Idlemind: Yeah, is it on anyone's to do list to make physical objects not freeze when you select them?

[13:00] Falcon Linden: I just use the kart one and add super gravity :)

[13:00] Stickman: Who many users are still on 1.23 (like me) or clients that don't support alpha skin? I don't know.

[13:00] NullBit String: falcon, also out of this information vacuum on best pracice, grows horrible monster jrewlery with 1024x2 textures and 512 scupt prim shoes

[13:00] Oskar Linden: for the betterment of all SLkind

[13:00] Falcon Linden: Tournesol: don't get me started

[13:00] Void Singer: pauline, only as used in avatar masking.... there are still other uses

[13:01] Desdemona Enfield: Are there any projects for obsoleting the 1.23.5 viewer?

[13:01] Stickman: At an old Q office hours, someone presented a sneaker shoe. 30MBs of textures.

[13:01] Falcon Linden: All right, folks, thanks for coming! (Wait, Andrew leads this User Group...)

[13:01] Vincent Nacon: all cars will still be driving with rotation from the center instead of from the rear

[13:01] Desdemona Enfield: porojections?

[13:01] Pauline Darkfury: Yeah, fair point, Void, there are some interesting cases for a rezzed invisiprim

[13:01] Desdemona Enfield: nvm.. that is a viewer question.

[13:01] Oskar Linden has to head off

[13:01] Vincent Nacon: so I'll be the only one who fixed that

[13:01] Stickman: Qarl, not Q. Excuse me.

[13:01] Oskar Linden: thanks for coming guys

[13:01] Stickman: Thanks for the usergroup!

[13:01] NullBit String: stickman, it was me, (lonetorus)

[13:01] Vincent Nacon: thanks

[13:01] Rex Cronon: tc falcon, oskar

[13:01] Desdemona Enfield: thank you.

[13:01] Andrew Linden: So much work to do. So little time. Sigh...

[13:02] Pauline Darkfury: Thanks, techy Lindens :)

[13:02] Stickman: Hah, ain't that the truth.

[13:02] Special Tail: Ardy Lay pulls F L I P's tail while eating a cookie

[13:02] Void Singer already has an engine tuning algorithm for vehicles... took a while to find all the curves (stupid wiki)

[13:02] Andrew Linden: Thanks for coming everyone.

[13:02] Rex Cronon: tc andrew

[13:02] Simon Linden: Des - I don't think there is a solid date or anything, but we're definitely trying to get all viewers onto the newer codebase. Let's not discuss the UI, but the underlying code is definitely better there

[13:02] Pauline Darkfury: Have a nice afternoon, Lindens :)

[13:02] Stickman: Later, thank you!

[13:02] Ardy Lay: Looks like I need a new UPS.

[13:02] NullBit String: yes, thank you for the office hour :)

[13:02] Qie Niangao: thanks Andrew, Simon, Kelly. :)

[13:02] Simon Linden: Thanks everyone for coming and the good discussion

[13:02] NullBit String: err

[13:02] Desdemona Enfield: LOL. well, part of the UI issue is resistance to change.

[13:02] NullBit String: user group

[13:02] Rex Cronon: tc simon

[13:02] NullBit String: ..hour

[13:02] Draconis Neurocam: thanks all the lindens

[13:03] Simon Linden: Bye all :)

[13:03] Rex Cronon: tc everybody and have nice day:)

[13:03] Desdemona Enfield: bye


Prev 2011.02.25 Next 2011.03.04