Simulator User Group/Transcripts/2011.08.05

From Second Life Wiki
Jump to: navigation, search

Simulator_User_Group

Prev 2011.08.02 Next 2011.08.09

List of Speakers

Aelfwine Crystal Andrew Linden Arawn Spitteler
Ashiri Sands Helena Lycia Jane1 Bookmite
Kelly Linden Leonel Iceghost Liisa Runo
Odessa Ibanez Pauline Darkfury Rex Cronon
Simon Linden Tortured Scientist xstorm Radek

Transcript

[16:00] Meeter: Welcome to the Server User Group

[16:00] Helena Lycia: Hello Andrew

[16:00] Aelfwine Crystal: Hello everyone, Andrew....all :)

[16:00] Andrew Linden: Hello.

[16:01] Ashiri Sands: hi Andrew

[16:01] Andrew Linden: Simon Linden says he'll be a little late.

[16:01] Andrew Linden: He's downloading a viewer update.

[16:01] Jane1 Bookmite: Hi Simon

[16:02] Helena Lycia: Don't tell me he's secretly switching to Firestorm?

[16:02] Jane1 Bookmite: oops, Hi andrew

[16:02] Helena Lycia: Hiya Rex

[16:02] Rex Cronon: hello everybod

[16:02] Ashiri Sands: Viewer updates have been very frequent

[16:02] Rex Cronon: hi helena

[16:03] Liisa Runo: hiyas

[16:03] Rex Cronon: hi

[16:03] Simon Linden: Hello everyone

[16:03] Helena Lycia: Hiya Simon

[16:03] Jane1 Bookmite: Hi Simon

[16:03] Tortured Scientist: hello

[16:03] Rex Cronon: hi simon

[16:03] Andrew Linden: Jane1, have you ever read _Neuromancer_ by William Gibson?

[16:04] Jane1 Bookmite: I have not.

[16:04] xstorm Radek: rez dam it

[16:04] Simon Linden: llRezDamnIt()

[16:04] Andrew Linden: There was a character in that book named "Lady 3Jane", or just "3Jane".

[16:04] Andrew Linden: You could have picked 1Jane.

[16:04] xstorm Radek: hi kelly

[16:04] Kelly Linden: Hello

[16:05] Jane1 Bookmite: ahhh, things one knows after a hasty pick 3 years later

[16:05] xstorm Radek: hi Andrew and simon

[16:05] Andrew Linden: Ok first some news...

[16:06] xstorm Radek: im to ask about what will happen to all the old LSL scripts

[16:06] Andrew Linden: I'm still working on bug fixing in mesh, so nothing exciting to contribute to any announcements.

[16:06] Andrew Linden: xstorm, could you elaborate on that?

[16:06] Rex Cronon: mesh bugs can be so exciting:)

[16:07] Andrew Linden: what do you mean "old LSL scripts"?

[16:07] Andrew Linden: Nah, just don't bother using mesh and you won't hit any of the bugs ;-)

[16:07] xstorm Radek: some old LSL scripts will not convert to mono so will they convert to C# when its added

[16:07] Andrew Linden: ah yes

[16:08] Leonel Iceghost: whats the status of C#?

[16:08] Andrew Linden: are we planning on abandoning support for legacy LSL compiled scripts Kelly?

[16:09] Helena Lycia: Oh... Hiya Kelly, didn't see you there

[16:09] Kelly Linden: We are not

[16:09] xstorm Radek: oh good to know

[16:09] Kelly Linden: We don't have any plans right not to stop support for LSL.

[16:09] Kelly Linden: right now*

[16:09] Helena Lycia: When do we get C#?

[16:10] Andrew Linden: I suppose that in the distant future, if we were to make a survey and find that only 10% of all scripts were using legacy LSL-compiled bytecode we could start making plans.

[16:10] Kelly Linden: There is no ETA on C# right now.

[16:10] Kelly Linden: Yeah, that is a possibility, but that would be quite a ways off before we could even start thinking about it.

[16:10] Pauline Darkfury: Is it still a goal to eventually disable the ability to compile new non-Mono scripts? (seems to me like that would be a beneficial step once we're confident that Mono actually is at least as good in all cases)

[16:11] xstorm Radek: if we had a code or script base converter i think that be the next best step

[16:11] Kelly Linden: Those kinds of issues are very much to-be-determined, we don't have plans along those lines at this time.

[16:12] xstorm Radek: and yet there has been so much talk around using c#

[16:12] Leonel Iceghost: what happened with the project, have you found any big problems?

[16:13] Pauline Darkfury: it's not been talked about all that much recently, Xstorm

[16:13] Pauline Darkfury: yes, it was talked about as the next big thing a long time ago, before it got shelved

[16:14] Simon Linden: Reality seemed to have interrupted those plans

[16:14] Helena Lycia: Damn reality

[16:14] xstorm Radek: shelved like some one asking a linden to quit or be fired type of shelved ?

[16:14] xstorm Radek: or just on a back burner ?

[16:15] Kelly Linden: C# is on a back burner right now.

[16:15] xstorm Radek: ok thank you

[16:16] Liisa Runo: i understood that Babbage got the C# quite ready before it got shelved. He already ran it in SL. Anyone have estimate how ready he got it?

[16:16] Kelly Linden: It still has a bit to go.

[16:16] Pauline Darkfury: it never made it as far as beta, if memory serves

[16:17] Liisa Runo: bit = 8% ?

[16:17] Kelly Linden: There is work around the security aspects and the micro-thread injector as well as some further passes and revision on the general SL C# API

[16:17] Kelly Linden: I'm pretty sure I'm not confident enough to give a % complete number on a project not currently active.

[16:18] Kelly Linden: Correct - it never made it to public beta.

[16:18] Pauline Darkfury: yeah, I'd avoid giving a precise number there, tbh, Kelly, if I understand the status of it correctly

[16:18] Kelly Linden: :)

[16:18] Pauline Darkfury: solid proof of concept, a good way from production is my understanding

[16:18] Rex Cronon: micro-thread injector by user code?

[16:19] xstorm Radek: yes but there be holes in the http profile system and assets system and that has not stop any thing yet

[16:19] Kelly Linden: The uthreadinjector is what makes 10k-20k concurrent scripts possible at all.

[16:19] Kelly Linden: It needs to know about all possible opcodes, and there is a big difference in the opcode profile that LSL uses and the full C# language.

[16:19] Rex Cronon: nothing like having somebody inject rogue code in the the server code:)

[16:20] Kelly Linden: The uthreadinjector injects code into the user script, not the other way around.

[16:21] xstorm Radek: is it not better to make a call library for simple commands in C# on the user side and keep heavy scripts coomand only server side ?

[16:21] xstorm Radek: commandss

[16:21] Kelly Linden: I don't follow xstorm.

[16:22] xstorm Radek: think of it this way server side has the full command codes and user just needs the basic caller code for it

[16:22] xstorm Radek: that way it will be limited

[16:22] xstorm Radek: and safer

[16:22] Pauline Darkfury: C# was entirely server-side as far as execution goes. scripting that runs on the viewer is a different thing, could have some interesting possibilities, but not part of the C# scripting

[16:23] Kelly Linden: The idea for using a language like C# is to get the added functionality of C#. If we limit it to the opcodes used by LSL you don't get any of that benefit. It is entirely server side.

[16:23] xstorm Radek: yes but limit it more to start with

[16:23] Simon Linden: yeah, that's off on a totally different tangent, but I'd love to see a scriptable viewer

[16:23] Kelly Linden: It is quite possible that the first release, or at least the first beta, would only support a small subset of possible opcodes.

[16:23] Pauline Darkfury: And, if you're talking about some code running on the viewer, safer for who? I'm sure as hell not allowing random SL creators to run C# code on my hardware, without a lot of protection

[16:23] Rex Cronon: isn't there already a scriptable viewer available?

[16:24] Andrew Linden: What kind of viewer scripts would you write Simon?

[16:24] xstorm Radek: thats be great to have built in scripting for off line work

[16:24] Leonel Iceghost: wow has scripting in the viewer and it is pretty secure

[16:24] Leonel Iceghost: world of warcraft, that is

[16:24] Simon Linden: The ability to pack up everything (or maybe a subset) on a parcel, and restore it later

[16:24] xstorm Radek: eeeeeeeewwwwwwww not wow

[16:24] Leonel Iceghost: I would write real user interfaces Andrew

[16:25] Ashiri Sands: What would you do with scripting client side?

[16:25] Pauline Darkfury: Yes, client-side scripting can be done relatively safely, e.g. browser JavaScript, WoW's LUA

[16:25] Simon Linden: Custom interfaces for game-based regions

[16:25] Rex Cronon: i don't think the ll viewer has lua scripting:)

[16:25] Simon Linden: For security, it just needs a well-designed sandbox

[16:25] Leonel Iceghost: no more 300kbps to transmit some color params to prims

[16:25] xstorm Radek: that sounds geat simo

[16:25] xstorm Radek: great

[16:26] Pauline Darkfury: it's just non-trivial to implement that type of scripting with reasonable security

[16:27] Leonel Iceghost: Pauline, you just modifi prims, or draw custom things.. bug locally, nothing to be worried about really, perhaps some data about resolution to get

[16:27] Simon Linden: I used to work on a server project that allowed user-written javascript ... it's not trivial, but if you begin the design with it in mind, it's not that hard to do

[16:27] Pauline Darkfury: Modifiying prims is something to worry about. I don't want to give random client-side scripts that I didn't write myself the ability to modify any prim I own

[16:28] xstorm Radek: i can only type with one hand so this will take longer but i do see how great it be to have what simon says as long as they do not get god commands over database items such as inventory items or users

[16:28] Leonel Iceghost: modify prims in your huds.. Pauline

[16:28] Leonel Iceghost: no need to add extra functionality than current lsl

[16:29] Pauline Darkfury: Same issue, Leonel., that can only be for scripts I write myself, unless there's a strong security model to restrict it to which prims it can edit

[16:29] Leonel Iceghost: your scripts modify your huds if they want..

[16:29] xstorm Radek: lsl is lacking the real power that is needed for some in world programing

[16:29] Pauline Darkfury: It can't be used for products if the product's viewer-scripts can modify any HUD, as that would allow them to break a HUD they have no business touching

[16:30] Meeter: Timecheck : User Group is half over

[16:30] Leonel Iceghost: Pauline, not any hud, just the one they are made for

[16:30] Simon Linden: Right, it needs a careful sandbox design

[16:30] Pauline Darkfury: Right

[16:30] Andrew Linden: Someone sent me an IM today and said:

[16:30] Pauline Darkfury: That's what I'm saying about it being non-trivial to get the security model

[16:30] Andrew Linden: the biggest immersion killer, and the area where SL is lagging behind other MMOs the most, right now is not having seamless sim crossings, and not being able to traverse the seas

[16:30] xstorm Radek: thats why off line programing is a must

[16:31] Andrew Linden: He wanted to add that topic to the User Group agenda

[16:31] Liisa Runo: and i want all mainland continents connected together with water sims

[16:31] xstorm Radek: can it be called MMU ?

[16:31] xstorm Radek: lol

[16:32] Helena Lycia: I would love to be able to travel between sims on a vehicle without the risk of crashing

[16:32] Rex Cronon: if could make a sim 1024m each side, u would have less problems with the crossings:)

[16:32] Helena Lycia: Or for my idea for enhanced agent movememnt to be implemented

[16:32] Pauline Darkfury: and, region crossings, good topic. I _thought_ that LeTigre had introduced a horrible 50% failure rate with TPs and crossings to/from main yesterday, but it seemed to calm down later. I think now that it was unfortunate timing, in terms of a general problem hitting just after LeTigre rolled, but not 100% sure

[16:32] Andrew Linden: yes, larger region sizes would reduce region crossings, and would be a partial solution

[16:33] Pauline Darkfury: Larger regions would also be a nightmare for some existing scripts

[16:33] Rex Cronon: it migth mess some scripts as they would expect each sim to be 256

[16:33] xstorm Radek: your talking about diva mega sim and opensim thats a no no

[16:33] Andrew Linden: good to know that LeTigre might not be introducing TP failures

[16:33] Andrew Linden: what is on LeTigre? ... Mesh probably.

[16:33] xstorm Radek: lol

[16:33] Pauline Darkfury: Yes, it's mesh on there

[16:33] Odessa Ibanez: magnum of is mesh

[16:34] Helena Lycia: I thought Mesh had its own RC?

[16:34] Helena Lycia: Last time I checked my sim it sed "Mesh RC" where it used to say "Magnum"

[16:34] xstorm Radek: do not mix mesh and mesh

[16:34] Leonel Iceghost: bigger sims could be made for legaly lsl functions of 256 parts.. and put new functions

[16:34] Liisa Runo: magnum and tigre ahve mesh, yes

[16:34] Pauline Darkfury: Well, I'm still not 100% sure that it's not LeTigre. I was getting horrible issues going main->LeTigre->main, but they seemed to go away.

[16:34] Helena Lycia: Why not make a big sim that pretends to be 4 normal sims?

[16:35] Kelly Linden: Mesh is on RC Mesh, RC_Mesh, LeTigre and Magnum

[16:35] Andrew Linden: We've kicked that idea around Helena. The maint obstacle is rigidity of the simulator code design.

[16:35] Pauline Darkfury: What I'd suggest is rather than a 1024m sim, to have a say to tightly couple 16 256m sims. Keep them as individual regions, but set them up so that agent handoff between them was much closer to seamless

[16:35] Andrew Linden: It's just harder than it sounds because of assumptions made within the code.

[16:35] xstorm Radek: aditi was having lag problems do to a mega prim if you was there testing but sl sees fine

[16:36] Andrew Linden: We have to find all of those assumptions and fix them. And they are numerous.

[16:36] Pauline Darkfury: Yup. If you can stick with the 256x256 model, but add a product that allows a group of regions to work tightly together, that seems potentially easier to me

[16:36] Andrew Linden: I tend to agree with the resident who raised that issue. I think we need to fix the region crossing problem.

[16:37] Andrew Linden: I'd love to find the time to try to work on it.

[16:37] Andrew Linden: There is some code in the Mesh branch that might make it easier to tackle certain types of region crossing bugs

[16:37] Helena Lycia: Do it in the time spent between sims

[16:37] Andrew Linden: in particular the ones about crossing into parcels where you're not allowed

[16:38] Pauline Darkfury: Yes, getting crossings to be closer to seamless in general would be great. Other than the issue that hit me just after LeTigre rolled, I've personally been finding crossings much improved these days, with approx 125 scripts crossing with me

[16:39] Simon Linden: I've been working on those cross-region-into-parcels-that-should-be-blocked bugs ... that was most of yesterday andtoday for me

[16:39] Helena Lycia: Vehicles are still very dicey

[16:39] Andrew Linden: Recent improvements on region crossing with scripts is probably related to work Kelly has done.

[16:39] Liisa Runo: yes, crossing are better, im been traveling the grid with phys vehicle this whole day. and been unseated only 3 times, logged out once

[16:39] xstorm Radek: may i ask why are people using mass move on flying ships and boats that are over 32 prims then reporting theyy crashed that was never ment to move prims over 32 and now people move 100 prims or more and do jiras about crossing and crashing

[16:39] Pauline Darkfury: yes, high speed vehicle crossings are still an issue (slow speed seems ok, tbh)

[16:40] Andrew Linden: well, I think 100-prim objects should be able to cross region boundaries

[16:40] Pauline Darkfury: Total number of prims really shouldn't b e a big deal, Xstorm. Prim data is _tiny_

[16:40] Andrew Linden: so those are legit bugs

[16:40] Leonel Iceghost: Prim data is huge at least to send to the viewers

[16:40] Pauline Darkfury: Scripts are the bigger issue, far more complex than prims, to unload them and reload them on crossing

[16:41] Pauline Darkfury: No, prim data is tiny even when sent to viewers

[16:41] Pauline Darkfury: Texture data is large, but that's a different thing. A texture on a prim is 36 bytes plus a small amount of overhead, per face

[16:42] xstorm Radek: i think its sill many sims are prim heavy and your going to move 100 prims on some ones land by flying on it then report when it chashed ?

[16:42] xstorm Radek: with mesh you will no longer need more them 5 prims

[16:43] Pauline Darkfury: moving objects around even on a prim heavy sim should be extremely easy. You update 3x float32, then broadcast that number to all connected agents if it meets interest list criteria

[16:43] Simon Linden: The problem with a lot of prims is that you end up wiht a complex scene to draw ... just a ton of triangles for the graphics system to render

[16:43] Pauline Darkfury: Yup, the rendering is where it becomes very much non-trivial

[16:43] Andrew Linden: heh, if only it was about updating 3xfloat32 when moving an object...

[16:44] Pauline Darkfury: Well, yeah, check parcel perms, physics, etc

[16:44] Pauline Darkfury: But we're talking 100 prim things, so should be non-physics (just other physics prims that need to move)

[16:44] Andrew Linden: yup, and then there is some spatial sorting to maintain

[16:45] Andrew Linden: it all adds up, but it is definitely more than just 3xfloat32 updates

[16:45] Pauline Darkfury: Ok, I over-simplified ;)

[16:45] xstorm Radek: i can see it now aspeople abuse report people using mass move and prim heavy builds

[16:45] Leonel Iceghost: btw, if 20 agents move around it should transmit their key, their pos and vel, 10 times per second right? it is like 15kbps to send.. but with 20 agents moving my internet connection goes 200kbps... are you going to work on it? people with packet loss, lose a lot more important packets and see a lot less things the more bandwidth they use

[16:45] Andrew Linden: well... you're right in one sense... it shouldn't cost much more to move 100 linked prims than it does to move one unlinked prim

[16:45] Andrew Linden: within the region

[16:46] Pauline Darkfury: yup, that was the real point, Andrew :)

[16:46] Andrew Linden: oh yeah... there is the script resource pools to maintain too

[16:46] Andrew Linden: so add a bunch of scripts and its even more interesting

[16:47] xstorm Radek: your forgeting people using old pc's and having to rez that type of updated data and assets calls on every sim server

[16:47] Andrew Linden: also, the scripts are really packaged in an efficient way, so 10 scripts scattered across various linked prims is more expensive to move than one script on the root prim

[16:48] Andrew Linden: er... scripts *aren't really packaged in an efficient way*

[16:48] Pauline Darkfury: The data sent from server to client is still pretty small, even for 100s of prims, if you ignore textures

[16:48] Rex Cronon: there r linked objects that have at least on script per ptirm

[16:48] Leonel Iceghost: if 20 agents move, your bandwidth is 200kbps.. how is that small

[16:48] Rex Cronon: per prim*

[16:48] xstorm Radek: Pauline test it on aditi first

[16:48] Pauline Darkfury: If the 100s of prims were already in view, or had been seen recently, many of the textures will be cached (hopefully)

[16:49] Andrew Linden: Oh yeah... the physics engine maintains some spatial data structures too.

[16:49] Pauline Darkfury: 20 agents moving is 3x float32 for position, 3x float32 for rotation (or is it 4x?)?

[16:49] Andrew Linden: There is a cost involved in moving an object within those

[16:49] Pauline Darkfury: Per-agent, that is

[16:50] Andrew Linden: and the bigger the object, and the more pieces in the shape, then the more expensive it can be.

[16:50] Rex Cronon: btw simon. u should try Radegast. i heard it has local c# scripting:)

[16:50] Pauline Darkfury: and for the data that should need to be sent to the client, not all the other bits

[16:50] Andrew Linden: For example... moving a 100-prim static object into a dense pile of dynamic objects can be quite expensive...

[16:50] Leonel Iceghost: right Pauline, it shouldn't be more than 15kbps

[16:50] Pauline Darkfury: re-drawing the scene should be the heavy bit, not the bandwidth

[16:50] Andrew Linden: because the phys engine will suddenly have to create a bunch fo collision points for the new interpenetrations

[16:51] Simon Linden: I'm actually not a particular C# fan ... I had more previous scripting experience with Javascript and ACtionScript

[16:51] Arawn Spitteler: Would 15,000 ten meter objects make a laggier sim than 15,000 decimeter objects.

[16:51] Andrew Linden: that burst of collision points can burst the simulator's memory consumption...

[16:51] Andrew Linden: and that can cause problems

[16:51] Pauline Darkfury: yeah, if you trigger a huge amount of new physics activity, that's going to get ugly server-side

[16:52] Andrew Linden: for other processes on the same host as well

[16:52] Andrew Linden: I'm just geeking out on how expensive it *can* be to move 100 linked prims within the same region.

[16:52] Andrew Linden: Not always expensive, but the possibility is there

[16:53] xstorm Radek: yes

[16:53] Pauline Darkfury: yup, the basics shouldn't be costly, but the consequences may well be costly

[16:53] Simon Linden: That is so often like the problems we face ... things can be fine 95% of the time, but that 5% where it get complex can just kill the sim

[16:53] Andrew Linden: But I'm interested in how to improve region crossings. I can't work on it right now, but I've been pondering the problem today.

[16:54] Andrew Linden: I've been thinking of strategies that we've already kicked around User Group before

[16:54] Tortured Scientist: with mesh you can have a complex looking vehicle and have the controls in one linked cube. should not be so bad. except for the PE issues where parcels are concerned. (sat upon vehicles)

[16:54] Andrew Linden: and wondering if I could take a swing at any of them.

[16:54] xstorm Radek: i own my own server and been looking at grid calls on the database and its heavy even if its known uuid calls

[16:55] Meeter: Timecheck : User Group is almost over

[16:55] Liisa Runo: Lindens, look at the picture behind me, what you can say about this? that total frame time and the stuff below it dont add up

[16:55] Tortured Scientist: i guess people want a pirate ship with 24 guns all individually manned crossing regions

[16:55] Pauline Darkfury: You can't compare the operation of OpenSim in terms of DB load for particular activities to SL. Some stuff might be quite similar, but there could be huge differences

[16:55] Rex Cronon: and tanks:)

[16:56] xstorm Radek: yes i own better 5t servers by ibm

[16:56] Crash Meeroo whispers: Simon Linden will you please pick me up and snuggle me?

[16:56] Pauline Darkfury: OpenSim is only similar at the high level (overall behaviour characteristics). Under the hood, it's a completely different beast, any similarities are coincidental

[16:56] Andrew Linden: Liisa, I think that means there are some "chunks" of the simulator frame that are not properly wrapped in a timer.

[16:57] Simon Linden: Liisa - I'm not sure where that extra time went, but the viewer stats are not all-encompassing

[16:57] xstorm Radek: lol

[16:57] Liisa Runo: been having some strange things going on lately. Sometimes we also get most of the time as spare time, but scripts get slice only once in 10min

[16:57] Andrew Linden: But I have to agree... the times should add up, otherwise it can prevent proper diagnosis of the source of the lag.

[16:58] Simon Linden: We also sometimes see delays that are from other server proceses ... something else hogs the CPU, and the system scheduler is not giving out the CPU evenly

[16:59] Pauline Darkfury: If scripts are only getting a slice in terms of per-minute, that's very badly wrong

[16:59] Rex Cronon: the viewers must be connecting to another dimension. quite soon we might endup opening a portal to it, and let monsters in:)

[16:59] xstorm Radek: hhhmmmm i seen that problem befor when a new island was put in sl back in 2009 with a network timing problem

[16:59] xstorm Radek: but that was fix

[17:00] Meeter: Thank you for coming to the Server User Group

[17:00] Andrew Linden: Liisa, what region are those stats from?

[17:00] Pauline Darkfury: Thanks, Lindens

[17:00] Liisa Runo: sandbox goguen or sandbox cordova

[17:00] Helena Lycia: Just had someone ask me a really important set of questions

[17:00] Helena Lycia: Hi Hi Helena. What's the deal wif mesh? Can we eat it? Does it reduce prim count or increase it? Is it good with cookies???

[17:00] Helena Lycia: I couldn't answer the last one though

[17:00] xstorm Radek: omg not cordova

[17:00] Liisa Runo: goguen

[17:00] Pauline Darkfury: 2 topics we didn't get to, but friendly reminders for: encroachment-return (on mainland, plus cross-region); and PE-sudden-bloat return

[17:01] Helena Lycia: Awwwww... I thoguht we'd avoid mentioning PE today

[17:01] Ashiri Sands: lol

[17:01] xstorm Radek: lol

[17:01] Pauline Darkfury: we did, in the meeting ;)

[17:01] Andrew Linden: Pauline, encroachment return would have to be enabled on the mainland once Mesh is out. But yeah, that is something I need to follow up on.

[17:01] Pauline Darkfury: I just tacked it on once the meeting had cloased

[17:01] Andrew Linden: I'll try to have an update next week.

[17:01] Pauline Darkfury: Thanks :)

[17:01] Tortured Scientist: yea i'm wondering how PE bloat is gonna work in rental groups. Will only the mesh get booted or will their be a prim massacre?

[17:02] Andrew Linden: As to sudden PE bloat causing returns... that is specifically what I'm working on these days.

[17:02] Andrew Linden: I've got some fixes for some of the modes that would trigger return

[17:02] Pauline Darkfury: cool

[17:02] Andrew Linden: but those will have to go into a mesh update. Not sure when they'll be added.

[17:02] xstorm Radek: next up stock crash and usa money going down the tubes

[17:03] Helena Lycia: Please keep the PE calculation simple Andrew. So far, when I try to explain it to friends their heads explode and it's getting really messy

[17:03] Andrew Linden: I've got most of the sudden bloat modes fit with pre-checks and they fail if they would overload the parcel

[17:03] xstorm Radek: lol

[17:03] Arawn Spitteler: U$ going south will be a problem, if the servers are in Holland

[17:03] Pauline Darkfury: We need the return fix as a priority, Andrew. It has the potential for RL financial loss for people

[17:03] Rex Cronon: exploding heads? now that is funny picture. lol

[17:03] Andrew Linden: and also a change coming into the system will be to return the object that pushed the parcel over, rather than random stuff belonging to other people

[17:04] Pauline Darkfury: Yeah, by all means drop-kick the bloating object out of the sim, that's just fine

[17:04] Arawn Spitteler: But the thing coming in, will be a vehicle

[17:04] Ashiri Sands: just mentioning anything vaguely technical can result in cerebral explosions

[17:04] Andrew Linden: that is, identify the object that caused the excess and return it, even if it is owned by the parcel owner and there are things there owned by others

[17:04] Andrew Linden: however, a trick will be to prevent grief modes

[17:04] Tortured Scientist: thanks. don't need the PE griefer drama all over the place lol

[17:04] Pauline Darkfury: Yup, that's my other concern with it

[17:05] Andrew Linden: where a griefer would be able to overfill the parcel and then let a regular object change and get itself returned.

[17:05] Andrew Linden: Hence the pre-change checks for enough parcel resources.

[17:05] Pauline Darkfury: Some parcel owners have a bunch of "owned by others", but rez & entry disabled. If a griefer can sneak something in and then bloat it, big trouble

[17:05] xstorm Radek: hhhmmm evil thinking some evil hacker will find a way to get region move commands from linden admin and move evey ones sims now thats scary to think about

[17:06] Andrew Linden: yes Pauline. I fixed a couple modes by which griefers could get objects into a parcel where they are banned.

[17:06] xstorm Radek: thats good

[17:06] Ashiri Sands: Saves the Meeroos too

[17:06] xstorm Radek: :)

[17:07] xstorm Radek: yes

[17:07] Andrew Linden: BTW, Prokofy Neva has reported problems where griefers have been able to flood a parcel and somehow get the simulator to return the "legit" content first, rather than the other stuff owned by the griefers.

[17:07] xstorm Radek: save meeroos

[17:07] Pauline Darkfury: The RL financial loss is breedables that (stupidly) die on return, or if it nukes someone's store and they are not online for a few days

[17:07] Helena Lycia: I still worry about the complexity of the calculation and all the ways the PE on an object can change

[17:07] Andrew Linden: I think maybe I found the "race condition" that was causing that bug.

[17:07] Rex Cronon: that was before mesh got here

[17:07] Pauline Darkfury: Yeah, Prok is a bit of a target for stuff like that

[17:07] Arawn Spitteler: So, I drive a megaprim vehicle over a parcel, and toggle it static-bloated

[17:08] Andrew Linden: And I expect it to be better in the mesh code. But I'm going to have to try to repro some grief attacks if I can find the time.

[17:08] Ashiri Sands: only a bit of a target?

[17:08] Pauline Darkfury: heh

[17:08] xstorm Radek: well my wife yelling for me to get rest

[17:08] Pauline Darkfury: I do feel a little sorry for him at times, tbh

[17:08] xstorm Radek: **ok i wuv u byby!!**

[17:08] xstorm Radek: every one

[17:08] Rex Cronon: tc xstorm

[17:08] Andrew Linden: Alright. I've got to go too.

[17:08] Andrew Linden: Thanks for coming everyone.

[17:08] Rex Cronon: tc andrew

[17:08] Pauline Darkfury: People going in and killing his chickens, or whatever is just evil

[17:08] Pauline Darkfury: Thanks, Andrew

[17:08] Liisa Runo: thanks everyone

[17:08] Simon Linden: Me too ... thanks everyone for the good discussion today

[17:09] Pauline Darkfury: Take care, have a good weekend, all :)

[17:09] Rex Cronon: tc simon

[17:09] Rex Cronon: tc all

[17:09] Andrew Linden: BTW, I've been publishg the User Groups transcripts in a timely manner the past week, so you can find them on the wiki.

[17:09] Andrew Linden: I still need to back-publish a bunch.

[17:09] Leonel Iceghost: I saw, thanks for that Andrew

[17:09] Andrew Linden: But I'm making progress.

[17:09] Rex Cronon: keep up the good work:)



Simulator_User_Group

Prev 2011.08.02 Next 2011.08.09