Simulator User Group/Transcripts/2011.08.26

From Second Life Wiki
< Simulator User Group
Revision as of 16:51, 26 August 2011 by Andrew Linden (Talk | contribs)

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


Prev 2011.08.23 Next 2011.08.30

List of Speakers

Andrew Linden Arawn Spitteler Eddi Decosta
Elisha Richez Flip Idlemind Jonathan Yap
Kadah Coba Kaluura Boa Kelly Linden
Latif Khalifa Leonel Iceghost Liisa Runo
Lomgren Smalls lonetorus Habilis Mastorian Kingsford
Memorial Dae Morgaine Dinova Sahkolihaa Contepomi
Simon Linden Vincent Nacon


[16:02] Sahkolihaa Contepomi: Hey Andrew.

[16:02] Andrew Linden: Hello

[16:02] Morgaine Dinova: Hi Andrew, Simon :-)

[16:03] Andrew Linden: Ok time for the news...

[16:03] Memorial Dae: Ha! Nice Av Kel!

[16:03] Morgaine Dinova: Bad new first :P

[16:03] Memorial Dae: Ok

[16:03] Andrew Linden: Last Tuesay I mentioned that I had a "faster scripts" project that was up on aditi

[16:03] Andrew Linden: and I asked people to test it out

[16:04] Andrew Linden: in particular, to let me know if their scripts are any faster there

[16:04] Vincent Nacon: sorry I didn't get the memo

[16:04] Andrew Linden: and also I challenged people to try to crash it and offered a reward

[16:04] Vincent Nacon: bad iea?

[16:04] Vincent Nacon: idea*

[16:04] Andrew Linden: whenever someone found a new way to crash with scripts

[16:04] lonetorus Habilis: the awsome sum of 20k

[16:04] Andrew Linden: actually it worked out well

[16:05] Andrew Linden: Flip Idlemind found two distinct crash modes

[16:05] Andrew Linden: so he got two rewards (yes, of L$20k)

[16:05] Sahkolihaa Contepomi: Heh...

[16:05] Vincent Nacon: knew he was such a griefer muhaha!

[16:05] Simon Linden: This weeks refreshments are on Flip!

[16:05] Latif Khalifa: lol

[16:05] Sahkolihaa Contepomi: Haha

[16:05] Memorial Dae: Sweet

[16:06] Andrew Linden: the best thing was that his first crash find was a particular flaw pattern in the code

[16:06] Morgaine Dinova: Gratz Flip :P

[16:06] Andrew Linden: which helped me audit the code and find other places where I had messed up

[16:06] Andrew Linden: the reason the flaws were introduced was that I was trying to work fast and convert a rather big chunk of code quickly

[16:07] Morgaine Dinova: Didn't unit tests catch it?

[16:07] Andrew Linden: so I wrote some vim "recordings" (keystroke macros) to help me out and used them to speed through the code

[16:07] Andrew Linden: What unit tests? ;-)

[16:07] Morgaine Dinova: Ouch

[16:07] Andrew Linden: Anyway, I've updated the deploy on aditi, and so it is open for playing with if anyone is interested.

[16:08] Andrew Linden: I never did get any feedback on whether anyone's benchmarks were running faster

[16:08] Andrew Linden: but the few benchmarks I ran myself showed improvements

[16:08] Vincent Nacon: before I'm confused, this is about Mono2?

[16:08] Lomgren Smalls: I don't have any benchmarks, but I'm sure there are quite a few out there. And is this just Mono improvements, or Mono and LSL both?

[16:09] Andrew Linden: ranging from 99% to 60% of the old time

[16:09] Latif Khalifa: the main problem with script performance is the scheduler inefficiency, especially on class 7 hardware

Total Avatar Scripts - 506

Highest - Morgaine Dinova: 91 (18%)

Lowest - Vincent Nacon: 0 (0%)

[16:10] Latif Khalifa: if a sim has a lot of scripts, it will spend more time deciding what to run, than actually executing scripts

[16:10] Andrew Linden: No, this isn't Mono specific, however it appears to affect Mono the most for some benchmarks because Mono is already much faster than legacy LSL2.

[16:10] Latif Khalifa: but that's the baffling case of some hardware quirk

[16:11] Andrew Linden: Both Mono and LSL2 execute some callbacks when doing something like llGetPos()

[16:11] Andrew Linden: My work was to try to speed up llGetPos() itself, not the bytecode execution.

[16:11] Latif Khalifa: So you were optimizing the api implementation in the sim?

[16:11] Andrew Linden: Yes.

[16:11] Latif Khalifa: cool

[16:11] Leonel Iceghost: 99% means twise?

[16:11] Leonel Iceghost: as fast

[16:11] Vincent Nacon: no

[16:11] Vincent Nacon: 200% would be 2x faster

[16:12] Andrew Linden: No, 99% means it took 99/100th to get done what the old code took 100/100 to get done.

[16:12] Leonel Iceghost: right.. so 99% slower :P

[16:12] Leonel Iceghost: oh got it

[16:12] Andrew Linden: 1% less time, I guess.

[16:12] Vincent Nacon: maybe he meant 199%

[16:12] Vincent Nacon: or... I dunno

[16:12] Vincent Nacon: and any good news?

[16:12] Andrew Linden: All of the benchmarks I ran were measuring the time to get some fixed amount of work done.

[16:13] Leonel Iceghost: so it is 1% to 40% faster

[16:13] Andrew Linden: So improvement translated into shorter times.

[16:13] Andrew Linden: A benchmark that was just trying to get as much done within some fixed amount of time would have improvements translate as higher scores.

[16:14] Simon Linden: The high end of that range is impressive ... was that any particular kind of script?

[16:14] Andrew Linden: Yeah, the best time I had was just a simple loop that called three different simple calls:

[16:14] Andrew Linden: llGetPos(), llGetMass(), and llGetRot() or something like that

[16:15] Andrew Linden: Calls such as llSetPos() have a built-in delay afterward

[16:15] Simon Linden: That makes sense, it was running a lot of bytecode and thus hitting your faster version a lot

[16:15] Andrew Linden: which means they'll run faster, and relenquish priority to the next script faster

[16:15] Andrew Linden: but they won't necessarily loop faster

[16:15] Liisa Runo: will the improvements also affect to the llGetPrim.. ways of doing those things?

[16:15] Liisa Runo: llGetLinkPrim*

[16:15] Andrew Linden: Yes Liisa.

[16:15] Liisa Runo: good

[16:16] Andrew Linden: Anyway, the experiment is up on a very small channel of aditi: bispinor

[16:16] Andrew Linden: The two regions are: "Sandbox - Weapons Testing" and "Sandbox - Bispoinor"

[16:16] Flip Idlemind: Any significance to the name?

[16:17] Vincent Nacon: could I be added to estate's management list for top scripts?

[16:17] Andrew Linden: If anyone tests some benchmarks there then I'd love to get an email with the results.

[16:17] Andrew Linden: Vincent, the estate is mainland. I'll have to modify the estate of those regions to add you.

[16:17] Vincent Nacon: oh yeah, n/m

[16:17] Andrew Linden: I'll look into that after this meeting. If I figure out how to change the estate before I have to leave then I'll send you an IM.

[16:18] Vincent Nacon: long as it doesn't take up too much of your time

[16:19] Andrew Linden: Other than that, I've been trying to wrap up some bug fixes around parcel access permissions that were introduced by the mesh code

[16:19] Andrew Linden: We've got a maint-server for mesh pretty much done but have to get it up for test.

[16:20] Andrew Linden: It will be known as DRTSIM-85 on aditi when it gets there.

[16:20] Andrew Linden: I've plugged some access holes for accounts that are specifically banned from parcels there.

[16:20] Andrew Linden: But there is a bit more work to do before it is truly tight.

[16:21] Vincent Nacon: "Dragon Runs This SIM"?

[16:21] Vincent Nacon: heheh

[16:21] Flip Idlemind: Deploy Request Tracker

[16:21] Andrew Linden: Maybe when I think I've got it all sealed up I'll pose another challenge and offer rewards for anyone who can find a hole in my defenses.

[16:21] Andrew Linden: But I know I'm not ready for that yet.

[16:22] Andrew Linden: That's all the news I've got.

[16:22] Liisa Runo: SCR-176

[16:22] JIRA-helper:

[#SCR-176] llAvatarOnLinkSitTarget is not working properly

[16:23] Vincent Nacon: I don't think that's up yet...

[16:23] Andrew Linden: is llAvatarOnLinkSitTarget() a new function?

[16:23] Vincent Nacon: haven't heard about it

[16:23] Flip Idlemind: It was in the "Scripting Awesomeness" project

[16:23] Kelly Linden: Yea, couple weeks ago

[16:23] Liisa Runo: it is new and i tested it when it was still only in RL, and it worked, but now it dont work anymore

[16:24] Morgaine Dinova: Can only estate managers determine how many scripts a person runs, and their CPU usage? We can't obtain that figure for ourselves?

[16:24] Arawn Spitteler: I think that would be an SEC issue

[16:24] Kelly Linden: I'll take a look at 176

[16:24] Sahkolihaa Contepomi: Morgaine - scripts can read scripts people are wearing, their script time and memory usage. :o

[16:24] Sahkolihaa Contepomi: My radar does all that.

[16:24] Kelly Linden: You can get it via LSL.

[16:24] Liisa Runo: im not sure if i have just lost too many marbles or if LL changed the code after the RC and the grid wide release

[16:25] Vincent Nacon: It can't read me cause I'm awesome.

[16:25] Morgaine Dinova: Cool Sahk. Got an ll function name I can look up?

[16:25] Lomgren Smalls: You can tell if it is an LSL (16k) or Mono (64k) script as well

[16:25] Flip Idlemind: llGetObjectDetails

[16:25] Vincent Nacon: Liisa, did it let you compile?

[16:25] Morgaine Dinova: Thanks Flip :-)

[16:25] Flip Idlemind: can be used to get script count and time

[16:26] Kelly Linden: llGetObjectDetails

[16:26] Object: Hello, Avatar!

[16:26] Simon Linden: Morgaine ... the scanner on the fence gets that info, or do you mean something else?

[16:26] Liisa Runo: yes, it compiles

[16:26] Lomgren Smalls: There's an object that uses count and memory

[16:26] Vincent Nacon: regardless if viewer is up to date with that code

[16:26] Morgaine Dinova: Oooh, thanks Lom!

[16:26] Lomgren Smalls: should be full perm

[16:26] Lomgren Smalls: I haven't seen it be able to get time, I need to look that up

[16:26] Kelly Linden: Also this HUD gets it as well:

[16:27] Morgaine Dinova: Thanks Simon, I'll poke around the scanner

[16:27] Simon Linden: I'm not sure what the perms are on that, I didn't write it

[16:27] Liisa Runo: Simon, i hope you didnt give her my toy? my toys are not be passed onward

[16:27] Vincent Nacon: it doesn't support alpha channel, doesn't it?

[16:28] Simon Linden: Liisa - no, I haven't changed perms or handed it out

[16:28] Liisa Runo: okie, good

[16:28] Morgaine Dinova: Thanks for keeping you code secret, Liisa. Good thing that you could read other people's code when you were learning, huh?

[16:28] Kadah Coba: Whats going on with SH-2255?

[16:29] Sahkolihaa Contepomi:

[16:29] Sahkolihaa Contepomi: And yes, I'd like to know this one.

[16:29] Andrew Linden: Dunno yet myself. I haven't looked into that yet.

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

[16:30] Andrew Linden: I think I heard someone talking about it in some triage meeting.

[16:30] Kadah Coba: Gawd help us is someone figures out a repro for that and is able to drop a regions Fps to 1 at will for as long as they want.

[16:30] Simon Linden: Are those full regions or homesteads?

[16:30] Leonel Iceghost: full regions

[16:30] Sahkolihaa Contepomi: Full.

[16:30] Kadah Coba: Full private regions

[16:30] Sahkolihaa Contepomi: The one region that was affected didn't have many scripts in it at all, and was cut off from the public.

[16:31] Andrew Linden: The parcel permissions problems and "mode rights" based on parcel settings are known bugs, I've fixed them in the aforementioned maint-mesh-server probject

[16:31] Kadah Coba: Randomly TD will drop to >0.1 and FPS to around 1.

[16:31] Kadah Coba: And on the region we were having it happen to, it was like that for over an hour.

[16:31] Andrew Linden: It appears SH-2255 has become a dumping ground for all sorts of unrelated bugs introduced by mesh.

[16:31] Andrew Linden: I think we have a few duplicate general perupose bug entries too.

[16:31] Andrew Linden: We

[16:32] Flip Idlemind: Did you look at the stats bar to see what was taking up the most time? (Scripts, Physics, etc)

[16:32] Kadah Coba: Its very hard to us to tell the different performance issues since mesh appart.

[16:32] Sahkolihaa Contepomi: Physics always dropped to 0.0

[16:32] Andrew Linden: We'll have to unravel those eventually -- pieces will be fixed shortly and we'll have to track down all the others mentioned in comments.

[16:32] Leonel Iceghost: Andrew, do you have those graphs showing performance since meshes comparing with before?

[16:33] Andrew Linden: No, I don't have any of those graphs handy.

[16:33] Andrew Linden: I'll have to hunt through our dashboards for them.

[16:33] Liisa Runo: (i have not read too much of other peole's code, that is why my toys are not identical to other ppls toys. i do share some code i write, but not all the best things i have spent months to write, enough many people have already sold things created by me as their own, some people are so much in a hurry with their copy paste that their overpriced products still carry my "Liisa's example script" in them)

[16:33] Flip Idlemind: I've seen regions with "Spare Time" through the roof, that also had "Pump IO" really high (I think Pump IO is actually part of "Spare time" but I'm not sure)

[16:33] Leonel Iceghost: I would bet it will be lower, physics doesn't seem very nice when many things are happening

[16:33] Andrew Linden: Or maybe Alex Linden has a graph handy. I try to remember to ask him next week.

[16:33] Flip Idlemind: But Viewer 2 no longer shows "Pump IO" as far as I can tell

[16:33] Kelly Linden: sometimes pumpio is part of spare time

[16:33] Simon Linden: Flip - it is _sometimes_

[16:34] Simon Linden: I don't know why those stats were taken out of the viewer ... we should put 'em back in sometime

[16:34] Simon Linden: The server is still sending the data, afaik

[16:35] Flip Idlemind: "Spare time" is actually missing from Viewer 2's stats bar (at least, last time I checked) but I put it back in. All that needed to be added was some XML-y stuff

[16:36] Simon Linden: it looks like it never made it into the main branch ... I don't see it in my V3

[16:36] Leonel Iceghost: spare time also shows 0 when many scripts are in the sim, even if the sim is perfect.. it never worked very well

[16:37] Simon Linden: That's actually how it is supposed to work

[16:37] Leonel Iceghost: sometimes you get 20 script time and you swim in lag, and sometimes you get 20 script time and everything works fine

[16:37] Simon Linden: the spare time is extra time at the end of the simulator frame after everything has been run

[16:37] Andrew Linden: same scripts Leonel?

[16:37] Leonel Iceghost: no

[16:37] Simon Linden: So if it runs scripts up to the end, spare time will be 0

[16:37] Leonel Iceghost: light vs doing things scripts

[16:38] Andrew Linden: yup

[16:38] Kadah Coba: Is spare time derived by the viewer or is it sent by the sim?

[16:38] Andrew Linden: time spent running scripts usually runs for a minimum amount of time, but up to whatever time is still available to fill up a 45fps simulator mainloop

[16:39] Simon Linden: It's from the sim, Kadah

[16:39] Latif Khalifa: if you pass certain number of scripts all the spare time is going to be used by the script scheduler

[16:40] Simon Linden: It's more of a 'headroom' ... it's extra time you could use with more scripts or physics and still keep at 45 fps

[16:41] Simon Linden: During that spare time, the region will briefly sleep, then check for incoming messages. Sometimes those messages (in pumpio) can take a long time, so you'll get a sudden jump in both values

[16:41] Flip Idlemind: Is there anything you can think of that would make a region have huge Pump IO time (And could it be related to the drop in performance since mesh came out. Idk...this is all wild speculation)

[16:41] Eddi Decosta: zzzzzzzzsorry Simon

[16:41] Eddi Decosta: sorry*

[16:41] Eddi Decosta: said*

[16:41] Simon Linden: no problem :)

[16:41] Vincent Nacon: and don't go thinking about stacking more sims on a blade server

[16:41] Vincent Nacon: muhaha!

[16:42] Eddi Decosta: loll hi there

[16:42] Simon Linden: heck, I wish we had one region per server :)

[16:42] Leonel Iceghost: I wish we had all regions in a cloud of servers :P

[16:42] Vincent Nacon: 4 was good but 8 is overkill

[16:42] Eddi Decosta: e, sorry my interupt

[16:42] Memorial Dae: Was good seeing you all, have a great weekend!

[16:43] Andrew Linden: hrm... pumpio is time spent handling http data, and transmitting objects/avatars across region boundaries, right Simon/Kelly?

[16:43] Liisa Runo: (Kelly are you making particles to work in HUD's?)

[16:43] Eddi Decosta: well, have a question about linkability ruls, why its changed just to link som prim in a prims of 64 meter?

[16:43] Kelly Linden: I'm just playing with forms and html-on-prim and particles

[16:44] Andrew Linden: Eddi, did linkability rules change recently?

[16:44] Kelly Linden: Andrew: Some more, potentially. PumpIO wraps some of our tcp handling.

[16:44] Andrew Linden: Or are you talking about larger prim sizes?

[16:44] Vincent Nacon: I used to remember particle having render glow on beta once.... I wonder what happened to it?

[16:44] Eddi Decosta: i do a prims of 64 meter and i can link some thing on ..

[16:44] Andrew Linden: What do we send over TCP?

[16:45] Eddi Decosta: but if i do two all the linkability is the same

[16:45] Simon Linden: All sorts of external web service requests

[16:45] Vincent Nacon: render glow + particle = nice flamethrower

[16:45] Eddi Decosta: wall*

[16:45] Kadah Coba: Linkabality rules were simplified a while ago, but I thought that was last year.

[16:45] Andrew Linden: Eddi, the linkability rules changed a while ago... 8 months? A year? Two years? I forget.

[16:45] Kelly Linden: Quite a bit of intersim messaging, including all that you mentioned.

[16:45] Liisa Runo: about an year

[16:46] Andrew Linden: But the result was that the linkability now depends only on the centers of the prims involved

[16:46] Andrew Linden: not their scales

[16:46] Eddi Decosta: Andrew i dont know but it ned to be changed i think ..

[16:46] Andrew Linden: so a 64m prim could link to any other prim, as long as that prim's center is within 54 meters of the other

[16:46] Latif Khalifa: how about relaxing 10m limit on llRez. seems kind of limiting with 64m prims

[16:46] Lomgren Smalls: I've been tempted to play around and see what can be linked (before the 64 meter prims)

[16:47] Andrew Linden: at the moment the "max linkability bounding sphere diameter" is 54 meters, but I think we'll eventually bump that up to a nice round number like 64

[16:47] Kelly Linden: I actually need to run a bit early today. Check out my html hud if you have time and a viewer that supports it. Updated since yesterday:

[16:47] Liisa Runo: yes, the distance limit for Rez is not working, there are tricks around it, and the limit dont serve any purpose

[16:47] Vincent Nacon: wait.. is that a redefined linkset limit now? 54m range?

[16:47] Sahkolihaa Contepomi: See you, kelly.

[16:47] Vincent Nacon: not percentage?

[16:47] Latif Khalifa: thanks Kelly, have a nice weekend :)

[16:47] Vincent Nacon: take care Kelly

[16:47] Andrew Linden: yes Vincent. The linkability rules were simplified a while ago -- no longer a function of prim scales

[16:48] Simon Linden: We can't raise that limit until there are other fixes in with rez permissions first

[16:48] Vincent Nacon: right but I meant it's not using percentage?

[16:48] Eddi Decosta: sorry the viewer just close like that without warning

[16:48] Flip Idlemind: They just have to be less than 54 meters away to be linked

[16:48] Andrew Linden: I don't understand what you mean by "percentage".

[16:48] Kadah Coba: The complicated link rules were slowing down llSetLinkPrimPramsFast

[16:49] Leonel Iceghost: to me too Eddi

[16:49] Andrew Linden: Oh yeah, that's why we did it Kadah. I had forgotten!

[16:49] Vincent Nacon: ahh no worry, the way I remembered the linkset rules might be different in point of view

[16:49] Flip Idlemind: Because they had to re-check the rules every time

[16:49] Eddi Decosta: Leonel , you have kisten s21?

[16:49] Morgaine Dinova: Is the linkability algorithm specified somewhere?

[16:49] Leonel Iceghost: phoenix

[16:49] Eddi Decosta: kirsten*

[16:49] Eddi Decosta: ah okey

[16:50] Eddi Decosta: ummm, i have the latest kirsten

[16:50] Andrew Linden: I think so Morgain. But it is pretty simple these days.

[16:50] Andrew Linden: I can find it...

[16:50] Morgaine Dinova: Cool!

[16:50] Andrew Linden:

[16:50] Eddi Decosta: well, Andrew about the linkability , i think it could be very nice to change this ..

[16:51] Andrew Linden: Eddi, what do you want changed?

[16:51] Arawn Spitteler: So, have we an llAvatarOnLinkSitTarget?

[16:51] Kaluura Boa: 64m instead of 54 would be a first step...

[16:51] Eddi Decosta: cause with the prims of 64 meter its more simple if you do < 64 , in remplacement of < 54

[16:52] Leonel Iceghost: it would be nice to stop all linkability, rez, llSetPos limits.. after all there are workarounds for everything...

[16:52] Morgaine Dinova: What yields the magic number "54"? (256 is obvious)

[16:52] Andrew Linden: Oh yeah. I expect we'll eventually bump it up to 64.

[16:52] Vincent Nacon: I'm surprised it was 42

[16:52] Leonel Iceghost: why not unlimited?

[16:52] Vincent Nacon: it wasn't*

[16:52] Andrew Linden: 54 was the minimum sphere that still contained all of the legacy linkable content.

[16:52] Leonel Iceghost: or 256

[16:53] Andrew Linden: I wanted to keep it as small as possible until we had encroachment return and collision across region boundaries

[16:53] Eddi Decosta: oh, and miew for Morgaine

[16:53] Arawn Spitteler: Didn't include my legacy content

[16:53] Morgaine Dinova: Hey Eddi :-)

[16:53] Vincent Nacon: <-- way past bed time, don't mind my typo and out of loop moment

[16:53] Andrew Linden: Meanwhile... the linkable limit is now a config setting rather than hardcoded into the C++

[16:53] Andrew Linden: so we can change it pretty easily when we're ready

[16:53] Eddi Decosta: Andrew do you know the date of previous change at linkability?

[16:54] Andrew Linden: the wiki page says "Dec 2010"

[16:54] Liisa Runo: good limit for link distance would be llVecDist(ZERO_VECTOR,<256,256,4096>)

[16:54] Vincent Nacon: what make it you guy be ready for it though?

[16:54] Eddi Decosta: the new* change*

[16:54] Eddi Decosta: sorry

[16:54] Morgaine Dinova: Is that page editable so that we can add "54 == minimum sphere that still contained all of the legacy linkable content"?

[16:54] Andrew Linden: We'd have to enable encroachment return, and maybe implement collisions across region boundaries.

[16:55] Andrew Linden: It is a wiki. I think it is editable.

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

[16:55] Andrew Linden: You probably just have to login

[16:55] Vincent Nacon: yeah

[16:55] Eddi Decosta: oh and Andrew, i have updated some thing here,

[16:55] JIRA-helper: [#CTS-743] Mesh vehicles axes float / axes friction

[16:55] Eddi Decosta: in other subject

[16:56] Vincent Nacon: so any thoughts about porting LSL functions for density, friction, rep. ?

[16:56] Liisa Runo: and since it is friday, ill throw a lil dream on the table: would be cool to have "science sandboxes" where we would not have all kinds of limits all over. We could play with particle accelerators and otehr fun toys without limits on stuff like velocity

[16:56] Vincent Nacon: and gravity of course

[16:57] Eddi Decosta: i see that but liisa? but the scipt of movement need to play with the gravity parameter to?

[16:57] Vincent Nacon: for prim,sorry

[16:57] Elisha Richez: for science!

[16:57] Andrew Linden: Liisa, sounds like a good application of OpenSim, where you can hack the phys engine to support special simulations

[16:57] Flip Idlemind: So, SCR-162. Next week maybe? (/me crosses fingers)

[16:57] JIRA-helper:

[#SCR-162] Bounds error when calling PRIM_LINK_TARGET in a child prim

[16:57] Andrew Linden: such as the N-body gravitational simulation that was done once

[16:57] Leonel Iceghost: Liisa, that would be a good exercice to lindens, to improve physics too

[16:58] Leonel Iceghost: to check everything is right with changed params

[16:58] Eddi Decosta: hmmm, interesting

[16:58] Morgaine Dinova: I added a Note to , quoting you Andrew. Hope it's OK

[16:58] Andrew Linden: Yup, looks good.

[16:58] Eddi Decosta: thanks morgaine

[16:59] Andrew Linden: A better science simulation would be nice.

[16:59] Eddi Decosta: well, the thing need to do 53.999 , cause when is 54.000 just i cant link

[16:59] Andrew Linden: it would be cool to be able to do magnetism, E-field, and other things

[16:59] Vincent Nacon: sounds like you might need to split another office meeting for physic department itself some day

[16:59] Morgaine Dinova: Oooh, yes Andrew!

[16:59] Leonel Iceghost: by doing that you ensure nothing is hardcoded and waiting for bugs

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

[17:00] Liisa Runo: thanks everybody ^^

[17:00] Andrew Linden: However really, it would probably be best to do it as an OpenSim mod for a niche market for virtual physics lags

[17:00] Andrew Linden: labs, even

[17:00] Andrew Linden: Hrm... maybe there is a niche market there for the OpenSim folks.

[17:00] Morgaine Dinova: Andrew: air with pseudo-real dynamics, so that one can fly properly after turning that option on.

[17:00] Arawn Spitteler: Density might be rendered knowable

[17:00] Andrew Linden: sure, that too

[17:01] Arawn Spitteler: Bernouli simulation?

[17:01] Andrew Linden: Although... fluid dynamics can really eat the CPU cycles

[17:01] Vincent Nacon: yeah, it's a wishful dream

[17:01] Andrew Linden: you'd need a devoted machine with lots of computing power

[17:01] Simon Linden: Falcon was discussing gravity settings yesterday at the beta user group ... I don't know if they might be useful

[17:01] Arawn Spitteler: One sim to a supercomputer sounds about right

[17:01] Sahkolihaa Contepomi: Will a proper physics engine ever be added into the viewer to replace the hacky flexi?

[17:01] Vincent Nacon: just have to make a basic version of it

[17:01] Andrew Linden: Flexi is a viewer-side feature.

[17:02] Andrew Linden: Flexi won't collide right until the viewer is doing its own physics simulations.

[17:02] Morgaine Dinova: Not true scientific simulation of the physics, that would be WAY too CPU intensive. But some minimal cheats can be done cheaply, such as air-like forces to affect flying, giving thrust and drag. Cheats are fine. :P

[17:03] Andrew Linden: yes, more interesting "game physics" (with cheats and approximations galore) might be interesting

[17:03] Arawn Spitteler: Scriptable lift?

[17:03] Vincent Nacon: but you don't really need Lindens to create Bernouli simulation physic when you can just make basic function on your own

[17:03] Andrew Linden: A long time ago I wanted to add an "airplane vehicle simulation" that used lift/stallspeed/etc

[17:03] Andrew Linden: but I never was able to knock that out as a demo

[17:03] Vincent Nacon: but of course.... we'd need those LSL function to change Gravity, friction, density, rep. to prim

[17:04] Vincent Nacon: you could simulate airlift with gravity

[17:04] Morgaine Dinova: Well Rod did come from EA ... he must be interested in better support for game development.

[17:04] Liisa Runo: LL could just take the easy path, mark couple limits in the server code, and let automated program delete those limits after every server update, and have that server version run in small selection of sandboxes for the ppl who are not afraid of getting pushed with highvelocity experiments

[17:05] Vincent Nacon: with relative velocity as requirement for airlift

[17:05] Simon Linden: Thanks everyone for coming, see you next week

[17:05] Sahkolihaa Contepomi: See you, Simon.

[17:05] Vincent Nacon: thanks Lindens

[17:06] Kaluura Boa: Good night...

[17:06] Vincent Nacon: see you on Tuesday, hopefully

[17:06] Eddi Decosta: thanks andrew to awnser me

[17:06] Liisa Runo: ninie

[17:06] Vincent Nacon: now I'm off to bed X_x

[17:06] Morgaine Dinova: No need to actually do fluid dynamics on lift surface. Just let a surface of any shape claim that it provides a lift vector with a given vector, and then the simulation merely give it that lift when it moves, according to speed. Ditto for drag, etc.

[17:07] Eddi Decosta: well night could be very long here ..

[17:07] Andrew Linden: Yes, we're interested in making SL better for game development

[17:07] Kadah Coba: Yay

[17:07] Morgaine Dinova: Good to hear!

[17:07] Andrew Linden: The guy to talk to for that is Gez Linden, if anyone is really interested.

[17:07] Leonel Iceghost: for game development?

[17:07] Eddi Decosta: game developement? is not a social network?

[17:08] Andrew Linden: He's working on a Linden-developed "Welcom Experience" that is a bit more gamey than most of what you see in SL

[17:08] Lomgren Smalls: heh, rubik's cube.

[17:08] Andrew Linden: and I expect a few game-dev friendly features to be developed to support it, that will help others who want to make games

[17:08] Eddi Decosta: Andrew first good welcome experiance you could have its to erease the basic mode, in viewer ..

[17:08] Mastorian Kingsford: wow 80s flashback

[17:09] Morgaine Dinova: Well virtual worlds comprise everything. Games, social networking, scientific simulation, are all just subsets of VW.

[17:09] Vincent Nacon: about time... but promise there won't be any god damn talking bird to start with like last time? ;)

[17:09] Eddi Decosta: honestly .. its a very poor mode

[17:09] Eddi Decosta: no way to buy l$ devise and no way to see this last if you buy in market

[17:09] Andrew Linden: Personally I agree with you Eddi. Basic mode gives me convulsions.

[17:10] Eddi Decosta: Andrew, you lost one chance to get more custumer to biuy the l$ devise

[17:10] Andrew Linden: I hear others like it, but I hear they are going to be tweaking it again.

[17:10] Andrew Linden: perhaps for the better

[17:10] Eddi Decosta: Andrew, i bet on a davenced mode

[17:10] Vincent Nacon: fire UI designer/coder! need easy customizable UI workflow for UI skin designer

[17:10] Eddi Decosta: i dont know if you see kirsten viewer ..

[17:11] Andrew Linden: I haven't seen any 3rd party viewers.

[17:11] Lomgren Smalls: Yes, what I would love to see is a fully skinnable UI

[17:11] Vincent Nacon: no wonder

[17:11] Andrew Linden: Too busy working on the simulator.

[17:11] Eddi Decosta: but , you it have some icon in nav bare to do the preferance and other thing like that , it a very good developement

[17:11] Morgaine Dinova: Firestorm is good. Hopefully Kokua will be as good or better.

[17:11] Arawn Spitteler: Is Andrew Tester allowed 3rd party?

[17:11] Eddi Decosta: Basic mode could be switch to Personal mode

[17:11] Morgaine Dinova: It was great to hear Rod say that he used all the TPVs.

[17:11] Vincent Nacon: Firestorm is good but still has that bitter taste from 2.0v

[17:12] Morgaine Dinova: Hopefully the Linden restrictions on using TPVs are now gone. The only way you can make yours better is by looking at others.

[17:12] Eddi Decosta: like firefox 6, you can do any setup to get the best browser

[17:12] Flip Idlemind: The problem is it (basic mode) does the opposite of what its goal is (help new people). Because people ask "How do you do this?" And you say "Go to <insert menu here>" And they say "I don't have any menus o.O" So you say "Ohhhh you must be on basic mode. Well log out, open the viewer, switch to advanced mode, close the viewer again, open the viewer again, log back in, and I'll help you"

[17:12] Andrew Linden: Unfortunately I'm too paranoid to use a 3rd party viewer from any computer.

[17:12] Flip Idlemind: Most people stop at "log out"...

[17:13] Vincent Nacon: learn to trust and mingle with other people, Andrew

[17:13] Eddi Decosta: Andrew, sorry but the v3 and the v2 is the same thing exept the mesh add

[17:13] Kadah Coba: :(

[17:13] Eddi Decosta: no?

[17:13] Sahkolihaa Contepomi: The most recent Snowstorm works well for me.

[17:13] Sahkolihaa Contepomi: Deferred runs nice and fast.

[17:13] Vincent Nacon: of course, don't use it on your Linden account

[17:13] Arawn Spitteler: That's why you're supposed to use Andrew Tester

[17:13] Vincent Nacon: get an alt

[17:13] Vincent Nacon: or that

[17:13] Flip Idlemind: scramble

[17:13] Morgaine Dinova: Andrew: compile your own from known (and respected) sources. Paranoia is healthy for security, but compiling your own overcomes most of the issues.

[17:14] Andrew Linden: I have a couple alts I could use.

[17:14] Andrew Linden: If I were working on the viewer I'd take the time.

[17:14] Jonathan Yap: Andrew, run a TPV in a virtual machine

[17:14] Vincent Nacon: muhaha!

[17:14] Arawn Spitteler: I don't know if Andrew has the time, but compiling his own would be good excersize

[17:14] lonetorus Habilis: you could always get a new pc, install a tpv though a seperate residential internet connection, and afterwards you can always burn the pc and chop the hdd to bits

[17:14] Sahkolihaa Contepomi: lol.

[17:14] Andrew Linden: But I'd also need an isolated machine yeah, or a virtual machine.

[17:15] Lomgren Smalls: virtual machines are very useful

[17:15] Vincent Nacon: and isolated interest?

[17:15] Liisa Runo: hmm hmm, it seems BB is returning wrong values again, i guess ill need to reopen SCR-90

[17:15] JIRA-helper:

[#SCR-90] llGetBoundingBox() returns wrong values on Magnum RC

[17:15] Vincent Nacon: internet*

[17:15] Andrew Linden: Yes, my paranoia level requires something like that lonetorus

[17:15] Vincent Nacon: my bad

[17:15] lonetorus Habilis: connect the vm to a vpn provider

[17:15] Jonathan Yap: Andrew, you could also image your HD, do tpv stuff, and reload from the backup

[17:15] Jonathan Yap: Good to make a backup every so often in any case

[17:15] Eddi Decosta: i love that word, paranoia

[17:15] Vincent Nacon: ok, enough with the stalling, my eyes are dry. Going to bed this time.

[17:16] Vincent Nacon: later all

[17:16] Andrew Linden: Yup, time to head out for me.

[17:16] Morgaine Dinova: NN Vincent :-)

[17:16] Andrew Linden: Thanks for coming everyone.

[17:16] Eddi Decosta: ty Andrew night and see you next time

[17:16] Sahkolihaa Contepomi: See you, Andrew.

[17:16] Flip Idlemind: Baiii guize

[17:16] Morgaine Dinova: Thanks Andrew, Simon :-)


Prev 2011.08.23 Next 2011.08.30