Simulator User Group/Transcripts/2011.11.18
|Prev 2011.11.15||Next 2011.11.22|
List of Speakers
|Broken Pipe||Homere Figaro||Howdy Colter|
|Jonathan Yap||Kaluura Boa||Kelly Linden|
|Leonel Iceghost||Moundsa Mayo||Nalates Urriah|
|NielArcher||Pauline Darkfury||Rex Cronon|
|Sahkolihaa Contepomi||Simon Linden||Sniper Siemens|
|Squirrel Wood||Tankmaster Finesmith||Techwolf Lupindo|
[16:01] Simon Linden: Hello everyone
[16:01] Rex Cronon: hi simon
[16:01] Rex Cronon: hi everybody
[16:01] Sniper Siemens: hi Simon
[16:02] Simon Linden: Hello ... Andrew is not available today, so I'm the host and M.C>
[16:02] Howdy Colter: Okay
[16:02] Simon Linden: For news, there's not much ... there will not be any server updates next week
[16:02] Vincent Nacon: Andrew went to MineCon?
[16:02] Vincent Nacon: :P
[16:03] Simon Linden: We've been working on some of the usual maintenance work and pulling together another chunk for the release pipeline
[16:04] Vincent Nacon: nice chair...
[16:04] Simon Linden: The next round has passed QA and should hit an RC slot a week and half from now
[16:04] Vincent Nacon: no pose, found your problem
[16:04] Simon Linden: I expect the current maintenace work will get promoted to the main channel then
[16:04] Simon Linden: but of course, something may happen between now and then
[16:05] Simon Linden: Kelly, did you have any news or announcements?
[16:06] Vincent Nacon: seem like a quiet week
[16:06] Pauline Darkfury: YAY for that, Kelly, looks like it should be good
[16:06] Kelly Linden: I think we've talked about it before, but if there are any questions I can talk more about llTransferLindenDollars
[16:06] Squirrel Wood: Woo!
[16:06] Kelly Linden: If you would like to try this feature out it is currently active on the channel DRTSIM-111 on regions Bethel, Fortuna and Sandbox Wanderton on Aditi.
[16:06] Simon Linden: Has anyone been able to try it out yet? It's on a few Aditi regions
[16:06] Techwolf Lupindo: That is for object to pay some, not someone paying object, correct?
[16:07] Kelly Linden: Correct
[16:07] Vincent Nacon: I hope that won't courage more gamling slots in SL with these new update ;)
[16:07] Rex Cronon: can u transffer money to prims?
[16:07] Rex Cronon: ok. u just answerd it
[16:07] Kelly Linden: Prims don't hold money, they act as a proxy for their owner if the appropriate permission is set
[16:07] NielArcher: I've been trying it
[16:07] Pauline Darkfury: Objects & groups have never been able to actually hold cash
[16:08] Tankmaster Finesmith wonders what happens if someone is able to spoof the owner name on the object to someone else and use that
[16:08] Squirrel Wood: I wonder if there are plans to add an option as a player to automatically refuse money payments above a certain amount ?
[16:08] Kelly Linden: Pretty sure you can't spoof the owner id.
[16:08] Techwolf Lupindo: I notice there is a =<0 error flag.
[16:08] Vincent Nacon: like the no-name creator object?
[16:08] Pauline Darkfury: One thing I'd kinda like to see sometime in the future is PERMISSION_DEBIT_LIMITED, e.g. may amount capped by the last money event
[16:09] Kelly Linden: interesting idea pauline.
[16:09] Squirrel Wood: I've had it happen to me that someone paid me 600k L$ of fraudulent money and believe you me. I was not happy about that.
[16:09] Tankmaster Finesmith: either that vincent, or i seem to recall in the past some were able to use a greifer client to get someoen else's name on the prim so they werent the ones that were blamed for the problems cuased by it
[16:09] Vincent Nacon: some people like seeing their money go red!
[16:09] Pauline Darkfury: That's basically the remaining risk in the process, that a misbehaving, but non-malicious script can empty your account
[16:10] Vincent Nacon: yeah Tank, it was the creator name but they couldn't spoof off the owner
[16:10] Tankmaster Finesmith: ah ok
[16:10] Pauline Darkfury: Owner should be impossible to spoof, and PERMISSION_DEBIT would have to be granted again (if the system is working properly)
[16:11] Simon Linden: Does anyone know why there's a chair on the table?
[16:11] Vincent Nacon: for me to sit on
[16:11] Techwolf Lupindo: Soneone blow it up. :-0
[16:12] Jonathan Yap: Squirrel owns it
[16:12] Kelly Linden: a proxy shape for physics on that would drop the PE to 18 or 19.
[16:12] Vincent Nacon: better not be some kind of show and tell
[16:12] Vincent Nacon: what's wrong with it?
[16:12] Jonathan Yap: Can you (Lindens) tell how much mesh is being used for static objects (vs what is being worn)?
[16:12] Pauline Darkfury: So, question on the hardware maint. What issue(s) is intened to be fixed by it?
[16:13] Simon Linden: Do you mean global usage, Jonathan?
[16:13] Jonathan Yap: yes simon, in general, but that is probably off topic for this meeting
[16:13] Pauline Darkfury: Or is it just getting the fuzz balls out from the oldest stuff? ;)
[16:13] Vincent Nacon crings as he saw the rivets on the chair....
[16:13] Simon Linden: There's probably some involved technical way to dig that data up, but I haven't seen any reports like that
[16:14] Simon Linden: I did hear recently that the mesh usage as measured by "number of regions with some mesh" has been growing at a fairly steady and constant rate
[16:14] Pauline Darkfury: Could presumably do some sort of SQL COUNT on each region's object DB, or the asset DBs
[16:15] Kelly Linden: There are no region object DBs
[16:15] Techwolf Lupindo: Mesh usage will go up a lot once phoenix and firestorm with mesh is releasted.
[16:15] Tankmaster Finesmith: yeah
[16:15] Jonathan Yap: I was just curious about mesh items put out, since their PE costs may or may not be higher than the equivalent work done in prims
[16:15] Nalates Urriah: Any word on how many people are using mesh capable viewers?
[16:15] Vincent Nacon: muhaha!
[16:15] Simon Linden: So I think we'll continue to see gradual adoption ... building good mesh objects is a new skill set, so I don't think there will be a sudden rapid rise unless someone invents some really new and exciting content
[16:15] Tankmaster Finesmith: Nal, about 1/3 of the logins are on mesh cablible viewers
[16:16] Vincent Nacon: yeah but that's with FS
[16:16] Tankmaster Finesmith: thats should double next month
[16:16] Rex Cronon: imo the problem with mesh adoption is that u must have PIOF
[16:16] Squirrel Wood: heh. couldn't rez my mesh squirrel here if I wanted.. it uses close to 1k PE
[16:16] Simon Linden: I don't know that stat either, Nal
[16:16] Pauline Darkfury: Yeah, it's still pretty huge effort to build with mesh, compared to prims, plus the implicit no-mod-ish nature of it will keep folks using prims for some stuff
[16:16] Rex Cronon: that is if u want to create any mesh
[16:17] Simon Linden: Right, mesh isn't going to be a cure-all or replacement for all building
[16:17] Nalates Urriah: Thx Rocks & Simon
[16:17] Squirrel Wood: If you can do it cheaper using prims or sculpties - use those.
[16:17] Squirrel Wood: Mesh is simply put just another asset one can opt to use
[16:17] Leonel Iceghost: any news about the gamming stuff?
[16:18] Vincent Nacon: which a lot of professional gonna do, pretty ironic since Mesh is generally for professionals
[16:18] Tankmaster Finesmith: the land impact costs for mesh i think are holding its adoption back as much as the late adopters to a mesh capable viewer
[16:18] NielArcher: Squirrel, did you make that shirt yourself?
[16:18] Squirrel Wood: the texture, yes. the mesh - its a template.
[16:18] Kelly Linden: Could this chair be made with <18 prims?
[16:18] Simon Linden: What gaming stuff do you mean, Leonel? I'm pretty sure the answer is 'no', however
[16:18] NielArcher: template? from where?
[16:18] Jonathan Yap: Simon, parts of what Esbee demonstrated at SLCC last summer
[16:18] Vincent Nacon: it could... too much polygons in the rivets
[16:18] Leonel Iceghost: the teleport, auto attach, path finding stuff
[16:19] Jonathan Yap: forced TP, etc
[16:19] Pauline Darkfury: Yeah, the way PE works is a fairly big turn off. I have seen some mesh stuff that is very low cost, but 100% of the cases where I've tried to "do the right thing" by converting objects over to PE and no-physics prims has given either no saving or negative saving, so have pretty much given up on that aspect of it, tbh
[16:19] Tankmaster Finesmith: lindends can already do forced TP :P
[16:19] Vincent Nacon: and a basic physic shape
[16:19] Howdy Colter: Simon... in regards to SVC-472, I did a stress test of the Blake Sea earlier today... just wanted to tip my hats to you guys, 50 sims in 7 minutes at full throttle and it was flawless.
[16:19] JIRA-helper: http://jira.secondlife.com/browse/SVC-472
[#SVC-472] Region Crossings Fail
[16:19] Simon Linden: Yeah, the short answer is 'no news' ... but there are discussions going on for how and when to relese parts of it
[16:20] Simon Linden: That's great to hear, Howdy. Thanks for the update
[16:20] Vincent Nacon: Howdy, try doing that in homesteads?
[16:20] Jonathan Yap: I was having a horrible time with region crossings today. I would not make a transition and end up watching my coordinates march to crazy numbers before being logged out
[16:20] Vincent Nacon: or was it that?
[16:20] Howdy Colter: The Blake are Homesteads if I'm not mistaken
[16:20] Vincent Nacon: ah ok
[16:20] Simon Linden: Most of the Blake Sea regions are homesteads or openspace regions
[16:20] Vincent Nacon: with how many prims?
[16:21] Howdy Colter: It was a 31 prim boat dressed just as I am now
[16:21] Simon Linden: Those (homesteads and openspaces) have had the most problems with the new kernel and thus region crossings
[16:21] Howdy Colter: I was doing a sim crossing every 6 seconds
[16:21] Vincent Nacon: that should be ok but what about over PE and con-hul?
[16:21] Vincent Nacon: hull*
[16:21] Pauline Darkfury: How many scripts should be a more relevant question, I'd have thought. I've never had crossing issues due to prim quantity, but too many scripts will still kill, especially if there's any speed involved
[16:22] Moundsa Mayo: Heee. Should have been in Purple, Umber, around there todat. TD of 0.18, and twice Imprudnce crashed and i got a "NVidia OpenGL driver lost connection with video adpater message. I'd never seen that one before.
[16:22] Tankmaster Finesmith: you got that om inprudence??
[16:22] Moundsa Mayo: Yus
[16:22] Tankmaster Finesmith: that suprised me
[16:22] Simon Linden: Yes, I think scripts are the biggest problem for crossings ... each one adds a good-sized chunk of memory that has to be packaged, sent and un-packaged for the crossing
[16:23] Tankmaster Finesmith: its been a huge issue with V3, bit havent heard of anyone on V1 with that problem
[16:23] Moundsa Mayo: Yeah, dual GTX 460's, too
[16:23] Tankmaster Finesmith: well, if its any card, its the 460
[16:23] Sahkolihaa Contepomi: There's still the driver bug in the 460s though, Tank.
[16:23] Moundsa Mayo: I think an exploit was ongoing around there.
[16:23] Tankmaster Finesmith: thats been the most netorious card coming across that driver bug
[16:23] Jonathan Yap: Moundsa, there is a windows registry patch you can apply to help with that issue
[16:23] Moundsa Mayo: Ah, cool! Thanks.
[16:24] Tankmaster Finesmith: no mondusa, the problem is that LL is using really olf OGL calls, and drivers are starting to no longer support them
[16:24] Moundsa Mayo: Proof again that I'm smartest when picking others' brains B^D
[16:24] Tankmaster Finesmith: thats where the underlying issue is coming from for the driver issues
[16:24] Jonathan Yap: olf=old
[16:24] Tankmaster Finesmith: LL is close to releasing clode thatll update the openGL calls
[16:24] Tankmaster Finesmith: which will fix that
[16:24] Moundsa Mayo: If there's a Jira for it I'll comment
[16:24] Simon Linden: There is some work in the viewer team to clean up the OpenGL usage, but it's not out yet
[16:25] Sahkolihaa Contepomi: There is a JIRA, but it's already fixed.
[16:25] Tankmaster Finesmith: it almost pasted QA yesterday
[16:25] Pauline Darkfury: I find it rather odd that drivers are failing to at least provide compatibility with older OpenGL calls. There's going to be tons of legacy software using them, seems very odd to pull them completely
[16:25] Sahkolihaa Contepomi: Yeah, couple of other blocking bugs (one big one with performance when lights and shadows are enabled).
[16:25] Moundsa Mayo: Is this always associated with the TD issue?
[16:25] Moundsa Mayo: I use lights but no shadows.
[16:25] Tankmaster Finesmith: paul, you can only keep backwards compatibility so long
[16:25] Jonathan Yap: Your graphics issue is unrelated to TD values
[16:26] Moundsa Mayo: I've mostly never had problems involving graphics, even when on an older system.
[16:26] Simon Linden: Backwards compatibity is really difficult to program with complex systems like that
[16:26] Moundsa Mayo: Then something else was in that area.
[16:26] Tankmaster Finesmith: using older vid drivers works for now
[16:26] Moundsa Mayo: ANd another i was with experienced repeated crashes.
[16:26] Sahkolihaa Contepomi: v1, v2 and current v3 have some .. really weird stuff going on with OpenGl.
[16:26] Sahkolihaa Contepomi: Some stuff is fixed functions (deprecated), some is modern shaders (shadows) and newer drivers aren't liking it.
[16:26] Pauline Darkfury: Well, I just find it odd that the driver folks don't have a library mapping the old calls onto new, even with less than 100% efficiency
[16:27] Tankmaster Finesmith: driver crashins, moundusa?
[16:27] Moundsa Mayo: +1 pauline.
[16:27] Tankmaster Finesmith: the do, paul
[16:27] Moundsa Mayo: Vierw crashes, Tank
[16:27] Sahkolihaa Contepomi: Pauline - most NVIDIA cards pull that off OK. But Runitai found a specific bug with 400 series cards.
[16:27] Leonel Iceghost: Pauline they do that since forever, that is their way to force you buy new graphic cards
[16:27] Tankmaster Finesmith: but theres a point where it hurts performance on the new capibility
[16:28] Moundsa Mayo: Sure, but degraded performance preferable to crashing.
[16:28] Jonathan Yap: your crashing is a timeout issue Moundsa
[16:28] Simon Linden: Hopefully that will be out soon ... a pretty high percentage of viewer crashes are related to the graphics drivers and cards. It would be great to get that crash rate to drop
[16:28] Sahkolihaa Contepomi: It should be soon, hopefully.
[16:28] Meeroo Home V1.0: Sim Time Dilation has dropped to 0.74, putting meeroo in reduced activity mode!
[16:28] Vincent Nacon: I'm sure PS3 programmers would crings to talk about these
[16:28] Techwolf Lupindo: Regine frozze
[16:28] Moundsa Mayo: I'm not comlaining - was more curious if the TD was part and parcel of the driver issue.
[16:29] Tankmaster Finesmith: agreed simon
[16:29] Sahkolihaa Contepomi: Performance wise, there's quite an increase in that too.
[16:29] Squirrel Wood: old ogl calls that have been declared "deprecated"... that was the issue, right =
[16:29] Rex Cronon: make a text only viewer, and no more crashes;)
[16:29] Sahkolihaa Contepomi: Squirrel - yes.
[16:29] Tankmaster Finesmith: yes, squuirl
[16:29] Squirrel Wood: and deprecated simply means that drivers do not have to support them anymore :p
[16:29] Moundsa Mayo: Rex, use dice, pencil, and paper and save on electricity B^D
[16:29] Pauline Darkfury: TD should be totally unrelated to client issues (other than an object which is torturing the server may also be torturing the client)
[16:29] Meeroo Home V1.0: Resuming normal meeroo activity!
[16:29] Tankmaster Finesmith: some of the calls were depricated 5 years ago
[16:30] Homere Figaro: ooops
[16:30] Homere Figaro: hi
[16:30] Jonathan Yap: No meeting next week due to the holiday?
[16:30] Moundsa Mayo: Yes, Pauline. Thats why we suspected some new griefer activity.
[16:30] Simon Linden: We'll have the Tuesday noon user group, but not a week from today
[16:31] Sahkolihaa Contepomi: Ah, Turkey Week as Kelly called it.
[16:31] Vincent Nacon: don't tell the sailors about it
[16:31] Tankmaster Finesmith: ?
[16:31] Vincent Nacon: ;)
[16:31] Simon Linden: There may not be much news because we're not rolling out new code, but we should have a better idea of the following week's releases
[16:32] Moundsa Mayo: We were going to test a KeyframedMotion train there but had to go to Heterocera.
[16:32] Pauline Darkfury: Would be really great if LeTigre could get promoted. The llMinEventDelay bug has been open rather too long, given it causes significant breakage of products
[16:32] Simon Linden: Let's see ... I don't have the jira number offhand, but there was a bug about being able to enter the premium sandboxes that was getting attention today
[16:32] Vincent Nacon: didn't suffer any prim drifting effect, Mayo?
[16:32] Simon Linden: I _think_ that's next in line, Pauline
[16:32] Moundsa Mayo: No, kitto's build was fine.
[16:33] Kelly Linden: (technically actually the bug is about *not* being able to enter premium sandboxes when you should be able to)
[16:33] Jonathan Yap: svc-7459 has been getting a ton of comments
[16:33] JIRA-helper: http://jira.secondlife.com/browse/SVC-7459
[#SVC-7459] Drop In Traffic Calculations
[16:33] Kelly Linden: A relog usually fixes it
[16:33] Simon Linden: Right
[16:33] Rex Cronon: u need a gold card, or something?
[16:33] Pauline Darkfury: Mmm, yeah, there's a lot of folks unhappy with traffic stuff
[16:33] Tankmaster Finesmith: golden ticket
[16:33] Simon Linden: I believe it's caused by the sequence of regions you visit ... regions on one of the RC's is dropping a chunk of required information
[16:33] Moundsa Mayo: Two issues were evident, though - lack of endpoint precision related to speed, and region crossing hangs.
[16:34] Techwolf Lupindo: If eveyone traiff stats drop, then the seearch order remains the same.
[16:34] Vincent Nacon: that's the same issue, Mayo. The prim drift
[16:34] Pauline Darkfury: Yeah, but some people are abandoning malls, when the traffic drops below a threshold, etc
[16:34] Pauline Darkfury: Plus I'm not convinced that 100% of parcels are dropping at the same rate
[16:35] Tankmaster Finesmith: their loss
[16:35] Moundsa Mayo: The endpoint precision? My mistake - i though you meant prims in the vehicle relative to one another.
[16:35] Pauline Darkfury: SL's loss, if that causes someone to close a region
[16:35] Simon Linden: I haven't seen that traffic bug before ... but I don't know of any work in that area
[16:35] Vincent Nacon: yeah, not really sure if they are actually related but
[16:35] Kelly Linden: I need to head out early today. Have a good weekend everyone.
[16:35] Sahkolihaa Contepomi: See you, Kelly. Enjoy your weekend.
[16:35] Tankmaster Finesmith: tc kelly
[16:35] Moundsa Mayo: Thanks for your work, Kelly!
[16:36] Pauline Darkfury: Take care, Kelly, have a good weekend
[16:36] Jonathan Yap: bye Kelly
[16:36] Tankmaster Finesmith: enjoy your weekend ?
[16:36] Rex Cronon: tc kelly
[16:36] Simon Linden: Actual SL usage has been pretty constant, so I suspect that's just numbers changing. It's obviously important, but I don't think the drop actually indicates a change in usage patterns
[16:36] Pauline Darkfury: Yeah, I think it's a counting error, rather than actual change in usage
[16:36] Tankmaster Finesmith wonders why a bunch of land in the distance just poped into jiew
[16:37] Vincent Nacon: SVC-7415
[16:37] JIRA-helper: http://jira.secondlife.com/browse/SVC-7415
[#SVC-7415] llSetKeyframedMotion skipping position distance worst over given time
[16:37] Tankmaster Finesmith: view*
[16:37] Leonel Iceghost: the jira says it is not proportional, so low traffic places lose "less"
[16:37] Moundsa Mayo: Thanks - was just searching Jira
[16:37] Vincent Nacon: it's getting harder to make a repo of that bug
[16:38] Broken Pipe: Your Meeroo is digging and seems to have found something!
[16:38] Moundsa Mayo: Is it being tuned out? Or is it an artifact of other network behaviors changing?
[16:39] Vincent Nacon: I dunno, the last test, I was alone in a sim
[16:39] Leonel Iceghost: the llSleep(0.0001) is going back to skipping a frame?
[16:40] Simon Linden: It looks like Maestro has been working on SVC-7415 so I'm not sure we'll be able to do anything until it's more reproducible
[16:40] Moundsa Mayo: Good to know it's in the mill!
[16:41] Vincent Nacon: not really
[16:41] Vincent Nacon: they can't do anything about it
[16:42] Vincent Nacon: it's gonna need someone who got the time and experienced in Havok's source
[16:42] Howdy Colter chuckles @ the names of Simon's Meeroos
[16:42] Techwolf Lupindo: hehe
[16:42] Vincent Nacon: don't think he spent enough time with it
[16:42] Techwolf Lupindo: Laggi, Restart, Ruth, and Crash.
[16:43] Simon Linden: The other thing we've learned about llSetKeyframedMotion() is that if you're very concerned about accuracy, pick time intervals that are multiples of 1/45
[16:43] NielArcher: need a fifth Meeroo. TD only visible IF the sim has enough resources
[16:44] Vincent Nacon: 250m/s isn't the problem if that's what you're trying to say
[16:44] Moundsa Mayo: Ah, thanks Simon. Will pass that along.
[16:44] Simon Linden: Well, trying to do anything in SL at 250m/sec is probably going to have problems :P
[16:45] Moundsa Mayo: Heeee.
[16:45] Simon Linden: But it's just some basic math .. the world steps in 1/45 of a second inverval
[16:45] Pauline Darkfury: Mmm, what actually needs to move at that speed, and needs KFM?
[16:45] Vincent Nacon: yeah but it shouldn't have issue when moving 8meter in 0.9 second
[16:45] Simon Linden: if you were thus trying to do something in 100ms, taht's not exactly on one frame or the next
[16:46] Moundsa Mayo: Sure, always try to match the granularity of the simulation.
[16:46] Vincent Nacon: ok but you're talking about timing... we're talking about the end position result
[16:46] Moundsa Mayo: IF you can avoide edge effects.
[16:46] Simon Linden: 0.9 seconds is 40.5 frames
[16:46] Simon Linden: So you're exactly in-between 2 frames
[16:46] Tankmaster Finesmith: what if the region isnt running at 45FPS?
[16:46] Vincent Nacon: yeah but why 0.2 meter difference?
[16:47] Squirrel Wood: back in ye olde hacok 1 days one could cross sims so quickly, one would actually skip some sims in the path right away
[16:47] Tankmaster Finesmith: lol
[16:47] Simon Linden: well, 8 meters over 40.5 frames is 0.19753086419753086419753086419753 meters per frame
[16:47] Simon Linden: (give or take a bit)
[16:47] Squirrel Wood: havok..
[16:48] Simon Linden: That looks suspiciously close to your .2 question
[16:48] Pauline Darkfury awards Simon today's excessive precision award ;)
[16:48] Tankmaster Finesmith: ?
[16:48] Vincent Nacon: so it's a rounding up issue at the end of each position loop?
[16:48] Jonathan Yap: Is .2 in binary a "problem" number?
[16:49] Simon Linden: I'm not sure ... just playing with the math here. I don't know the code, nor Havok, well enough to know how it handles these differences
[16:49] Jonathan Yap: I would think it would be even, as it's a power of 2
[16:49] Jonathan Yap: so no rounding issue there
[16:49] Vincent Nacon: yeah, think you could add that thought in the comment for him to read?
[16:49] Squirrel Wood: sum up 0.01 + 0.01 + .. and print the result. you will know immediately if it is a problem number
[16:49] Vincent Nacon: might be a start for him
[16:50] Simon Linden: There was some discussion last time along these lines ... if it's rounding or truncating the time to a specific frame interval, it couild produce errors like this
[16:50] NielArcher: Jonathan, what's a power of two?
[16:50] Pauline Darkfury: if it's off by 0.2 at the end of movement, and it's approx 0.2 per frame, an off by 1 type error prematurely ending a loop?
[16:50] Kaluura Boa: Stupid question: Why wouldn't llSKfM() simply round up the time internally to match its needs of time expressed in frames
[16:50] Kaluura Boa: ?
[16:50] Squirrel Wood: avoid the error by multiplying 0.01 by the wanted amount.
[16:50] Moundsa Mayo: Edge effect
[16:51] Simon Linden: That could be, Pauline, but I'm guessing ... but from the reports on accuracy, it seems like they are down to the size and scale of what one frame's motion would look like
[16:51] Vincent Nacon: yeah
[16:51] Jonathan Yap: Sorry, I am tired, .2 is 00101110 00110010 in binary
[16:51] Simon Linden: It would be interesting to see if you changed that interval to 1 second, for example, and if it suddenly becomes more accurate
[16:52] Vincent Nacon: might be better for some people to have position based on region instead of local
[16:52] Kaluura Boa: It should... 1 second is 45/45...
[16:52] Pauline Darkfury: yeah, trying time & distance values which are a small range either side of the problem values should be enlightening. Possibly also 0.5x and 2x the values
[16:54] Simon Linden: So for now, if you really need accuracy on that, I'd suggest using some time value that's an even multiple of 1/45. I don't have proof that will really help, but it makes some sense if Havok our our code doesn't interpolate carefullly
[16:54] Object: Hello, Avatar!
[16:54] Simon Linden: * havok OR our code...
[16:55] Leonel Iceghost: also llGetVel reports wrong when sim fps goes down
[16:55] Leonel Iceghost: you can be running and a hud reporting 30m/s
[16:55] Vincent Nacon: really?
[16:55] Simon Linden: That's an interesting one, Leonel.
[16:56] Leonel Iceghost: yes, it is random but I can reproduce it with a hud more than with others
[16:56] Leonel Iceghost: it uses llSetForce
[16:56] Leonel Iceghost: a constant one
[16:58] Simon Linden: Are there any last bugs or issues to bring up before the hour is over?
[16:59] Pauline Darkfury: So, before the end, quickly, is the hardware maint today & Monday supposed to fix anything we'd notice, or just random preventative maint?
[16:59] Simon Linden: It's not _completely_ random :P
[16:59] Pauline Darkfury: Beyond the regions offline for longer than expected, which I think was mentioned for it
[16:59] Tankmaster Finesmith: lol
[17:00] Pauline Darkfury: Just wondering if it's supposed to fix any glitches when regions are not offline?
[17:00] Simon Linden: It's some work trying to help with our memory usage and performance problems
[17:00] Pauline Darkfury: Ahh, ok, so might squish some of the obscure perf bugs, maybe, hopefully, perhaps? ;)
[17:00] Simon Linden: It may help a bit with region crossings and overall performance, but it's not going to be a dramatic change
[17:00] NielArcher: hope not. I now have a horrible image of a blind fold Linden walking into the colo and pinning a label on the first rack they bump into for doing maintenance
[17:01] Pauline Darkfury: Ok, cool, thanks, pretty much what I'd guessed at
[17:01] Vincent Nacon: muhaha!
[17:01] Simon Linden: We spin the tech around a few times before they stagger towards the racks...
[17:01] Squirrel Wood: So, what would be the result if I ran a Qualys Scan against LL servers? :)
[17:01] NielArcher: Simon, but that makes it COMPLETELY random
[17:02] Simon Linden: lol, I'll pass that along...
[17:02] Rex Cronon: they send the men in black after u:)
[17:02] Simon Linden: Thanks everyone for coming today and the good discussion
[17:02] Tankmaster Finesmith: have a great weekend everyone ?
[17:02] Moundsa Mayo: Simon, thanks for your time and all your hard work!
[17:02] Squirrel Wood: Qualys = Vulnerability scan
[17:02] IelArcher: thanks for having us ;-)
[17:02] Pauline Darkfury: To make that random, you have to blindfold them, put them on a wheely office chair, wheel around the racks randomly, spin them around, then let them go ;)
[17:02] Rex Cronon: tc simon and everybody else
[17:02] Squirrel Wood: to find actual and possible security issues
[17:02] Simon Linden: I think Andrew and I will both be here next week, but a reminder that one week from now we'll skip the user group
[17:02] Nalates Urriah: Thx all
[17:03] Simon Linden: Feel free to show up if you want, but I don't think there will be LIndens here
[17:03] Sahkolihaa Contepomi: Have a good weekend, Simon.
[17:03] Tankmaster Finesmith: ill be out of town for both
[17:03] Vincent Nacon: muhaha!
[17:03] Simon Linden: Thanks ... have a good weekend, everyone
[17:03] Sahkolihaa Contepomi: Gonna be a quiet week for us Europeans.
[17:03] Pauline Darkfury: Running unapproved vuln scans can get you an early morning visit from the FBI if they upset the wrong person, not a recommended activity
[17:03] Simon Linden: Best of luck with your rezzing and region crossings :)
[17:03] Moundsa Mayo: B^D
[17:03] NielArcher: bye all
[17:03] Pauline Darkfury: Thanks, Simon, have a good weekend
[17:03] Rex Cronon: bye *
|Prev 2011.11.15||Next 2011.11.22|