User:Andrew Linden/Office Hours/2007 10 09
Jump to navigation
Jump to search
Transcript of Andrew Linden's office hours:
[16:29] | Andrew Linden: | hey Ryozu |
[16:29] | Ryozu Yamamoto: | Heya |
[16:29] | Ryozu Yamamoto: | How's work going? |
[16:29] | Andrew Linden: | looks like it just you and me today. |
[16:29] | Ryozu Yamamoto: | Kitto was here just a few minutes ago |
[16:29] | Andrew Linden: | Things are moving along. I'm currently fixing some avatar movement bugs |
[16:29] | Ryozu Yamamoto: | I know I'm not the only one who cares O_o |
[16:29] | Ryozu Yamamoto chuckles | |
[16:29] | Andrew Linden: | we'll wait a little longer |
[16:30] | Andrew Linden: | Ryozu Yamamoto, you related to Ryozu Kojima? |
[16:30] | Ryozu Yamamoto: | Same Me =) |
[16:30] | Kitto Flora: | Eeeep! |
[16:31] | Ryozu Yamamoto: | Up until recently, I was the only Ryozu in SL |
[16:31] | Kitto Flora: | Meeting where lindens almost outnumber residents! |
[16:31] | Ryozu Yamamoto chuckles | |
[16:31] | Andrew Linden: | Ah, well I'm working on SVC-785 right now |
[16:31] | Andrew Linden: | turns out there are two bugs involved there |
[16:31] | Sidewinder Linden: | hi guys |
[16:31] | Andrew Linden: | hello sidewinder |
[16:31] | Ryozu Yamamoto: | Heya Sidewinder =) |
[16:31] | Kitto Flora: | Hello hello hello |
[16:32] | Sidewinder Linden: | hi kitto, ryozu |
[16:32] | Andrew Linden: | alright, I'll just start with a quick summary update... |
[16:32] | Sidewinder Linden: | maybe i shouldn't have sat way over here - thought we'd fill up with folks ;) |
[16:32] | Andrew Linden: | There were four different crash modes discovered on Friday |
[16:32] | Ryozu Yamamoto: | 785 would explain why flight assists act wonky too |
[16:32] | Kitto Flora: | Spect everyone waiting for an update |
[16:32] | Andrew Linden: | I've fixed one (I think) |
[16:32] | Ryozu Yamamoto: | Ouch, 4 |
[16:33] | Sidewinder Linden: | well, better to find them now ryozu, eh? |
[16:33] | Andrew Linden: | another is deep in Havok4, however it is related to the RCCS system |
[16:33] | Andrew Linden: | I've filed a ticket with Havok about that one. |
[16:33] | Ryozu Yamamoto: | Indeed =) |
[16:33] | Ryozu Yamamoto: | If we don't find them now, someone else will after it's live |
[16:33] | Ryozu Yamamoto: | And from what I've learned, half the time, they won't report them |
[16:33] | Andrew Linden: | The other two I haven't really dug into yet since I don't have repro recipe's for them yet |
[16:33] | Ryozu Yamamoto nods. | |
[16:34] | Ryozu Yamamoto: | If you have any clues at all, I'm willing to give it a go |
[16:34] | Andrew Linden: | since there were so many avatar motion bugs I figured I'd try to knock some of those down first and push an update sooner |
[16:34] | Andrew Linden: | I've fixed several avatar motion bugs... |
[16:34] | Ryozu Yamamoto: | Awesome |
[16:35] | Ryozu Yamamoto: | Say, did you clean the super glue off our feet yet? XD |
[16:35] | Andrew Linden: | currently working on SVC-785 which is fixed... except that it is actually two bugs at work and I haven't quite fixed the second half. |
[16:35] | Sidewinder Linden: | super glue? that sounds like a good story... |
[16:35] | Kitto Flora: | Any news of 756? |
[16:35] | Andrew Linden: | meanwhile, I've got 1.5 other developers helping me out and they've fixed a few bugs I think |
[16:35] | Ryozu Yamamoto: | Awesome, SVC-785 explains a lot of problems with flight assists, motion enhancers, all sorts |
[16:36] | Ryozu Yamamoto: | 1.5? Put the chainsaw away =P |
[16:36] | Andrew Linden: | Kelly is working on some linking bugs, and Simon was tracking down some LSL action bugs |
[16:36] | Ryozu Yamamoto nods. | |
[16:36] | Andrew Linden: | so we should have an update tomorrow |
[16:37] | Andrew Linden: | I'm thinking maybe I'll disable the RCCS tomorrow, at least in a few regions maybe, so we can try to find some repro recipes that are unrelated to RCCS trickery. |
[16:37] | Ryozu Yamamoto: | Something about the repro on llCreateLink making links out of order... Did a recent maingrid update break llCreateLink here? |
[16:37] | Andrew Linden: | so that's the summary |
[16:37] | Andrew Linden: | Anybody here play with vehicles over the weekend? |
[16:37] | Ryozu Yamamoto: | The script I made exhibited the behavior as described on the Beta grid |
[16:37] | Kitto Flora: | I did only a bit |
[16:37] | Andrew Linden: | ? Which script? |
[16:38] | Ryozu Yamamoto: | But it wouldn't even create a link on main grid |
[16:38] | Kitto Flora: | Car script |
[16:38] | Ryozu Yamamoto shouts: http://jira.secondlife.com/browse/SVC-768 | |
[16:38] | Sidewinder Linden: | i don't see how a main grid change could change link behvaior here or vice versa |
[16:38] | Kitto Flora: | The problem is not loss of energy |
[16:38] | Ryozu Yamamoto: | I mean, I can't seem to use llCreateLink here at all |
[16:38] | Ryozu Yamamoto: | So I couldn't verify that on the main grid, a new link goes in right after the root |
[16:38] | Andrew Linden: | hrm... |
[16:39] | Ryozu Yamamoto: | 99% sure it does, but I'm not sure if it's me that busted the script or the main grid that did it |
[16:39] | Ryozu Yamamoto: | Ah, as far as vehicles go, there's the video I uploaded to uh |
[16:39] | Ryozu Yamamoto: | SVC-772 |
[16:39] | Andrew Linden: | I didn't see that video yet. |
[16:40] | Ryozu Yamamoto: | http://youtube.com/watch?v=N5JEM3SLNOA |
[16:40] | Kitto Flora: | Vehicle problem : Someone said to me they reconed it was prim over prim friction set way too high (one prim physical) |
[16:40] | Andrew Linden: | oh right, i haven't looked into that problem yet |
[16:40] | Ryozu Yamamoto: | Yeah, that's 772 |
[16:40] | Andrew Linden: | I don't think it is very hard to fix. I'll dig into it tomorrow morning. |
[16:40] | Kitto Flora: | Which, Andrew? |
[16:40] | Ryozu Yamamoto: | That's the Dominus there of course |
[16:40] | Troy Finsbury shouts: hello? | |
[16:41] | Sidewinder Linden: | i guess it's about time to try the f-16 again with this build, eh? |
[16:41] | Ryozu Yamamoto: | Also funny how the car bounced when it hit the ground ;) |
[16:41] | Sidewinder Linden: | hopefully it will be automagically fixed with these :) |
[16:41] | Andrew Linden: | you know Sidewinder, I never got around to messing with the f-16 |
[16:41] | Ryozu Yamamoto: | f-16? |
[16:41] | Ryozu Yamamoto: | plane? |
[16:41] | Sidewinder Linden: | it's a huns valen f-16 |
[16:41] | Sidewinder Linden: | yup jet |
[16:41] | Ryozu Yamamoto: | Ah |
[16:41] | Andrew Linden: | Sidewinder found a plane that acted broken, but was also broken on SL proper. |
[16:42] | Ryozu Yamamoto: | Heh |
[16:42] | Ryozu Yamamoto: | I recognize the Huns Valen name |
[16:42] | Ryozu Yamamoto: | Don't think I've flown it before however |
[16:42] | Sidewinder Linden: | well on sl proper it would break if you push hard aross region boundaries, but would work in one region pretty well |
[16:42] | Sidewinder Linden: | it's very complex |
[16:42] | Andrew Linden: | lessee... I had an issue to raise here... trying to remember what it was |
[16:42] | Kitto Flora: | I tried my F20 in a H4 sim - seemed to work ok inside the sim |
[16:42] | Ryozu Yamamoto: | Welcome Troy, we're discussing Havok4 |
[16:42] | Troy Finsbury: | ah |
[16:43] | Sidewinder Linden: | that's good |
[16:43] | Sidewinder Linden: | which f20 is that? |
[16:43] | Ryozu Yamamoto: | I've never had much luck with sim crossings in any aircraft =( |
[16:43] | Kitto Flora: | My F20 |
[16:43] | Troy Finsbury: | I have a techinal software question |
[16:43] | Sidewinder Linden: | this is the f-16 i'm talking about ryozu |
[16:43] | Troy Finsbury: | cool jet |
[16:43] | Ryozu Yamamoto: | Nice |
[16:43] | Ryozu Yamamoto: | I should pick up a copy |
[16:43] | Andrew Linden: | ah yes... how much buoyancy should we allow? |
[16:43] | Ryozu Yamamoto: | Hmm, Buoyancy |
[16:44] | Kitto Flora plonks down a F20 | |
[16:44] | Ryozu Yamamoto: | Well |
[16:44] | Andrew Linden: | turns out currently the value that can be passed to llSetBuoyancy() is unbounded -- there is no sanity check |
[16:44] | Ryozu Yamamoto: | I know a handful of scripts that use it |
[16:44] | Andrew Linden: | but we have to bound it in Havok4 |
[16:44] | Andrew Linden: | so... what is a reasonable limit? |
[16:44] | Troy Finsbury: | where do I go for questions? |
[16:44] | agame Weston: | hi |
[16:44] | Ryozu Yamamoto: | Reasonable? Somewhere in the hundreds |
[16:44] | Andrew Linden: | what kind of questions Troy? |
[16:44] | Ryozu Yamamoto: | Feel free to ask here Troy, we might be able to help, might not. I honestly don't know where to go for general questions though |
[16:45] | Troy Finsbury: | whe I start SL now I get two screens |
[16:45] | Sidewinder Linden: | is it a ratio? i'm not sure how to interpret that number |
[16:45] | Ryozu Yamamoto: | oh, sounds like you have Console enabled |
[16:45] | Andrew Linden: | hundreds?!!! I was thinking maybe 2 |
[16:45] | Troy Finsbury: | one looks like a data screen |
[16:45] | Kitto Flora: | I use buoyancy on sim crossing - neutral... Cant test that yet though |
[16:45] | Troy Finsbury: | if I close it the game closes |
[16:45] | Ryozu Yamamoto: | Andrew: One example usage: The Jflash uses extra buoyancy to return an avatar from orbit. |
[16:45] | Ryozu Yamamoto: | It could also be used as a kind of movelock in Havok4 more so than Havok1 |
[16:45] | Troy Finsbury: | started about a week ago |
[16:46] | Sidewinder Linden: | andrew - do you have a sense from teh scripts you've seen what people actually have used for values? |
[16:46] | Andrew Linden: | how does buoyancy > 1.0 help an avatar in orbit? |
[16:46] | Ryozu Yamamoto: | Well, I can orbit you and you can try it out XD |
[16:46] | Ryozu Yamamoto: | Just kidding |
[16:46] | Sidewinder Linden: | hehe |
[16:46] | Kitto Flora: | Isnt buoyancy relly set the value of 'g' ? |
[16:46] | Sidewinder Linden: | what is the problem if we let the ceiling be too high? |
[16:46] | Ryozu Yamamoto: | That about sums it up Kitto |
[16:46] | Sidewinder Linden: | is it just that we need "something" as a sanity check |
[16:46] | Andrew Linden: | Anybody know the answer to Troy's question about the log console showing up? |
[16:46] | Kitto Flora: | Put an Av in 100g's and he heads for ground fast? |
[16:47] | Andrew Linden: | Anybody know how to turn it off? |
[16:47] | Ryozu Yamamoto: | That's a debug setting |
[16:47] | Ryozu Yamamoto: | Client->Console WIndow |
[16:47] | Troy Finsbury: | did I accidently turn something on |
[16:47] | Ryozu Yamamoto: | Probably turned it on when aiming for "Debug Settings" |
[16:47] | Ryozu Yamamoto: | Right Kitto |
[16:47] | Andrew Linden: | Yes Troy, there is a setting near the bottom of the ClientMenu called "Console Window" |
[16:47] | Troy Finsbury: | or cam retraints I pick |
[16:48] | Andrew Linden: | you probably want to untoggle it. |
[16:48] | Ryozu Yamamoto: | Uncheck it, then relog |
[16:48] | Troy Finsbury: | it's still there, I should restart? |
[16:48] | Troy Finsbury: | ok thanks alot |
[16:48] | Ryozu Yamamoto nods. | |
[16:48] | Ryozu Yamamoto: | No problem =) |
[16:48] | Troy Finsbury: | ta ta |
[16:48] | Andrew Linden: | No Sidewinder, i don't know what values are being used in llSetBuoyancy() but I've got to bound it with something. |
[16:48] | Ryozu Yamamoto: | Most physics calls are bound by max-int |
[16:48] | Sidewinder Linden: | is there any reason to bound it much below the size of the datatype? |
[16:49] | Ryozu Yamamoto: | 2bil or so |
[16:49] | Kitto Flora finally gets to SVC-772 - its the same as SVC-756. | |
[16:50] | Ryozu Yamamoto: | Oh, is it a dupe? |
[16:50] | Andrew Linden: | Hrm... I'll look at it more carefully. As I recall buoyancy is limited by the "script energy" system, so that may be enough to curb very enthusiastic values. |
[16:50] | Kitto Flora: | Yes |
[16:50] | Ryozu Yamamoto: | Well, on normal objects, it should be |
[16:50] | Ryozu Yamamoto: | Avatars have a limitless energy supply |
[16:50] | Ryozu Yamamoto: | I think |
[16:50] | Ryozu Yamamoto: | I've never really figured out if it's just limitless or extremly high |
[16:50] | Andrew Linden: | very large buoyancy would push the avatar up very fast |
[16:51] | Ryozu Yamamoto: | I mostly see it being used to push avatars down |
[16:51] | Andrew Linden: | I think you can set negative buoyancy, so a very large negative value would push objects down |
[16:51] | Ryozu Yamamoto: | Right |
[16:51] | Ryozu Yamamoto: | I use it in the OmniPhaze "Anti-gravity" |
[16:51] | Andrew Linden: | (sorry, buoyancy only affects the avatar if it is called on an attachment) |
[16:51] | Ryozu Yamamoto: | Right |
[16:51] | Andrew Linden: | yes Ryozu, but do you use very large values? |
[16:51] | Ryozu Yamamoto: | I should say "Attachments" have seemingly limitless energy |
[16:52] | Andrew Linden: | hrm... sounds like a bug |
[16:52] | Ryozu Yamamoto: | I do some math to turn 1g into 100% and let the user input custom values |
[16:52] | Ryozu Yamamoto: | bug? ;_; |
[16:52] | Ryozu Yamamoto: | That's one I'd hate to see fixed, hehe |
[16:52] | Ryozu Yamamoto: | It's been true for as long as I can remember |
[16:53] | Andrew Linden: | yeah, well I'm torn... for one the script energy system caps the power of some operations |
[16:53] | Kitto Flora: | notes say that buoyancy is between 1 and -1 |
[16:53] | Ryozu Yamamoto: | Well, it gives 1 to -1 as example |
[16:53] | Kitto Flora: | neutral t o 2g |
[16:53] | Andrew Linden: | on the other hand... its implemenation is rather broken and inconsistent |
[16:53] | Ryozu Yamamoto: | 1 go to -1g actually |
[16:54] | Sidewinder Linden: | would limiting this limit the power of some of the heavy artillery style guns like the black widows? |
[16:54] | Ryozu Yamamoto: | I noticed llPushObject seemed to use a lot more energy |
[16:54] | Ryozu Yamamoto: | Black widows? |
[16:54] | Andrew Linden: | yes... that is what I thought. As I recall we had capped buoyancy to 1 and -1 |
[16:54] | Ryozu Yamamoto: | Those use the Blitz sphere, you know |
[16:54] | Kitto Flora: | Well an accell of -1g = netral |
[16:54] | Kitto Flora: | +1g = 2g total |
[16:54] | Ryozu Yamamoto: | Users will complain if they can't at least "Reverse" gravity |
[16:54] | Andrew Linden: | but somewhere along the way someone removed the sanity check on the input value |
[16:54] | Sidewinder Linden: | they have a mode - i forget the name - that will push you far enough up that the altitude wraps i think ;) |
[16:54] | Kitto Flora: | If people use values outside the spec range.. expect strange results? |
[16:54] | Ryozu Yamamoto: | Yeah Side |
[16:55] | Ryozu Yamamoto: | That's the Blitz sphere |
[16:55] | Ryozu Yamamoto: | 95 Hollow/-360 twist |
[16:55] | Andrew Linden: | well, I'm interested in nerfing the blitz sphere |
[16:55] | Ryozu Yamamoto: | You already have |
[16:55] | Ryozu Yamamoto: | Havok4 defeats it soundly |
[16:55] | Andrew Linden: | ok good |
[16:56] | Ryozu Yamamoto: | That will render a fair number of weapons unusable |
[16:56] | Ryozu Yamamoto: | But in this case |
[16:56] | Ryozu Yamamoto: | I'm happy about breaking content =P |
[16:56] | Andrew Linden: | I'm also interested in nerfing very large values for buoyancy. The question is... what should the nerf values be? |
[16:56] | Kitto Flora: | Nasty weapons are they? |
[16:56] | Ryozu Yamamoto: | The idiots selling these things |
[16:56] | Ryozu Yamamoto: | Don't care about crashing sims or exploiting bugs |
[16:57] | Ryozu Yamamoto: | I'll buckle and say a low value is alright, just as long as it's both sane AND useable |
[16:57] | Kitto Flora: | Why should it be other than +1 to -1? |
[16:57] | Ryozu Yamamoto: | I see no problem with "Reverse" gravity |
[16:57] | Ryozu Yamamoto: | Or extra gravity to defeat someone trying to push me |
[16:57] | Andrew Linden: | I was thinking about adding a flag to the avatar that would ignore external llPushObject() calls. What do you all think about that idea? |
[16:57] | Andrew Linden: | (not in Havok4... but right after) |
[16:57] | Ryozu Yamamoto: | Nah |
[16:57] | Ryozu Yamamoto: | Push restricted zones are good enough |
[16:57] | Kitto Flora: | I'd rather have that flag on a vehicle |
[16:57] | Ryozu Yamamoto: | The only real issue are exploits |
[16:58] | Kitto Flora: | It would save 30% + of the code in trains |
[16:58] | Ryozu Yamamoto: | Vehicles may find use, yeah |
[16:58] | Ryozu Yamamoto: | But for avatars and non-vehicle flagged objects, I don't see it as a good thing |
[16:58] | Ryozu Yamamoto: | Not when just disabling push on a parcel should block it |
[16:58] | Sidewinder Linden: | and i think it would mess with folks who want to do combat sims |
[16:59] | Kitto Flora: | yes |
[16:59] | Andrew Linden: | no Kitto, I'm talking about direct llPushObject() calls from external scripts on your avatar -- shouldn't it be a "right" of the avatar to not be pushed if they don't want? |
[16:59] | Sidewinder Linden: | if the av's can turn it off individually, it's too easy to "cheat" in that world |
[16:59] | Ryozu Yamamoto: | Otherwise, it may as well just be ApplyImpulse with rotational force |
[16:59] | Andrew Linden: | kitto, you don't have griefers pushing your trains to the ends of the world, do you? |
[16:59] | Kitto Flora: | Andrew - if LL supports combat sims, you gotta be able to push avs there |
[16:59] | Kitto Flora: | Otherwise all the fighters will turn off push |
[16:59] | Kitto Flora: | and then cant get 'shot' |
[17:00] | Sidewinder Linden: | right kitto.. .that's what i was trying to say... |
[17:00] | Ryozu Yamamoto: | I was thinking more about just neat toys |
[17:00] | Sidewinder Linden: | it disables a kidn of real game play here that people want |
[17:00] | Andrew Linden: | ok, so maybe have a per-parcel/region/estate setting that overrides avatar preferences? |
[17:00] | Ryozu Yamamoto: | Mario pipes that shoot avatars (Voluntarily) |
[17:00] | Sidewinder Linden: | that makes more sense to me andrew - letting the land owners set the behavior for their space |
[17:00] | Kitto Flora: | Well, simply setting the parcel no-push does stop llPush against Avs. |
[17:01] | Ryozu Yamamoto nods. | |
[17:01] | Andrew Linden: | soo... people would have to 'agree' to play the games where they get shot-pushed. What is wrong with that? |
[17:01] | Kitto Flora: | One agrees to get 'shot at' and thus shoved around by going to a combat sim |
[17:01] | Ryozu Yamamoto: | I don't suppose -128 to 128 are reasonable values for llSetBuoyancy? |
[17:01] | Sidewinder Linden: | right |
[17:02] | Sidewinder Linden: | and how about you're in a tight spot and about to get ambushed - you turn off push for one mintue and then turn it back on |
[17:02] | Ryozu Yamamoto: | As for llPushObject though, I think the current restrictions are alright |
[17:02] | Sidewinder Linden: | really not fair to the other players and very hard to detect |
[17:02] | Ryozu Yamamoto: | A parcel owner turns it off, it just doesn't work |
[17:02] | Andrew Linden: | Ryozu, I think those are rather large for reasonable values for llSetBuoyancy() but I'm still open to input |
[17:02] | Kitto Flora: | agrees with Ryozu, except for indirect on vehicles |
[17:02] | Andrew Linden: | +/-10 seems too large to me |
[17:03] | Ryozu Yamamoto: | Don't think we can come to an agreement on that one then |
[17:03] | Ryozu Yamamoto: | But, I can take a survey if you'd like |
[17:03] | Kitto Flora: | On bouyancy... |
[17:03] | Andrew Linden: | the extremely large values for buoyancy are useful because... ? |
[17:03] | Kitto Flora: | Maybe 'change / upgrade' it to +2 to -2, set hard limits and see who complains? |
[17:03] | Andrew Linden: | coming back from being blitzed? |
[17:04] | Ryozu Yamamoto: | That's the one use in particular I know of |
[17:04] | Andrew Linden: | yeah, i was thinking more long the line of +/- 2 |
[17:04] | Andrew Linden: | ok, so suppose we tackled the blitzing problem and solved it? |
[17:04] | Sidewinder Linden: | well here's another cut on this... that particular value would be very easy to change, right? |
[17:04] | Sidewinder Linden: | what if we set it to +/- 2 |
[17:05] | Kitto Flora: | whats blitzing? |
[17:05] | Ryozu Yamamoto: | But I don't claim to know all possible ideas and their implementations |
[17:05] | Andrew Linden: | yeah easy... would require a recompile (hard coded) |
[17:05] | Sidewinder Linden: | and then if people have an issue with it in post-beta (limited rollout) we could change it without major impact |
[17:05] | Ryozu Yamamoto: | Kitto: Ever heard of the PsiTek? |
[17:05] | Andrew Linden: | no what is PsiTek? |
[17:05] | Kitto Flora: | Some Co? |
[17:05] | Ryozu Yamamoto: | Ah, Luc Aubret not as well known as I thought maybe |
[17:05] | Ryozu Yamamoto: | jadz0r Conover? |
[17:05] | Kitto Flora shrugs | |
[17:06] | Andrew Linden: | pretend I'm out of touch and don't actually log in much to play around. |
[17:06] | Kitto Flora: | (We have to pretend that? :) |
[17:06] | Andrew Linden: | s/pretend/suppose/ |
[17:06] | Ryozu Yamamoto: | There's a physics glitch currently where a sphere with certain prim properties exhibits extreme force upon whatever it touches |
[17:06] | Kitto Flora: | ok |
[17:06] | Ryozu Yamamoto: | If you just drop it on the ground, it will go off world several sims away within moments |
[17:07] | Ryozu Yamamoto: | No scripts, just primage and Havok1 |
[17:07] | Sidewinder Linden: | hmmm which properties? |
[17:07] | Kitto Flora: | Is that the thing that was previously mentioned to be disabled by h4? |
[17:07] | Andrew Linden: | I think it is taking advantage of the "penetration resolution" code (I wrote that) |
[17:07] | Ryozu Yamamoto: | A couple scripters figured out that if you rez two of them on an avatar, they get shot to uh |
[17:07] | Ryozu Yamamoto: | A few trillion meters |
[17:07] | Andrew Linden: | and the fact that we have a bug in our LLVolume geometry generation that can confuse Havok1 (and Haovk4 too probably) |
[17:07] | Ryozu Yamamoto: | The SL client can't show however high that is |
[17:08] | Ryozu Yamamoto: | So it just shows it as -2bil |
[17:08] | Ryozu Yamamoto: | What they don't know |
[17:08] | Ryozu Yamamoto: | And what I hope very much very very few people know |
[17:08] | Ryozu Yamamoto: | Is that 30 of these can crash a sim faster than the sim can send the prim data to other clients |
[17:08] | Kitto Flora may have to vanish any second.... | |
[17:08] | Ryozu Yamamoto: | Instacrash |
[17:08] | Andrew Linden: | well, instacrash isn't very hard in Havok1 |
[17:09] | Ryozu Yamamoto: | Well, apparently someone found another way to do it |
[17:09] | Sidewinder Linden: | have you tried this on h4 ryozu? |
[17:09] | Ryozu Yamamoto: | the Sphere doesn't touch Havok4 |
[17:09] | Andrew Linden: | it certinaly doesn't require 30 prims |
[17:09] | Kitto Flora: | OK is that what blitzing is? Use of these objects? |
[17:09] | Ryozu Yamamoto: | Blitzing is the use of these prims to send an avatar to the height ceiling |
[17:09] | Kitto Flora: | Does this effect still exist in H$? |
[17:10] | Kitto Flora: | H4? |
[17:10] | Ryozu Yamamoto: | Nope |
[17:10] | Kitto Flora: | Goody |
[17:10] | Ryozu Yamamoto: | I've tried, lots. Hehe |
[17:10] | Ryozu Yamamoto: | It's a horribly annoying exploit |
[17:10] | Kitto Flora: | Sounds like a loophole plugged to me |
[17:10] | Ryozu Yamamoto: | There's a second loophole that seems plugged as well |
[17:10] | Andrew Linden: | ok so very high buoyancy is used to recover from extreme blitzing... so far that is the only suggested reason for keeping the values unbounded |
[17:10] | Ryozu Yamamoto: | With these two loopholes plugged, that covers almost every way to push an avatar in no push areas |
[17:10] | Kitto Flora: | AH - the connection |
[17:10] | Ryozu Yamamoto: | Well, I don't mean to keep them unbounded |
[17:10] | Ryozu Yamamoto: | just not meaninglessly restricted |
[17:11] | Andrew Linden: | assuming blitzing is solved in Havok4 then perhaps we can constrain buoyancy |
[17:11] | Kitto Flora: | take away the loophole and you dont need the out of range values to recover... |
[17:11] | Andrew Linden: | I'll see what kind of complaints I get when I nerf it to +/- 2 |
[17:11] | Ryozu Yamamoto: | I know people who just set gravity to -200% just to go into the air for no apparent reason, it amuses them |
[17:11] | Kitto Flora votes for that | |
[17:11] | Ryozu Yamamoto chuckles | |
[17:11] | Sidewinder Linden: | that sounds good to me andrew |
[17:11] | Ryozu Yamamoto: | I'd suggest at least another point or two higher |
[17:11] | Ryozu Yamamoto: | But that's just me |
[17:11] | Andrew Linden: | I'll consider that |
[17:12] | Andrew Linden: | my main worry is extreme velocities in the physics engine |
[17:12] | Ryozu Yamamoto: | Let me try 10 buoyancy real quick and see |
[17:12] | Andrew Linden: | if the avatar and objects can't be accelerated to horrific velocities then maybe I'll make it higher |
[17:12] | Ryozu Yamamoto: | Extreme velocity is a problem (putting sim crossings aside) |
[17:12] | Kitto Flora: | Did anyone try a sailboat yet? |
[17:12] | Ryozu Yamamoto: | I haven't yet |
[17:13] | Andrew Linden: | right... I'd like to cap the max velocity of objects |
[17:13] | Ryozu Yamamoto: | I'm avoiding vehicle bugs personally because I don't know how to script vehicles yet |
[17:13] | Andrew Linden: | and with unbounded impulses coming from thing like buoyancy it is difficult |
[17:13] | Ryozu Yamamoto: | Hmm |
[17:13] | Kitto Flora: | I do almost nothing but vehicles now |
[17:13] | Ryozu Yamamoto: | Then I appoint you czar of vehicle bug testing =P |
[17:13] | Kitto Flora: | car/sled |
[17:13] | Andrew Linden: | Kitto, what kind of vehicles do you make? |
[17:13] | Kitto Flora: | TY |
[17:13] | Ryozu Yamamoto: | I may be gone a moment |
[17:13] | Kitto Flora: | Ever seen the steam train? |
[17:14] | Ryozu Yamamoto: | testing just how fast this makes me go |
[17:14] | Sidewinder Linden: | brb guys |
[17:14] | Andrew Linden: | no not yet Kitto |
[17:14] | Kitto Flora: | never seen that? |
[17:14] | Ryozu Yamamoto: | Well |
[17:14] | Andrew Linden: | not yet... wow it takes 6 scripts |
[17:14] | Kitto Flora: | Theres 200 + running in SL |
[17:14] | Kitto Flora: | Actually more like 15 scripts |
[17:14] | Ryozu Yamamoto: | I can say this much, a buoyancy of 10 isn't extreme |
[17:15] | Kitto Flora: | and about 12 in each carriage |
[17:15] | Kitto Flora: | Sold over 700 of them now |
[17:15] | Andrew Linden: | What was your peak velocity Ryozu? |
[17:15] | Ryozu Yamamoto: | Not sure, but honestly, pretty slow |
[17:15] | Ryozu Yamamoto: | I couldn't find my velocity meter =/ |
[17:15] | Andrew Linden: | Are your trains broken in Havok4 Kitto? |
[17:15] | Kitto Flora: | Totally |
[17:15] | Kitto Flora: | See SVC-756 |
[17:15] | Andrew Linden: | Ok |
[17:16] | Kitto Flora: | All car scrips over prims - grind to a halt |
[17:16] | Kitto Flora: | So cars on roads dont work |
[17:16] | Kitto Flora: | trains on tracks |
[17:16] | Andrew Linden: | oh right, that one. I'm going to try to fix it tomorrow morning. |
[17:16] | Kitto Flora: | monorail on rail |
[17:16] | Kitto Flora: | Good |
[17:16] | Kitto Flora: | Once that problems fixed I can test more stuff |
[17:17] | Ryozu Yamamoto: | Be right back |
[17:17] | Kitto Flora: | Theres aload of collision stuff in it too |
[17:17] | Andrew Linden: | Lessee... I had another issue to bring up (I think we've discussed buoyancy enough for now) |
[17:17] | Ryozu Yamamoto: | Peak velocity: 56m/s |
[17:17] | Andrew Linden: | I should have written it down... |
[17:17] | Andrew Linden: | oh Ryozu, that sounds like there is a velocity limiter working on the avatar |
[17:17] | Ryozu Yamamoto: | Oh no |
[17:18] | Ryozu Yamamoto: | Coming back down I easily broke 250 |
[17:18] | Andrew Linden: | I think I know what is keeping the avatar velocity from going unbounded |
[17:18] | Ryozu Yamamoto: | I could go up much faster |
[17:18] | Ryozu Yamamoto: | I just used 10 as a baseline |
[17:18] | Andrew Linden: | theoretically buoyancy > 1 should accelerate indefinitely, but the avatar controller itself limits speed |
[17:18] | Ryozu Yamamoto: | Ah! I see |
[17:18] | Andrew Linden: | and there is other code for capping the velocity of normal objects |
[17:19] | Ryozu Yamamoto: | It always seemed assume that llSetBuoyancy(X) just meant X * 9.8 m/s push |
[17:19] | Andrew Linden: | that's probably why I haven't seen unbounded values on buoyancy kill too many simlators yet |
[17:19] | Ryozu Yamamoto: | More or less |
[17:19] | Ryozu Yamamoto: | Constant force, but not constant acceleration |
[17:20] | Andrew Linden: | it does, however there are other forces at work... we explicitly cap the velocities of objects elsewhere in the code |
[17:20] | Ryozu Yamamoto nods. | |
[17:20] | Ryozu Yamamoto: | All is well there then |
[17:20] | Kitto Flora: | any more? gtg - bad time for me... |
[17:20] | Ryozu Yamamoto: | Mind you, llGetVel only returns a max magnitude of 250 |
[17:21] | Sidewinder Linden: | thanks for staying as long as you could kitto... |
[17:21] | Ryozu Yamamoto: | So I'd never know exact speeds above that |
[17:21] | Andrew Linden: | but... if we don't cap the input to llSetBuoyancy() to *something* then there are ther dangers ... NaN injection |
[17:21] | Ryozu Yamamoto nods. | |
[17:21] | Ryozu Yamamoto: | I totally expect a capped value |
[17:21] | Ryozu Yamamoto: | Just as long as the cap doesn't end up killing a signifigant amount of use cases |
[17:21] | Ryozu Yamamoto: | IE, a decent range |
[17:21] | Ryozu Yamamoto: | Should be fine |
[17:22] | Ryozu Yamamoto: | I think no one would notice if capped at 10 |
[17:22] | Sidewinder Linden: | i guess the part i'm missing is, why not just cap the value to either the max value of the data type of the parm, or something that won't overflow with internal calcuations, and leave it at that? |
[17:22] | Ryozu Yamamoto: | Except jadz0r and me |
[17:22] | Sidewinder Linden: | heh |
[17:22] | Andrew Linden: | ok, I guess I'll play with buoyancy and see what kind of damage I can do |
[17:22] | Ryozu Yamamoto: | Sidewinder: I'm thinking, in the case of acceleration, the velocity wouldn't max out |
[17:22] | Andrew Linden: | no Sidewinder, the max of the data type is not good |
[17:22] | Ryozu Yamamoto: | Or rather, velocity would increase exponentially |
[17:23] | Andrew Linden: | we may multiply that value in the code somewhere and produce a NaN value |
[17:23] | Ryozu Yamamoto nods. | |
[17:23] | Andrew Linden: | which if we feed to the physics engine can break stuff |
[17:23] | Sidewinder Linden: | ahh yes - that's what i meant by the other part of the comment |
[17:23] | Sidewinder Linden: | right |
[17:23] | Ryozu Yamamoto: | I imagine it IS multiplied somwhere by 9.8 |
[17:23] | Sidewinder Linden: | at the very least ;) |
[17:23] | Ryozu Yamamoto: | 10 does seem like a sane value to me. |
[17:23] | Andrew Linden: | this was the problem with vary large values that were being fed to llTargetOmega() which I fixed a few months ago |
[17:23] | Sidewinder Linden: | ok |
[17:24] | Sidewinder Linden: | again, if it's an easy change, we can just pick something, like 10 |
[17:24] | Sidewinder Linden: | and if folks have a reasonable explanation of why it's way too low it's a quick alteration |
[17:24] | Ryozu Yamamoto nods. | |
[17:24] | Andrew Linden: | yes, you can think of the units of buoyancy as 'g' |
[17:24] | Sidewinder Linden: | given the current fps available... i would think that anything above 10 is pretty umm |
[17:24] | Sidewinder Linden: | meaningless in terms of anything visual :) |
[17:24] | Ryozu Yamamoto: | But an explcit example of how much complaint would happen if it was capped at 1.0 to -1.0 |
[17:24] | Ryozu Yamamoto: | It's already filed as a bug |
[17:25] | Sidewinder Linden: | ok |
[17:25] | Sidewinder Linden: | yup |
[17:25] | Andrew Linden: | right, and they used +10 as an example of something that worked in Havok1 |
[17:25] | Sidewinder Linden: | hehe if it's roughly correlated to g's.. .umm wouldn't an avatar get killed at around 7-8? |
[17:25] | Ryozu Yamamoto chuckles | |
[17:25] | Sidewinder Linden: | so 10 actually seems reasonable from the haha reality perpsective as well then :) |
[17:26] | Andrew Linden: | anything > 1 would theoretically accelerate to unbounded values... however |
[17:26] | Sidewinder Linden: | yup |
[17:26] | Andrew Linden: | there is other code for limiting the speed of the avatar |
[17:26] | Ryozu Yamamoto: | It's not real nice though if seemingly unrelated bits of code rely on each other to limit, eh? |
[17:26] | Ryozu Yamamoto: | Well, unrelated as far as the scope of the internals of the physics engine |
[17:26] | Andrew Linden: | so as long as we can't exceed such limits, and we can't create really annoying grief objects... we should probably leave the values failry liberal |
[17:27] | Sidewinder Linden nods | |
[17:27] | Ryozu Yamamoto: | I've never heard of buoyancy used to greif.. Actually! |
[17:27] | Andrew Linden: | well, I've forgotten the other thing I was going to bring up... |
[17:27] | Ryozu Yamamoto: | You just reminded me of a use case I use quite often |
[17:27] | Andrew Linden: | and we're almost over anyway |
[17:27] | Sidewinder Linden: | ryozu? |
[17:28] | Ryozu Yamamoto: | Have you ever experienced those rather irritating physics cubes that inundate a sim with particles and self replicate? |
[17:28] | Ryozu Yamamoto: | Often times, they're physical. I can usually sit ont hem |
[17:28] | Ryozu Yamamoto: | Sit on them, set an attachment's buoyancy to large values |
[17:28] | Ryozu Yamamoto: | And knock them offworld |
[17:28] | Andrew Linden: | No I haven't seen those objects specifically, but I can imagine. |
[17:28] | Ryozu Yamamoto: | While waiting for Lindens to come clean up |
[17:29] | Andrew Linden: | ah hrm... |
[17:29] | Ryozu Yamamoto: | But on the other hand, that's... |
[17:29] | Ryozu Yamamoto: | Eh |
[17:29] | Ryozu Yamamoto: | Could be just as bad |
[17:29] | Ryozu Yamamoto: | Oh, hmm, that's the other thing that limits buoyancy velocity.. objects are returned after they hit the ceiling |
[17:29] | Andrew Linden: | Ok, well that sounds like a pretty good use of large buoyancy, but rather advanced. |
[17:30] | Andrew Linden: | not something any newbie would try |
[17:30] | Ryozu Yamamoto: | Well, that's just my personal use.. never heard of anyone else doing it |
[17:30] | Andrew Linden: | but the best reason I've heard so far for letting the values be high |
[17:30] | Ryozu Yamamoto: | And in the end, it doesn't accomplish much when they're replicating |
[17:31] | Ryozu Yamamoto: | It's only effective when I run into a handful that stopped replicating due to grey goo fence |
[17:31] | Sidewinder Linden: | oh on a completely separate topice... andrew were you able to replicate that "unlink vehicle" exploit? that one sounded pretty odd |
[17:31] | Sidewinder Linden: | i don't have the jira number at hand - this was the link/unlink a newbie vehicle to generate a crash? |
[17:31] | Ryozu Yamamoto: | O_o |
[17:31] | Andrew Linden: | what issue is that Sidewinder? |
[17:32] | Ryozu Yamamoto: | Weird |
[17:32] | Sidewinder Linden: | i'm looking but jira's a bit slow atm for me |
[17:32] | Ryozu Yamamoto: | At the moment? |
[17:32] | Ryozu Yamamoto chuckles | |
[17:32] | Sidewinder Linden: | ok ok ok :) |
[17:32] | Ryozu Yamamoto: | It's always slow =P |
[17:32] | Sidewinder Linden: | ok "particularly slow" |
[17:32] | Andrew Linden: | well, send me an email.. I've got to run. |
[17:32] | Sidewinder Linden: | no problem |
[17:32] | Sidewinder Linden: | thanks |
[17:32] | Andrew Linden: | There is somewhere I've got to be at 18:30 |
[17:32] | Andrew Linden: | Thanks for coming Ryozu. |
[17:33] | Ryozu Yamamoto: | Alright =) |
[17:33] | Ryozu Yamamoto: | No problem, glad to be here |
[17:33] | Sidewinder Linden: | ok - thanks and thanks ryozu for keeping on picking on havok :) |
[17:33] | Ryozu Yamamoto: | If you guys ever need me to do anything, let me know |
[17:33] | Ryozu Yamamoto: | I'm here to help in every way possible |
[17:33] | Sidewinder Linden: | cool |
[17:33] | Sidewinder Linden: | well i should jump as well... see you around soon |
[17:33] | Ryozu Yamamoto: | I'm usually on Ryozu Kojima on beta though ;) |
[17:33] | Ryozu Yamamoto: | Take care |
[17:33] | Sidewinder Linden: | ok |