User:Andrew Linden/Office Hours/2008 02 12
Jump to navigation
Jump to search
Transcript of Andrew Linden's office hours:
[11:02] | Teravus Ousley: | .. a group of linked prim is a single object |
[11:02] | Teravus Ousley: | though.. that's not what the sim stats reports on under 'Objects' :P |
[11:02] | Teravus Ousley: | hence the descrepency in terms |
[11:02] | Arawn Spitteler has a teleporter, that depends on a sat on prim having link numbers other than unlinked prims | |
[11:03] | Ryozu Kojima waves. | |
[11:03] | Andrew Linden: | ok, I'm here, really |
[11:03] | Rex Cronon: | hi ryozo |
[11:03] | Teravus Ousley: | Hello :D |
[11:03] | Rex Cronon: | hi andrew |
[11:03] | Arawn Spitteler: | Hi, Andrew |
[11:03] | Andrew Linden: | hello everyone |
[11:03] | Renee Faulds: | : ) |
[11:04] | Becky Pippen: | Hi all |
[11:04] | Andrew Linden: | The next update to Havok4 is currently in QA |
[11:04] | Rex Cronon: | hi |
[11:04] | Arawn Spitteler has just been experiencing the weird phenomena of Beta Grid | |
[11:04] | Andrew Linden: | looks like the QA pass is half way done |
[11:04] | Arawn Spitteler: | That's the one where STATUS_ROTATE will function? |
[11:04] | Andrew Linden: | but considering that it was started yesteday... looks like it will be done this evening sometime rather than this afternoon |
[11:05] | Andrew Linden: | Last thursday, after the office hour I held another vehicle session |
[11:05] | Andrew Linden: | where i updated one region on the Havok4 preview to the tip of the branch (pretty much what QA is looking at right now) |
[11:06] | Andrew Linden: | and it appeared that several bugged vehicles are now working |
[11:06] | Gaius Goodliffe nods. | |
[11:06] | Andrew Linden: | otherwise... mostly fixes that I've pre-announced last week |
[11:07] | Andrew Linden: | there were two blockers that I became aware of yesterday |
[11:07] | Andrew Linden: | but I fixed those already and they are in the next update (they already passed QA) |
[11:07] | Andrew Linden: | the blockers were: SVC-1449 |
[11:07] | Andrew Linden: | first object in a coalesced group gets its name canged |
[11:08] | Andrew Linden: | and a problem where "locked" content could not be rezzed from inventory |
[11:08] | Renee Faulds: | I discovered a rez faux situation this morning |
[11:09] | Renee Faulds: | the linking limit on H4 is 50 meters now |
[11:09] | Renee Faulds: | but when rezzed on a H1 it falls to pieces |
[11:09] | Rex Cronon: | that was my question:) |
[11:10] | Gaius Goodliffe: | That's true of any object, rez faux or now... |
[11:10] | Gaius Goodliffe: | *or no |
[11:10] | Teravus Ousley: | :D |
[11:10] | Ryozu Kojima: | Indeed, depending on how it's linked. |
[11:10] | Gaius Goodliffe: | If it links under H4 but not H1, if you try to rez it in H1, it falls apart. |
[11:10] | Andrew Linden: | The linkability rules in Havok4 are more permissive than Havok1 for most cases |
[11:10] | Arawn Spitteler wonders what Rez Faux is | |
[11:10] | Renee Faulds: | a packageing system |
[11:10] | Ryozu Kojima: | Arawn: it's a box/script that allows you to package, rez, and position builds that require multiple linked sets |
[11:11] | Andrew Linden: | I think there are some slight subsets of the linkability space in each case that are not contained within the other... |
[11:11] | Gaius Goodliffe: | I think with the megaprim workaround, everything H1 legal is now H4 legal... |
[11:11] | Arawn Spitteler: | So, like a tree house, with several layers, would be packaged through Rez-Faux |
[11:11] | Andrew Linden: | that is, there is a little bit of linkable H1 content that does not link in H4, and visa versa |
[11:11] | Ryozu Kojima: | I think the best answer here is to simply advise against linking things in Havok4 if you plan to release them into Havok1 sims |
[11:12] | Gaius Goodliffe nods. | |
[11:12] | Gaius Goodliffe: | It's not a bug, just a consequence of the way things work. |
[11:12] | Andrew Linden: | No Gaius, there is a small subset of H1 content that theoretically doesn't link in H4, but I was waiting to see if anyone complained |
[11:12] | Ryozu Kojima: | Theoretically? |
[11:13] | Gaius Goodliffe: | Really? Hehe. Something to do with tiny prims? |
[11:13] | Andrew Linden: | it is a special case where you link to the limit, and have small pieces in the corners of the bounding box |
[11:13] | Andrew Linden: | however, such content should fail to link in H1 if you were to rotate it such that its Axis Aligned Bounding Box (AABB) were not lined up with the world axes |
[11:13] | Arawn Spitteler thinks that will be the case in the central build of the last sim to migrate | |
[11:13] | Gaius Goodliffe: | Ah. |
[11:13] | Rex Cronon: | actaully it could have some uses. right now u need to give permissions for objects to link/unlink |
[11:14] | Renee Faulds: | this was a house that is 30x30 meters all linked to one piece as an experiment if it would unpack on a H1 |
[11:14] | Renee Faulds: | 1218 meters high |
[11:14] | Renee Faulds: | 18 |
[11:15] | Renee Faulds: | just thought other creators should be aware before they start selling products that will fail on a H1 |
[11:15] | Ryozu Kojima nods. | |
[11:15] | Andrew Linden: | Renee, are you saying this is a tree house that uses megaprims? |
[11:15] | Renee Faulds: | no |
[11:15] | Renee Faulds: | regular prims |
[11:16] | Renee Faulds: | a 2 story pitch roof house structure |
[11:16] | Andrew Linden: | please explain more about Rez Faux, and this failure to link problem in H4? |
[11:16] | Andrew Linden: | could you perhaps, provide me with a copy of the unlinkable house? |
[11:17] | Andrew Linden: | or perhaps the location of one, I can grab the entire regions state and pull the house out |
[11:17] | Renee Faulds: | yes I can provide you a copy of the house - it is a 2 strory residential home 132 prims that under H4 will link size 30x30x18 but completely unlinks as re -rezzed on a H1\ |
[11:17] | Renee Faulds: | all 132 pieces are linked |
[11:18] | Gaius Goodliffe: | Note: It links fine in H4. |
[11:18] | Renee Faulds: | yes |
[11:18] | Gaius Goodliffe: | It's a "fails to link in H1" problem. |
[11:18] | Renee Faulds: | yes |
[11:18] | Andrew Linden: | oh... sorry, I forgot that detail already |
[11:18] | Gaius Goodliffe: | Which is to say, not at H4 problem at all. |
[11:18] | Andrew Linden: | yeah, forwards compatibility in H1 for H4 is not supported |
[11:18] | Renee Faulds: | right |
[11:19] | Gaius Goodliffe: | Moral of the story: H4 only content is H4 only. ;) |
[11:19] | Renee Faulds: | I just thought it prudent to make other creators aware of this failure before products are mass released |
[11:19] | Andrew Linden: | Simon, you weren't here last week, on Thursday |
[11:19] | Gaius Goodliffe nods. | |
[11:19] | Andrew Linden: | but we talked about boosting the vehicle speed limit |
[11:20] | Andrew Linden: | and by the end of the meeting I was of the opinion that it was a headache that I didn't want to try to push through the H4 transition |
[11:20] | Andrew Linden: | perhaps you would like to raise that issue again |
[11:20] | Simon Linden: | ok, sure. It would make the vehicle MOTOR stuff work better, but I'm sure it would break things |
[11:21] | Andrew Linden: | one argument for doing it now is that we probably won't have another chance to do it... H4 is going to break stuff, some subtle changes and other outright breakage |
[11:21] | Andrew Linden: | so... if we're going to break stuff... why not get this enhancement in at the same time |
[11:22] | Andrew Linden: | perhaps some of the breakage will look like just shear H4 incompatibility |
[11:22] | Andrew Linden: | hrm... I think maybe the batteries in my wireless keyboard are finally failing |
[11:22] | Andrew Linden: | as I type characters get lost |
[11:23] | Simon Linden: | The issue is that the vehicle MOTOR settings have a speed limit of 40m/sec, and we see developers getting by this via additional impulses |
[11:23] | Gaius Goodliffe: | The additional impulse trick will still work, right? |
[11:23] | Gaius Goodliffe: | Just be unnecessary... |
[11:23] | Simon Linden: | Right, but if we raise the speed limit on the motor, the older vehicles will go even faster :) |
[11:24] | Andrew Linden: | yes... but some vehilce scriptors have set the motor larger than 40... and it gets silently clamped to 40 |
[11:24] | Gaius Goodliffe: | It doesn't seem likely to me that it would break much. Scripts will suddenly get exactly the speed they asked for. Which sounds like a fix to me. There may be scripts that asked for more than they wanted but settled on what they got, I suppose, but how common is that really? |
[11:24] | Andrew Linden: | we don't know how common it is |
[11:24] | Gaius Goodliffe: | I think this may be something that, like Andrew said last time, we just need to test and see if it happens or not. |
[11:25] | Andrew Linden: | but when one in a million things break, we usually hear about it |
[11:25] | Arawn Spitteler imagines some scripter wanted 50, got 40, and tweaked for 40, making 50 broken | |
[11:25] | Gaius Goodliffe: | My prediction is you won't see much of anything, but predictions don't mean much without empirical testing... |
[11:25] | Simon Linden: | This does seem to be the right time to try it and see what types of problems ensue |
[11:25] | Simon Linden: | So, and I know this is opening Pandora's box, what's a more practical speed limit? |
[11:25] | Gaius Goodliffe: | 3.0 * 10^8 m/s? :D |
[11:25] | Teravus Ousley: | hehe |
[11:26] | Rex Cronon: | how fast can the fastest car go in rl? |
[11:26] | Gaius Goodliffe: | But I'm guessing we're a ways away from simulating relativistic effects... |
[11:26] | Simon Linden: | Gaius - yeah, we'd just have to start ramping up your mass when you got too fast ... |
[11:27] | Gaius Goodliffe: | hehe |
[11:27] | Gaius Goodliffe: | Play with the TD |
[11:27] | Gaius Goodliffe: | I think it needs to be a lot faster than the fastest car... |
[11:27] | Andrew Linden: | one reason the vehicle limit was kept low in H1 was the "deep think" problem. Slower vehicles would be less likely to be crashed into a state of penetration that would kill the simulator. |
[11:27] | Rex Cronon: | no, if u want faster, u add a rocket:) |
[11:27] | Andrew Linden: | That problem is aleviated in H4. |
[11:28] | Gaius Goodliffe: | One wants to be able to shoot items high into the air quickly. |
[11:28] | Gaius Goodliffe: | Ballistic teleports. ;) |
[11:28] | Ryozu Kojima: | Land speed record is 341.111...m/sec |
[11:28] | Gaius Goodliffe: | (aka "Gnome flingers"?) |
[11:29] | Teravus Ousley: | heh |
[11:29] | Simon Linden: | The high speeds aren't going to be very usable, due to sim crossings and asset loading delays, but it seems like this should be high enough to allow usage with cars and planes |
[11:30] | Gaius Goodliffe: | sim crossings aren't an issue for vertical movement. |
[11:30] | Ryozu Kojima: | I'd say 256m/sec or whatever the avatar cap is |
[11:30] | Gaius Goodliffe: | Being able to rocket to high altitude quickly would be nice. |
[11:30] | Ryozu Kojima: | That 341m/sec was done in a car powered by 2 jet engines ;) |
[11:31] | Simon Linden: | Yeah, it probably had too many prims :) |
[11:31] | Andrew Linden: | let's take a quick non-binding vote... if you think we should increase the vehicle speed limit now then say "YES", if not say "NO" |
[11:31] | Rex Cronon: | in rl there is a spherical metal cage that u can drive bikes, even cars, on the inside. can u do that here? |
[11:31] | Gaius Goodliffe: | What's Mach 1 in m/s? |
[11:31] | Ryozu Kojima: | Sure |
[11:31] | Renee Faulds: | yes |
[11:31] | Rex Cronon: | yes |
[11:31] | Darek Deluca: | Yes |
[11:31] | Ryozu Kojima: | Rex: Probably, if your vehicle is designed for it |
[11:31] | Arawn Spitteler: | Yes, Mach 1.0 sounds like a good start |
[11:32] | Ryozu Kojima: | I tried with a Dominus and a mega prim, didn't work too well though ;) |
[11:32] | Gaius Goodliffe: | Then you can break the sound barrier in H5. :) |
[11:32] | Renee Faulds: | :) |
[11:32] | Arawn Spitteler: | In SL, you wouldnt drive about, in a cage, but drive the cage about, and animate the textures to stand still |
[11:32] | Ryozu Kojima: | hah |
[11:33] | Simon Linden: | OK, sounds like agreement. I'll try 256 and we can see what blows up |
[11:33] | Rex Cronon: | can't h4 simulate that? |
[11:33] | Ryozu Kojima: | I think the biggest problem Rex was the fact that the vehicle didn't have enough centripital(sp) force to "stick" to the cage, and the undercarriage kept getting caught up |
[11:33] | Andrew Linden: | btw, no one has said anything about it, and I haven't really tested it in a while, but in H4 you should be able to stand on a large moving platform a little better. |
[11:33] | Teravus Ousley: | :D |
[11:33] | Andrew Linden: | has anyone noticed? |
[11:34] | Darek Deluca: | how large? |
[11:34] | Ryozu Kojima: | Also doesn't help that most vehicles in SL auto-correct for being upside down |
[11:34] | Andrew Linden: | it has to be larger than your avatar I think |
[11:34] | Andrew Linden: | or maybe it is twice the mass, can't remember |
[11:34] | Ryozu Kojima: | Andrew: I usually only get carried around in a UFO x_x |
[11:34] | Gaius Goodliffe: | Currently, trying to stand on my elevator while it goes down works in H1 but not really in H4... |
[11:34] | Ryozu Kojima: | Speaking of which |
[11:34] | Rex Cronon: | would be interested to do a loop with a car:) |
[11:34] | Rex Cronon: | interesting* |
[11:35] | Darek Deluca: | Will it not trigger the 'walk' anymore? I have that problem with my physical elevators |
[11:35] | Andrew Linden: | how does the elevator fail in H4 Gaius? |
[11:35] | Ryozu Kojima: | I don't know if this is a physics engine thing, but.. Andrew, when you stand on a moving platform or vehicle, it carries you with it but triggers walking animations. Is that something you could do something about? Maybe in the long run? |
[11:35] | Gaius Goodliffe: | You fall, brush yourself off, fall again, etc. all the way down. In H1 you just stand on it and "walk in place" as you go down. |
[11:36] | Andrew Linden: | I thought that "walk animation while on moving object" problem was not so bad in Havok4 |
[11:36] | Andrew Linden: | however, as I said, I haven't tested it in a while |
[11:36] | Gaius Goodliffe: | It's not a problem with the animation. |
[11:36] | Gaius Goodliffe: | You drop, are stationary for a second, fall again, are stationary, etc. |
[11:37] | Ryozu Kojima: | Well, the animation kidn of detracts from the effect |
[11:37] | Andrew Linden: | Gaius, give me a copy of the elevator sometime and I'll see if I can fix it. |
[11:37] | Gaius Goodliffe: | In H1 you move with the platform. In H4 you just kinda freeze in place, then freefall, then freeze in place. |
[11:37] | Andrew Linden: | How does the elevator move? |
[11:38] | Gaius Goodliffe: | A series of llMoveToTargets slowly spaced out over the distance. |
[11:38] | Andrew Linden: | ok |
[11:38] | Gaius Goodliffe: | It also overshoots at the end. |
[11:38] | Ryozu Kojima: | Overshoots? |
[11:38] | Gaius Goodliffe: | It goes past where it's supposed to stop, the turns around and come back. |
[11:38] | Ryozu Kojima: | On the way down? |
[11:38] | Gaius Goodliffe: | Both ways. |
[11:39] | Andrew Linden: | Ok, one item I wanted to talk about was something about the avatar motion... |
[11:40] | Andrew Linden: | there are a number of FPS (First Person Shooters) players at LL |
[11:40] | Andrew Linden: | and they often complain about the avatar motion and how it needs client side prediction |
[11:40] | Ryozu Kojima: | meh |
[11:40] | Rex Cronon: | hmm? |
[11:41] | Teravus Ousley finds that interesting | |
[11:41] | Andrew Linden: | what they don't realized, is that most of the avatar's clunkyness is designed in... the slow-to-get going effect and the slow-to-stop |
[11:41] | Kitto Flora: | Is there not too much (bad) client side prediction ? |
[11:41] | Andrew Linden: | that is, the server-side-only design is not the main limiting factor in the responsiveness of the avatar's motion |
[11:42] | Andrew Linden: | the rubber banding can be caused by server lag |
[11:42] | Andrew Linden: | and client-side prediction might help reduce that problem |
[11:42] | Andrew Linden: | but on a responsive simulator, you could theoretically achieve near FPS response |
[11:43] | Andrew Linden: | if the avatar motion were designed right |
[11:43] | Kitto Flora: | It appears that the client is predisting wrongly when the server updates(?) lag |
[11:43] | Gaius Goodliffe: | It does seem that the client-side prediction is pretty messed up... |
[11:43] | Renee Faulds: | how much more resources would this force to the client side computer system |
[11:43] | Andrew Linden: | one note about FPS's... you can typically run at 50 mph in an FPS. Anything slower is not fun. |
[11:44] | Renee Faulds: | a majority of residenent computers barely meet requirments now |
[11:44] | Ryozu Kojima: | I vote for keeping the client as dumb as possible |
[11:44] | Rex Cronon: | how, can u run at 50mph, when even a car doesn't go yet above 40? |
[11:44] | Saijanai Kuhn: | eh. IN the long run, this is an issue for the AWG, but would require tunable sims, I think |
[11:44] | Ryozu Kojima: | Rex: FPS games are all about the fast paced action is all |
[11:45] | Andrew Linden: | well, the rubber banding could be reduced by a smarter client/server protocol |
[11:45] | Rex Cronon: | i like playing them too:) |
[11:45] | Renee Faulds: | this is true Andrew |
[11:45] | Ryozu Kojima: | Looking at this from an open grid point of view, from the physics engine aspect, all you really need to do is make sure the client is supplied sufficent data. I imagine a lot of 3rd party clients popping up |
[11:45] | Gaius Goodliffe: | My big issue with client-side prediction is that it tried to be too clever and almost invariably gets it wrong. Rather than just keeping an object moving in the same direction at the same speed, it makes you drop in an arc like it's expecting your lift to have suddenly disappeared, or worse, on the sea, the ocean suddenly disappear or something. It's bizarrely using a more complicated algoritm to simulate unlikely events it seems. |
[11:46] | Ryozu Kojima: | And I don't see all of them wanting or caring about client side avatar movement prediction or whatnot |
[11:46] | Andrew Linden: | well ok. I got into the "we need client side prediction" discussion again this morning before this hour |
[11:46] | Saijanai Kuhn: | This is more an AWG question for long-term |
[11:46] | Andrew Linden: | so I figured I'd mention some of the issues around the SL avatar controller here |
[11:46] | Simon Linden: | I've always thought the client prediction code needs some basic checks - it will take you off a long ways with whatever the last velocity/direction is. It would be better to just stop than have the AV walking through mountains |
[11:47] | Rex Cronon: | can't the client do some simle ray casting, at least on objects bounding boxes? |
[11:47] | Renee Faulds: | my only concern would be the addition load on client machines |
[11:47] | Andrew Linden: | if we were to make the avatar move like an FPS then I'm sure I'd get all sorts of complaints about broken content... scripts and in-world obstacles |
[11:47] | Gaius Goodliffe: | Just stopping would be less jarring than sinking into the sea... |
[11:47] | Ryozu Kojima: | I'd just take out the whole gravity prediction is all |
[11:47] | Teravus Ousley: | heh |
[11:47] | Saijanai Kuhn: | texture drawing isliey the main cause of client-side lag |
[11:47] | Saijanai Kuhn: | is likely* |
[11:47] | Renee Faulds: | plus Andrew makes an excelent point on scripting |
[11:47] | Teravus Ousley: | I don't think there's graviti prediction.. it's probably velocity interpolation based on a vector on the simulator |
[11:48] | Arawn Spitteler: | A couple weeks ago, I lag-walked a long way out of sim. |
[11:48] | Saijanai Kuhn: | wear Qarl's ant. YOu can fly almost to the next sim |
[11:48] | Gaius Goodliffe: | It has to be gravity -- it doesn't take you in a straight line, you follow a parabolic trajectory until you reestablish contact with the sim. |
[11:49] | Rex Cronon: | it can't be gravity, if it was u would be accelerating |
[11:49] | Gaius Goodliffe: | You DO accelerate! |
[11:49] | Andrew Linden: | in H1 whenever we ask an object for its acceleration, we always get -9.8... even when the object is tumbling around on the ground |
[11:49] | Ryozu Kojima: | Parabolic curve ;) |
[11:49] | Gaius Goodliffe: | You fall faster and faster until the lag clears. |
[11:50] | Andrew Linden: | that is because the collisions don't count in the acceleration integration step |
[11:50] | Teravus Ousley: | hmm.. at least in OpenSimulator, there doesn't appear to be gravity prediction on the client. It's probably the acceleration interpolation! |
[11:50] | Ryozu Kojima: | It's discombobulating when you're in a vehicle |
[11:50] | Teravus Ousley: | :D |
[11:50] | Teravus Ousley: | there's that' too |
[11:50] | Andrew Linden: | in Havok4 they removed the RigidBody::getAcceleration() method |
[11:50] | Andrew Linden: | so we now compute our own acceleration, and that is what is supplied for interpolation on the client side |
[11:51] | Andrew Linden: | measuring acceleration is kinda tough, especially when the object is tumbling. And there is more than one way to measure it |
[11:51] | Andrew Linden: | I think I now measure it in two ways, and then pick one, or take the average, and sometimes discard them altogether |
[11:52] | Andrew Linden: | the logic is complicated, and it has been a while since I've looked at it |
[11:52] | Gaius Goodliffe: | All I know is, as soon as lag hits, your airship or sea ship starts sinking, even if it wasn't before. The "interpolation" includes gravity but not buoyancy it seems. |
[11:52] | Andrew Linden: | this is also in Havok4 Gaius? |
[11:52] | Gaius Goodliffe: | Actually, I'm not sure... I don't recall seeing any lag-behavior one way or the other. I'll keep an eye out. |
[11:53] | Andrew Linden: | yeah, I'd be interested to find out what you notice. |
[11:53] | Andrew Linden: | That's all I had to talk about... the table is open to any issues or questions you may have. |
[11:53] | Rex Cronon: | i do have something |
[11:53] | Gaius Goodliffe: | When do we see the new goodies we got a preview of on Thursday? |
[11:54] | Gaius Goodliffe: | (And by goodies I mean lucious bug-fixes. :) |
[11:54] | Arawn Spitteler: | I was just wearing my rail-rocket in Mauve Beta, and I flew upwards. |
[11:54] | Andrew Linden: | Very soon Gaius. That is the next update currently in QA I was talking about. |
[11:54] | Rex Cronon: | if i am killed in h4, and i sometimes crash |
[11:54] | Kitto Flora: | Me either... If the rubberbanding may have changed its worth taking a look to see how it is now |
[11:54] | Andrew Linden: | Arawn, sorry could you be more clear about what the problem was? |
[11:55] | Andrew Linden: | You weren't supposed to fly upwards? |
[11:55] | Kitto Flora: | Did the llPushObject() problem come up? |
[11:55] | Andrew Linden: | Rex, your client crashes when you die in Havok4? |
[11:55] | Arawn Spitteler: | It's not supposed to apply a force, but when I want it to. |
[11:55] | Rex Cronon: | sometimes |
[11:55] | Arawn Spitteler: | I think it's using llApplyImpulse |
[11:56] | Andrew Linden: | the Havok4 client is a bit behind the H4 simulator, I think |
[11:56] | Andrew Linden: | I wonder if it is caused by the mismatched state of the two, REx |
[11:56] | Rex Cronon: | possible |
[11:56] | Gaius Goodliffe: | It's now the same as the mono client, right? ADITI (or however it's spelled)? |
[11:57] | Andrew Linden: | Not quite, I think we were waiting for some UI update to be checked into the Havok4 branch for MONO support |
[11:57] | Darek Deluca: | Question: Are the remaining mega prims going to be supported. And should we report physics issues that are unique to megaprims and Havok4? |
[11:57] | Rex Cronon: | oh, andis kind of weird when u shower somebody with bullets and their av hardly moves |
[11:58] | Arawn Spitteler: | Behaviour isn't happening here, but happened after I had it returned to Lost and Found in Mocha Beta |
[11:58] | Andrew Linden: | Megaprims... not all megaprims are supported in Havok4. In particular megaprims with sides larger than 256 will have those sides clamped to 256. |
[11:58] | Andrew Linden: | There were some problems with linking some megaprims, we fixed some of those issues, but don't know if they are all fixed. |
[11:58] | Gaius Goodliffe: | Rex: That might just be realism. The amount of force of a bullet hitting someone is precisely equal to the kick of the gun that fired it. Hollywood likes to overplay it. |
[11:59] | Andrew Linden: | Otherwise... I'm not aware of any megprim problems. |
[11:59] | Renee Faulds: | is there any work being done on asset storage on client machines at least for a backup client /server protocol crosscheck? |
[11:59] | Rex Cronon: | i am shooting 1/2m spherical bullets:) |
[11:59] | Gaius Goodliffe: | Ah! Heheh |
[11:59] | Kitto Flora: | Did the llPushObject() problem come up? |
[12:00] | Darek Deluca: | i am not sure it is the megaprims, but i keep having physical objects imbed themselves in non-physical objects, not colliding and rebounding. |
[12:00] | Andrew Linden: | Rex, the avatar not moving from bullets is caused mostly by the fact that in Havok4 the avatar tries harder to move as you direct. So if you are "standing" it "stands a little harder" than it used to. |
[12:00] | Gaius Goodliffe: | hehe |
[12:00] | Gaius Goodliffe stands real hard. | |
[12:01] | Andrew Linden: | Simon, you think maybe the embedded objects is caused by the RCCS kicking in? |
[12:01] | Rex Cronon: | and shouldn't get damaged, by moving physical objects, even if their damage function was never called? |
[12:01] | Andrew Linden: | Darek, does this happen after extreme load on the system? |
[12:02] | Darek Deluca: | no, i don't think so, the load on my sim is always quite low |
[12:02] | Simon Linden: | Could be the RCCS - if the sim is getting bogged down. |
[12:02] | Darek Deluca: | My performance is always over 4500 using conovers performance meter. |
[12:02] | Andrew Linden: | RCCS = runtime collision control system |
[12:03] | Simon Linden: | Darek, what are the objects that become stuck? |
[12:03] | Andrew Linden: | Darek, what is the region name? |
[12:04] | Darek Deluca: | Xidd, is the region. I have physical pinballs on one level constantly moving and being kept moving by cannons. The pinballs sometimes get embedded in the walls (256x256 prims). |
[12:04] | Kitto Flora: | Did the llPushObject() problem come up? |
[12:04] | Andrew Linden: | Darek, are the walls made of concave objects? Hollow or otherwise? |
[12:05] | Rex Cronon: | what is the problem with that funciton kitto? |
[12:05] | Darek Deluca: | they are 256x256x1 prims, is that a hollow, path cut 256x256x256 prim? |
[12:06] | Andrew Linden: | concave objects use Havok's MeshShape, which is a collection of triangles |
[12:06] | Darek Deluca: | they are boxes though |
[12:06] | Kitto Flora: | llPushObject() By an object on an Av when they stand up - used to eject Avs from vehicles. It is failing totally in H4. Results in broken content on MG now. |
[12:06] | Andrew Linden: | if you get inside a large enough mesh shape then you don't collide... |
[12:06] | Kitto Flora: | I have enteres it in jira as a bug. |
[12:06] | Andrew Linden: | unless you try to get out, in which case you collide with the inside walls of the MeshShape |
[12:06] | Renee Faulds: | I have also noticed that prim alighnment on H4 is still off |
[12:07] | Ryozu Kojima: | Haha, Yeah, I run into that one sometimes |
[12:07] | Simon Linden: | I've got to run ... thanks everyone. bye! |
[12:07] | Rex Cronon: | bye simon |
[12:07] | Kitto Flora: | Bye Simon? |
[12:07] | Davie Zinner: | thanks Simon |
[12:07] | Andrew Linden: | That reminds me Kitto, there is new code in the next update for improved dismount of vehicles |
[12:07] | Kitto Flora: | Is it possible for a h4 early adopetr to get rolled back to H1? |
[12:07] | Rex Cronon: | kitto u could make vehicle phantom than use push |
[12:08] | Kitto Flora: | Rex - thats what I do |
[12:08] | Gaius Goodliffe: | llUnSitTarget? :D (I know, I know, not yet...) |
[12:08] | Andrew Linden: | the simulator will now try a little harder to find a location where your avatar is not penetrating anything, but are standing on something |
[12:08] | Kitto Flora: | The phantom works |
[12:08] | Kitto Flora: | The push no longer works |
[12:08] | Kitto Flora: | It works right in H1, maybe used to work right in H4 |
[12:08] | Andrew Linden: | Yes Kitto, all a H4 early adopter needs to do is to talk to Sidewinder and ask to be rolled back to H1 |
[12:08] | Kitto Flora: | When will that update be on Preview, Andrew? |
[12:08] | Andrew Linden: | very soon Kitto... today or tomorrow |
[12:09] | Kitto Flora: | OK |
[12:09] | Kitto Flora: | I will see what it does for the problem |
[12:09] | Renee Faulds: | I have also noticed that prim alighnment on H4 is still off |
[12:09] | Kitto Flora: | But I suspect all trampolines |
[12:09] | Kitto Flora: | Bouncie castle |
[12:09] | Kitto Flora: | all broken now |
[12:09] | Rex Cronon: | trampoines no longer work? |
[12:10] | Kitto Flora: | I dont have one to test |
[12:10] | Darek Deluca: | mine seems to work |
[12:10] | Kitto Flora: | But the script and requrements are the same |
[12:10] | Kitto Flora: | No bouncie |
[12:11] | Gaius Goodliffe: | THis one works... |
[12:11] | Andrew Linden: | ok Kitto, I'll keep an eye out for bug reports about that. |
[12:11] | Kitto Flora: | Its on Jira |
[12:11] | Andrew Linden: | I've got to go soon. |
[12:12] | Kitto Flora: | I dont have the number with me tho |
[12:12] | Gaius Goodliffe: | This one works a bit too well... |
[12:12] | Renee Faulds: | is there work being down to make wall prims line up better |
[12:12] | Andrew Linden: | I'm trying to make more passes of the public jira list. I think there are a number of bugs that are fixed but not closed yet. |
[12:12] | Rex Cronon: | /tp p 10000 |
[12:12] | Teravus Ousley: | heh |
[12:12] | Andrew Linden: | After this next update goes out I'll make another pass. |
[12:12] | Kitto Flora: | Well that is interesting |
[12:13] | Renee Faulds: | quite a trampoline huh Kitto |
[12:13] | Kitto Flora: | Yes, I've seen as bad :) |
[12:13] | Teravus Ousley: | heh, trampo-orbit :D |
[12:14] | Gaius Goodliffe: | New satellite launch system lol |
[12:14] | Kitto Flora: | zI can see the code in that |
[12:14] | Renee Faulds gave you Angels Trampoline. | |
[12:14] | Kitto Flora: | I wonder if its a magnitude problem |
[12:14] | Renee Faulds: | Andrew I still notice that my wall prims are not all the same sign and that butting edges still show a seam |
[12:15] | Andrew Linden: | Renee, not the same size? |
[12:15] | Rex Cronon: | this is weird |
[12:15] | Rex Cronon: | is used to work |
[12:15] | Renee Faulds: | yes same size according to edit but like a 25mm diff visually |
[12:15] | Kitto Flora: | Rex, does that one use a fixed number for the push? |
[12:15] | Rex Cronon: | yes |
[12:15] | Kitto Flora: | So does my vehicles |
[12:15] | Renee Faulds: | maybe 10mm |
[12:15] | Kitto Flora: | But the one that works here - its calculationg a number |
[12:15] | Rex Cronon: | i can set it how hard to push |
[12:16] | Kitto Flora: | Crank it up 100 |
[12:16] | Gaius Goodliffe: | It pushes it you bump the side! |
[12:16] | Rex Cronon: | is at 10000 |
[12:16] | Kitto Flora: | Erk |
[12:16] | Gaius Goodliffe: | lol |
[12:17] | Gaius Goodliffe: | I'm stuck! :: |
[12:17] | Gaius Goodliffe: | :o |
[12:17] | Kitto Flora: | Its pushing on the wrong vector |
[12:17] | Rex Cronon: | this is more like a repeller |
[12:17] | Rex Cronon: | but it worked on h1 |
[12:17] | Gaius Goodliffe: | Okay, it seems avatars interpenetrating objects don't get pushed out.. |
[12:17] | Kitto Flora: | Gais - no - they get trapped |
[12:17] | Gaius Goodliffe: | :o |
[12:18] | Andrew Linden: | They do Gaius, unless you're in a MeshShape (concave object) |
[12:18] | Gaius Goodliffe: | This doesn't look concave... |
[12:18] | Kitto Flora: | Would that be a like - a hollow vehicle? |
[12:18] | Gaius Goodliffe: | I guess it is. It's a hollow box... |
[12:18] | Andrew Linden: | anything hollow, or with a carve out side |
[12:18] | Kitto Flora: | Well therein is the problem |
[12:19] | Gaius Goodliffe: | Shouldn't I be shoved into the hollow area, though? |
[12:19] | Andrew Linden: | no, for mesh shapes you only collide with the inside and outside surfaces |
[12:19] | Kitto Flora: | I am trapped |
[12:20] | Kitto Flora: | Help! Help! |
[12:20] | Andrew Linden: | however, in this case you seem to be embedded... snagged |
[12:20] | Teravus Ousley: | heh |
[12:20] | Gaius Goodliffe: | The fountain just loves me too much to let go. It's understandable... |
[12:20] | Rex Cronon: | we have canibal cars:) |
[12:20] | Kitto Flora: | Oooh |
[12:20] | Kitto Flora: | Great Griefing device!!!!! |
[12:20] | Andrew Linden: | The omniphaze should be able to get you out |
[12:21] | Teravus Ousley: | It seems to use the opposite of the velocity |
[12:21] | Kitto Flora: | Just shove a hollow prim around an av - TRAPPED! |
[12:21] | Andrew Linden: | I'm currently trying to code up a fix for that problem. |
[12:21] | Kitto Flora: | Good on yer Andrew :) |
[12:21] | Gaius Goodliffe: | Now, that was cool... |
[12:21] | Arawn Spitteler will have to show Herm, before it's fixed. |