User:Andrew Linden/Office Hours/2008 05 10

Transcript of Andrew Linden's office hours:

[11:01]	VendorKing Held gave you Advantage Vendor V_RJ2 (split 50%).
[11:02]	JayR Cela:	hi there Andrew
[11:02]	Andrew Linden:	Hi JayR
[11:02]	Andrew Linden:	What is "Team SDN"?
[11:03]	Andrew Linden:	Hello Gaius
[11:03]	Gaius Goodliffe:	Hello!
[11:03]	JayR Cela:	i got ta say the havok 4 thing works very well
[11:04]	Andrew Linden:	Really? Much of whal we do at my office hours is talk about the various Havok4 bugs.
[11:04]	JayR Cela:	i not really finding many buggs
[11:05]	JayR Cela:	oh
[11:05]	JayR Cela:	the dissapearing skirt bugg
[11:05]	Gaius Goodliffe:	Speak of "is it a bug or a feature"...
[11:05]	Gaius Goodliffe:	*speaking of...
[11:05]	Andrew Linden:	Yeah well... that isn't actually a "Havok4 bug"
[11:05]	Rex Cronon:	hi everybody
[11:05]	JayR Cela:	Nicholaz Bersoferd fixed / whay cant you
[11:05]	Andrew Linden:	Hi Rex
[11:05]	JayR Cela:	Hi Rex
[11:05]	Gaius Goodliffe:	Are you guys aware that the constraints that used to stop people from making new megaprims are gone?
[11:06]	Rex Cronon:	hiii..
[11:06]	Phli Foxchase:	Hello :)
[11:06]	Andrew Linden:	Yes Gaius, I heard about that. Kelly Linden fixed it -- should go out in the next server update.
[11:06]	Gaius Goodliffe:	Ah okay. :)
[11:07]	Phli Foxchase:	so, it's an exploit ?
[11:07]	Andrew Linden:	Yes, that 'feature" was really a bug and was unintended
[11:07]	Gaius Goodliffe:	Thought it might have been part of the first steps towards megaprim liberation.
[11:07]	Phli Foxchase:	I thought it was the beginning of Megaprim Liberation
[11:07]	Phli Foxchase:	: )
[11:07]	Gaius Goodliffe sighs.
[11:08]	JayR Cela:	i am just upset my skirts keep dissapearing
[11:08]	Andrew Linden:	Meanwhile, Qarl Linden is working on the true beginning of "Megaprim Liberation" -- being able to return objects that overlap onto your land.
[11:08]	Gaius Goodliffe:	Cool. I really need to stop by Qarl's office hours one of these days...
[11:09]	Andrew Linden:	Cela, I don't know much about the skirt bug except for the fact that it isn't related to the code I'm working on -- it is entirely client-side, which is why Nicholas was able to fix it in his client.
[11:09]	Andrew Linden:	Steve Linden would know more about the status of that bug.
[11:09]	Cinthya Loveless:	does that mean we're going to be finaly free of the hassle of mega prims ?
[11:10]	JayR Cela:	Andrew / OK / so why you not fix it
[11:10]	Phli Foxchase:	Brad Linden did a fixe for this bug, it is in QA
[11:10]	Andrew Linden:	Ah ok, if Brad has fixed it then it is just taking time for the fix to propagate through the QA/testing pipeline
[11:11]	Andrew Linden:	however... I wonder what branch he fixed it in
[11:11]	Gaius Goodliffe:	If by "hassle of megaprims" you mean the way neighbors can have huge objects overlap your property (whether they include any megaprims or not is irrelevant), then yes.
[11:11]	Andrew Linden:	if so, it should get released soon
[11:11]	Cinthya Loveless:	what is the "Megaprim Liberation" you speak of Andrew?
[11:11]	Wolfhaven Teleportation HUD v3.0 is Ready
[11:11]	Andrew Linden:	Megaprim Liberation is the project where we allow SL residents to create prims with scale components > 10 meters on a side
[11:12]	Alyx Stoklitsky:	We can already do that :P
[11:12]	Andrew Linden:	Many SL Residents want to be able to use megaprims, and there aren't too many technical reasons for preventing them... but there are a few
[11:12]	Gaius Goodliffe:	Not after the next server update, Alyx. :p
[11:12]	Cinthya Loveless:	yes rendering is one
[11:13]	Alyx Stoklitsky:	Good job I made every concievable size
[11:13]	Andrew Linden:	There is currently an exploit by which megaprims can be created again.
[11:13]	Gaius Goodliffe:	hehe
[11:13]	Andrew Linden:	However we'll be plugging that hole soon
[11:13]	Cinthya Loveless:	I've wished they would be totaly removed form the grid that would make my Day
[11:13]	JayR Cela:	Cin / i agree
[11:13]	Rex Cronon:	why is it considered an exploit?
[11:14]	Andrew Linden:	Here is the quick outline of the "Megaprim Liberation" project. I've mentioned it a few times on the forums and/or these office hours:
[11:14]	Kitto Flora:	Megaprims have some good uses
[11:14]	JayR Cela:	is not ready yet
[11:14]	Alyx Stoklitsky:	I want all the megaprims I can get. My build in smuttynose would be utterly impossible without them.
[11:14]	Gaius Goodliffe:	There's no reason a prim shouldn't be able to be made just as large as a linkset.
[11:14]	Andrew Linden:	Before we can allow anyone to create megaprims there are some preliminary projects/features that need to be done:
[11:14]	Cinthya Loveless:	mega prims hurt sim performance Alyx
[11:14]	Gaius Goodliffe:	Megaprims help sim performance.
[11:15]	Andrew Linden:	(1) land/parcel owners should be able to return objects that overlap on their parcels
[11:15]	Cinthya Loveless:	thats a great idea Andrew
[11:15]	Rex Cronon:	what good is a megaprim, if u can't even link two prims that are less than 10m apart:(
[11:15]	Andrew Linden:	(2) such return should also work for sculptie prims
[11:15]	Alyx Stoklitsky:	They only hurt physics performance if they're unnecessarily overlapping and unlinked
[11:15]	Cinthya Loveless:	and rendering and some other things Alyx
[11:16]	Andrew Linden:	(3) maybe... some content will be immune to overlap return ("Linden Content", and some estate owners have asked for the ability to disable overlap return on their estates/regions)
[11:16]	Rex Cronon:	rezzing 1000 toruses also hurts sims performance
[11:16]	Alyx Stoklitsky:	I see no rendering issues with megaprims. I don't call draw distance an 'issue'.
[11:16]	Gaius Goodliffe:	Indeed, having four prims lags more than having one prim. If you can replace for 10x10 prims with one 20x20 prim, you *improve* performance.
[11:16]	Cinthya Loveless:	you don't deal with the issues they casue every day alyx such as people like my self who own many sims and have many residnets
[11:16]	Andrew Linden:	(4) ... perhaps some new/extra limitations should be present for megaprims... such as max size of object that can be attached
[11:17]	Gaius Goodliffe:	*four 10x10
[11:17]	Cinthya Loveless:	thats a great idea Andrew as well
[11:17]	Andrew Linden:	Uh... my list of megaprim prerequisites has morphed over time.
[11:17]	Andrew Linden:	I don't think that list is the same as when I started
[11:17]	Gaius Goodliffe:	hehe
[11:17]	Rex Cronon:	why doesn't h4 allow 2 objects to be linked if the distance between them is less than equal to 10m?
[11:18]	Andrew Linden:	but anyway, I really want to allow residents to make megaprims where they make sense
[11:18]	Gaius Goodliffe:	Rex: I do that all the time.
[11:18]	Cinthya Loveless:	thats a Great Idea and would help all the way around Andrew
[11:18]	Rex Cronon:	did u hear that error?
[11:18]	Lillie Yifu:	that's nonsense cinth
[11:18]	Lillie Yifu:	total utter nonsense
[11:18]	Andrew Linden:	oh yeah... another project is (6) static objects collide across region boundaries -- that is necessary for over-the-border prim overlap to be properly fixed
[11:19]	Cinthya Loveless:	what is?
[11:19]	Rex Cronon:	"[11:18] Link failed -- Unable to link any pieces - pieces are too far apart."
[11:19]	Lillie Yifu:	mega prims hurting sim performance.
[11:19]	Rex Cronon:	the distance between the 2 cubes is only 8m
[11:19]	Andrew Linden:	Uh... Rex that sounds like a bug, and I'm surprised that it exists.
[11:20]	Gaius Goodliffe:	Hmmm
[11:20]	Gaius Goodliffe:	They're too small?
[11:20]	Cinthya Loveless:	they do Lillie, ytou wouldn't understand since you don't deal with the iissues every day such as peopel like me who own many sims and have to deal with all of the issues
[11:20]	Alyx Stoklitsky:	Then you have them only placeable by estate managers. I personally require the biggest prims possible. I have 512 sphere in the sim I manage that is vital to the nature of the build, not to mention a 1024x1024 , which I could certainly replace with a 63536. I will find a use for every tool I'm given.
[11:20]	Andrew Linden:	What is the repro recipe? Just two cubes placed some distance apart?
[11:20]	Andrew Linden:	What size are the cubes?
[11:20]	Rex Cronon:	the 2 red cubes
[11:20]	Andrew Linden:	The linkability depends very much on the cube size.
[11:20]	Andrew Linden looks for red cubes...
[11:21]	Lillie Yifu:	then you arean incompettent builder and sim owner sinthyia
[11:21]	Andrew Linden:	ah yes...
[11:21]	Lillie Yifu:	that's not our fault
[11:21]	Rex Cronon:	behind your back
[11:21]	Andrew Linden:	those two red cubes are too small to link at that distance
[11:21]	Gaius Goodliffe:	Cinthya: It is total nonsense, and I do have to deal...
[11:21]	Rex Cronon:	why does size matter when the distance is less than 10m?
[11:22]	Andrew Linden:	the linkability limit for two 0.5^3 cubes is... about 2 + 3*(0.8) meters
[11:22]	Cinthya Loveless:	when you own as many sims as i do lets see you say its nonsense when you have so many issues regaurding mega prims
[11:22]	Andrew Linden:	so... 4.5 meters or so?
[11:22]	Alyx Stoklitsky:	Cinthya: Perhaps you should talk to Intlibber Brautigan...
[11:22]	Kitto Flora:	Why is it so short, Andrew?
[11:22]	Cinthya Loveless:	anyways stop trying to argue with me thats not what this meeting is about
[11:23]	Rex Cronon:	4m apart and still no link allowed
[11:23]	Gaius Goodliffe:	Cinthya: If I ever *misidentify* the source of my problems, I'm sure I'll dig a little deeper rather than just blame the biggest thing I see on my screen.
[11:23]	Andrew Linden:	The reason for the linkabiity rules are...
[11:23]	Kitto Flora:	(.5M cubes link at 3M, not at 4M)
[11:24]	Cinthya Loveless:	im ignoring you so stop wasting time trying to convence me differently and start paying attention to the meeting
[11:24]	Rex Cronon:	if the "link rules" don't allow objects that are less than 10m appart to be linked, than the "link rules" have to be changed, as they don't make sense
[11:24]	Andrew Linden:	(1) we felt it necessary for *some* limit very early on -- mostly for the non-collision over region borders, but also for simplification of rendering/UI/genera-physics
[11:26]	Andrew Linden:	(2) the final linkability rules were constructed out of an aesthetic sense that the linkability distance should scale with the size of the parts... very tiny prims should not be linkable at distances much much larger than themselves and the scale in which they are experienced
[11:26]	Gaius Goodliffe:	I think (2) is a bit whacked TBH.
[11:27]	Andrew Linden:	perhaps
[11:27]	Gaius Goodliffe:	Not that the principle is necessarily wrong, but the implementation is quite a bit off.
[11:27]	Kitto Flora:	Where did (2) come from? That would prevent some useful constructs. And for 'Aesthetic' reasins? Who's?
[11:28]	Andrew Linden:	The aesthetic was mine and Philip's. The first linkability rules were constructed waaaaay back in 2001 or so
[11:28]	Kitto Flora:	I find it very odd that one can link 3 10M prims spanning 20M but not two small cubes only 4M apart.
[11:28]	Rex Cronon:	u r not the only one kitto
[11:29]	Kitto Flora:	Humm - I really need to explain things to Philip, again ;)
[11:29]	Andrew Linden:	well, I would find it odd that you could link two 0.01 m cubes 20 meters apart.
[11:29]	Lillie Yifu:	or you can link two cut prims that are far apart, and are seen as visually small. A totured torus can be 10x10x10 made very small, and then linked to another 30M away.
[11:29]	Kitto Flora:	Not me --- making a planetary system?
[11:30]	Lillie Yifu:	linked messages are a huge reason to want to link small prims that are distant, but need to be in communication.
[11:30]	Kitto Flora:	Yes - another good reason
[11:30]	Andrew Linden:	perhaps... building planetary systems was outside the scope of the project when the virtual world was being conceived
[11:30]	Andrew Linden:	although it is no longer outside the scope
[11:31]	Kitto Flora:	I've seen quite a few orrarys here
[11:31]	Lillie Yifu:	but linkage isn't just about visual, there are special properties of linked sets, and those properties are useful.
[11:31]	Andrew Linden:	yes, and those link properties didn't exist when we created the linkability rules
[11:31]	Andrew Linden:	I think the linkability rules will eventually be relaxed
[11:32]	Rex Cronon:	i really can't see the downfalls of objects of any size being able to be linked if they are <= 10m appart
[11:32]	Alyx Stoklitsky:	then perhaps things need to be changed. I see few reasons to limit what is possible in regards to building in SL.
[11:32]	Andrew Linden:	they must if SL is to remain competitive in the realm of virtual worlds
[11:32]	Lillie Yifu:	hiearchical linking is anotherthing, I am sure you know, that there is a great deal of call for.
[11:33]	Andrew Linden:	however, I'm not very interested in boosting the linkability distances right now... there are quite a few prerequisites to be done first
[11:33]	Alyx Stoklitsky:	I hear LL wants to force all the megaprims down to 256?
[11:33]	Andrew Linden:	for example... objects should collide over region boundaries
[11:33]	Andrew Linden:	Alyx, that 256 meter limit is no longer in vogue here at LL
[11:33]	Morgaine Dinova:	I remember a Google video by Philip and Cory, in which they both accepted that the decision to not provide hierarchical linking was a mistake.
[11:33]	Saijanai Kuhn:	ouch (pun not intended)
[11:33]	Arawn Spitteler:	bi-regional obhects?
[11:33]	Cinthya Loveless:	256 is sim sized
[11:34]	Kitto Flora:	Andrew - would cross-region boundar collisions also apply to Avs running into prims?
[11:34]	Andrew Linden:	actually Alyx, that was, "Andrew Linden wants to clamp all megaprims to 256 meters on a side"
[11:34]	Phli Foxchase:	^^
[11:34]	Teravus Ousley:	:D
[11:34]	Saijanai Kuhn:	so multi-sim objects are a serious consideraton? 2 or 3 sim-covering?
[11:34]	Andrew Linden:	Yes Kitto re avatars and collisions across region boundaries.
[11:34]	Arawn Spitteler:	I thought that was the hollow, for creating a horizon
[11:34]	Alyx Stoklitsky:	Ah, I see. I have many uses for prims far, far larger than 256
[11:34]	Andrew Linden:	no Saijanai... it is just that >256 megaprims are no still on probation
[11:34]	Saijanai Kuhn:	arcosanti here we come
[11:35]	Kitto Flora:	Andrew: That would be really good, I'm tured of crossing reg.boundary and getting trapped in a building thats slow in rezzing.
[11:35]	Teravus Ousley:	Yeah, that's a hard problem to solve when regions' influence spans 256m
[11:35]	Gaius Goodliffe:	Andrew, I remember there was talk about forcing them to be phantom, anything come of that idea?
[11:35]	Saijanai Kuhn:	in fact, wonder if solari is still alive. He would enjoy working with megaprims I think
[11:35]	Saijanai Kuhn:	Soleri?
[11:36]	Andrew Linden:	Not yet Gaius. Forcing phantom for very large megaprims was an idea.
[11:36]	Andrew Linden:	I haven't really been focusing on megaprims lately, but I would like to finish Megaprim Liberation before the end of 2008
[11:37]	Gaius Goodliffe:	Considering we're getting a fresh batch, migth be a good idea. :)
[11:37]	Kitto Flora:	Andrew, do you know anything of the 'cast shadow' property for prims?
[11:37]	Andrew Linden:	It is not yet clear to me how much work features will have to be done to make megaprims less griefy
[11:38]	Andrew Linden:	Kitto, one of the devs at LL is experimenting with shadow casting... Runitai Linden
[11:38]	Rex Cronon:	i hope that "megaprim liberation" will not impose too many restrictions
[11:38]	Saijanai Kuhn:	there's a image foloating around
[11:38]	Teravus Ousley:	heh. it's computationally intensive to raycast for prims at borders :D
[11:38]	Andrew Linden:	He sent a few demo screnshots around a few months ago... haven't heard anything about it since.
[11:38]	Kitto Flora put Runitai on a golden pedestal.
[11:38]	Arawn Spitteler:	Some sort of Seniority License, would be appropriate. One little MegaPrim Construction Book could wreck routine havoc, at Help Island.
[11:39]	Teravus Ousley imagines that you could probably do that though... raycast along the bounary against the tri-mesh.
[11:39]	Teravus Ousley:	(parcel boundary)
[11:39]	Morgaine Dinova:	Andrew: done in shader hardware, hopefully. Shadow casting is pretty intensive.
[11:39]	Saijanai Kuhn:	http://picasaweb.google.com/Runitai/DeferredRendering/photo#5192943343115923442
[11:40]	Saijanai Kuhn:	shadowmapping, I believe
[11:40]	Andrew Linden:	Don't know much about how it is being done, however one reason why it probably has not been released yet is that it probably still eats up CPU resources.
[11:40]	JayR Cela:	ok / i seem to think that he video card driver problem is being addressed
[11:40]	Andrew Linden:	before I forget... I should make a few announcements...
[11:41]	JayR Cela:	the latest driver is not allways the best
[11:41]	Morgaine Dinova:	There are no spare CPU resources. They're all needed to run Vista ;-)
[11:41]	Saijanai Kuhn:	Direct shadowing (creating shadows from the sun) is done with hardware shadow maps, which complements SSAO very well but is a totally separate technique. That, too, is being researched internally with some good results, but it's too early to promise anything, so here's a picture you can take how you will:
[11:41]	Kitto Flora:	Shawdows is whats wrong with SL - visially. Chuck Windlight. Chuck voice. Gimme shadows :)
[11:41]	Andrew Linden:	The main thing is that LL's Havok4 team is going to break up within a week or two
[11:41]	Saijanai Kuhn:	/snif
[11:41]	Lillie Yifu:	They've doen a really great job
[11:41]	Andrew Linden:	Kelly and Simon Linden are going to move on to other projects soon.
[11:41]	Andrew Linden:	I think Sidewinder and Simon are going to work on some server optimization stuff.
[11:42]	Kitto Flora:	Whill physics load ever get fixed?
[11:42]	Andrew Linden:	Meanwhile, I'll still be fixing misc bugs for another several weeks.
[11:42]	Andrew Linden:	And I will still hold office hours, twice a week for now.
[11:42]	Andrew Linden:	Topic of conversation is "Whatever"
[11:43]	Morgaine Dinova:	lol
[11:43]	Lillie Yifu:	kk I have one
[11:43]	Alyx Stoklitsky:	Cool, fix the local rotation bug.
[11:43]	Arawn Spitteler:	Have others noticed that odd drifing stance, we sometimes take?
[11:43]	Sorrelle Poutine:	what is the local rotation bug? that might be the one i'm here about
[11:43]	Andrew Linden:	Kitto, the "physics load" problem I think would fall under Simon's optimization/lag-reduction project
[11:43]	Lillie Yifu:	would it be possible to spin out the characteristics of physical materials on prims and set them exactly, rather than having things like "wood" "metal" etc?
[11:43]	Rex Cronon:	and "sliding down the hill" is still here
[11:43]	Sorrelle Poutine:	and the drifting of objects once they collide
[11:44]	Lillie Yifu:	I should try and recreate my permanent temporary prim bug for Sidewinder if we are looking at little things.
[11:44]	Gaius Goodliffe:	And large vehicles being complete stopped dead in their tracks my a nearby hovering avatar.
[11:44]	Gaius Goodliffe:	*by a
[11:44]	Andrew Linden:	Sidewinder and Simon haven't said anything publicly about their project yet, but it is really just a spinoff of some Havok4 bug fixing that has picked up some other lag hotspots.
[11:44]	Kitto Flora wonders if sliding down the hill is linked to the discovered drifting down while flying bug
[11:45]	Saijanai Kuhn:	did anyone ever figure out why Qarl's ant lags so much?
[11:45]	Teravus Ousley:	hehe, Gaius, Opensimulator has that bug too :D
[11:45]	Andrew Linden:	Gaius, I think I may have fixed that bug about hovering avatar's stopping dynamic objects.
[11:45]	Gaius Goodliffe:	Ah, we've devolving into OpenSim? lol
[11:45]	Andrew Linden:	however, it won't be in this week's update, or next week's
[11:45]	Gaius Goodliffe:	As long as it gets there eventually. :)
[11:45]	Yuu Nakamichi:	any news on the cyclic TD, especally apparent in openspace regions?
[11:45]	Andrew Linden:	it just missed the deadline for our release pipeline codefreez
[11:46]	Saijanai Kuhn:	Andrew, I don't know if this is a commonly known thing, but my avie has ended up inside a hollow sphere before. In the volume, not the hollow part
[11:46]	Andrew Linden:	Yuu, Simon is working on the cyclic low time-dilation problem.
[11:46]	Yuu Nakamichi:	Is Simon goign to give us an update?
[11:46]	Alyx Stoklitsky:	happens to me a lot, Saijanai
[11:46]	Gaius Goodliffe:	I've found it's a lot easier these days to end up inside prims.
[11:46]	Andrew Linden:	That falls under his optimization/anti-lag work
[11:46]	Yuu Nakamichi:	ok
[11:47]	Kitto Flora:	More power to Simon! :)
[11:47]	Yuu Nakamichi:	Hi Sidewinder
[11:47]	Andrew Linden:	Eventually Yuu, but the Havok4 team has not yet fallen apart. Perhaps next week or two.
[11:47]	Teravus Ousley:	heh, Sajanai, face normals FTW!11!11!1!1 :D
[11:47]	Yuu Nakamichi:	ok great
[11:47]	Sidewinder Linden:	hi yuu
[11:47]	Saijanai Kuhn:	also, is there an intended difference in physics between the internal surface of a hollow object and the esxternal? Penetrations from inside to ouside are MUCH easier
[11:48]	Andrew Linden:	btw, if you ask a question and I don't 'answer it it may be that it got lost in the chat logs. Try re-asking...
[11:48]	Cinthya Loveless:	hey Sidewinder
[11:48]	Andrew Linden:	now about the avatar getting inside prims...
[11:48]	Sidewinder Linden:	/hi cinthya
[11:48]	Andrew Linden:	There are two things that make it easier for avatar's to get inside of prims...
[11:48]	Gaius Goodliffe:	I think Saijanai's observation there is related to mine from last time about getting easily stuck on the inside surface of hollowed cylinders and tubes and such.
[11:49]	Rex Cronon:	hi sidewinder
[11:49]	Sidewinder Linden:	/wow this connection is slow :p
[11:49]	Sidewinder Linden:	hi rex
[11:49]	Gaius Goodliffe:	Hi Side :)
[11:49]	Rex Cronon:	r u using a modem:)
[11:49]	Teravus Ousley:	I think I also suggested potentially raycasting along the parcel boundary against the faces of the prim to determine if it crosses a parcel boundary or not.
[11:49]	Andrew Linden:	(1) since the Havok4 penetration resolution code is not optimal for avatars, we currently disable collisions between the avatar and objects in which is is penetrating if the avatar is stuck long enough (> 1second) and deep enough
[11:49]	Elric Ember checks the new Avatar Rendering Cost feature
[11:50]	Saijanai Kuhn:	ah, OK
[11:50]	Morgaine Dinova:	It's the NSA deep packet inspection slowing everything down. They really need faster hardware.
[11:50]	Andrew Linden:	(2) hollow prims collide with Havok's MeshShape, which is really just triangle soup...
[11:50]	Andrew Linden:	but the triangles collide on their inside and outside surfaces
[11:50]	Teravus Ousley:	hehheh.
[11:50]	Andrew Linden:	so if you are completely inside a concave prim... Havok doesn't think you are penetrating at all
[11:51]	Andrew Linden:	unless you happen to overlap between the inside and outside
[11:51]	Andrew Linden:	so... you can walk inside hollow prims
[11:51]	Saijanai Kuhn:	hmmm... right. THing is sometimes I end up BETWEEN the inside and outside
[11:51]	Rex Cronon:	i think u can walk inside non-hollow prims too
[11:51]	Saijanai Kuhn:	in the shell...
[11:51]	Hulaboom Voom:	heya Kitto!
[11:51]	Teravus Ousley:	yeah, thats interesting having face normals pointing both ways
[11:51]	Kitto Flora:	Hi Hula
[11:51]	Andrew Linden:	I expect that inside collision property to eventually change... I have plans on replacing all MeshShape's with collections of convex pieces.
[11:52]	Andrew Linden:	But that project will take a while to get done.
[11:52]	Saijanai Kuhn:	KK. THis probably explains why followers inside a hollow sphere sometimes "slip out"
[11:52]	Hulaboom Voom:	(oops sorry... didint see all the other folks here ...)
[11:52]	Andrew Linden:	(Probably too late... people will probably build lots of content that relies on the MeshShape properties... sigh)
[11:52]	Sorrelle Poutine:	Andrew, i have a question about collisions between objects and resultant intersections of objects, although not people.
[11:53]	Andrew Linden:	Ask Sorrele
[11:53]	Teravus Ousley:	On the Megaprim tools I also suggested potentially raycasting along the parcel boundary against the faces of the prim to determine if it crosses a parcel boundary or not.
[11:53]	Sorrelle Poutine:	basically our script allows ppl to rotate objects (not in build mode) and they then intersect, get stuck, messed up, and eventually float away.
[11:53]	Kitto Flora:	I have what I believe is a general solution to the 'av gets trapped on stand up inside hollow vehicles' problem, if anyone is interested.
[11:53]	Andrew Linden:	We won't be using raycasting for prim-parcel collision queries on the server
[11:53]	Rex Cronon:	why do u need to raycast? when the bounding box is already there..
[11:53]	Sorrelle Poutine:	but they should be unable to intersect. in havok 1 they didn't intersect.
[11:53]	Gaius Goodliffe:	llUnSitTarget?
[11:54]	Kitto Flora:	Not Unsit target or llPushObject().
[11:54]	Andrew Linden:	Sorrelle, these are "dynamic" objects with the [x] Use physics checkbox enabled?
[11:54]	Gaius Goodliffe:	What's the solution, then?
[11:54]	Sorrelle Poutine:	correct... they're pieces of a puzzle that people can move around using their arrow keys
[11:54]	Sorrelle Poutine:	and they can be rotated 90 degrees at a time
[11:54]	Teravus Ousley:	the bounding box isn't a very accurate description of 'is this twistedt torus crossing my parcel; :D
[11:54]	Sorrelle Poutine:	and they can intersect into each other as well as other objects.
[11:54]	Teravus Ousley:	.. so that's why it was suggested.
[11:55]	Sorrelle Poutine:	and then they get stuck, although they try to wrioggle out (it's aliiiiive)
[11:55]	Andrew Linden:	and the problem is... they intersect in the physics engine but should not?
[11:55]	Andrew Linden:	Sorrelle, are you aware of the "collision tolerance" feature of the Havok physics engine?
[11:55]	Sorrelle Poutine:	correct. and once they collide and try to fix themselves after an intersection, they end up floating away
[11:55]	Rex Cronon:	than something wrong with either the bounding box, or the torus
[11:55]	Sorrelle Poutine:	Unfortunately, i'm not the one who programmed them, but I believe he has tried messing with that, yes...
[11:56]	Andrew Linden:	how big are the pieces Sorrelle?
[11:56]	Kitto Flora:	The problems with the Av is that they cannot get out of the hollow object. It traps them. Arrow keys are ineffectine.
[11:56]	Sorrelle Poutine:	as in how many prims? i think they're maybe 2 each
[11:56]	Andrew Linden:	Correct Kitto, I haven't found a good solution to the HollowCage problem yet.
[11:56]	Rex Cronon:	can u sit on them kitto?
[11:56]	Kitto Flora:	Andrew: I have :)
[11:56]	Andrew Linden:	No Sorrelle, how big as to size (meters across)
[11:56]	Gaius Goodliffe:	And it is...?
[11:57]	Arawn Spitteler:	Sitting just puts you back in the cage
[11:57]	Sorrelle Poutine:	normal behavior is that they would land in the puzzle board and stay there. and not be able to go into the walls or each other
[11:57]	Kitto Flora:	Rex: Trapped av and escape by sitting on another object. But thats ugly.
[11:57]	Morgaine Dinova:	On a slightly connected subject to boundary boxes/penetration ... will Havok4 eventually allow Argent Stonecutter's "Parcel Basements" to be implemented (ie. adding a parcel-shaped basementto every parcel, 1Km below zero altitude). Privacy is going to become very important in due course.
[11:57]	Sorrelle Poutine:	oh, maybe 1 meter. they're not very big
[11:57]	Teravus Ousley notes.. that simply cutting a cube to 3/4ths of a cube and then putting the open space over the edge of a parcel would be enough to put the bounding box over the parcel boundary.... however, no 'prim' is actually over the boundary
[11:57]	Sorrelle Poutine:	I'd have to go check
[11:58]	Sorrelle Poutine:	I have some movies but they're giant, but I can send them if you'd like
[11:58]	Andrew Linden:	well sorrelle... there were some slight differences between the collision shapes in going from Havok1 to Havok4
[11:58]	Andrew Linden:	and Havok4 still has a collision tolerance of 0.10 meters, so two dynamic objects will stack with a gap that big
[11:59]	Andrew Linden:	so dynamic pieces in a puzzle will also collide with a gap
[11:59]	Morgaine Dinova:	The Parcel Basements idea was rather stuck on the unknown question of whether Havok will still work at negative altitude.
[11:59]	Andrew Linden:	if you try to sqeeze them tighter then you will get "instabilities" and the objects will try to wiggle out
[12:00]	Andrew Linden:	for best results on dyamic puzzles or other dynamic mechanics...
[12:00]	Sorrelle Poutine:	well, this is even accounting for the gap though
[12:00]	Andrew Linden:	use as large of pieces as possible, and only use convex parts -- stay away from concave sub-prims
[12:00]	Saijanai Kuhn:	That would explain the behaviro of followers inside a hollow sphere I have seen. They bounce against the interior and often slip throuhg
[12:00]	Sorrelle Poutine:	we made the puzzle board bigger, for example...
[12:00]	Lillie Yifu:	So that means we still need to use phantom prims and sandwiching for things liek billiard's tables... hmmm
[12:01]	Andrew Linden:	Saijani, are any parts of your puzzle pieces concave?
[12:01]	Sorrelle Poutine:	and the pieces are just like an L or a square - don't believe they're concave or convex
[12:01]	Saijanai Kuhn:	not a puzzle.
[12:01]	Saijanai Kuhn:	that's Sorrelle's
[12:01]	Sorrelle Poutine:	logging on another av to see what size they are
[12:01]	Andrew Linden:	convex means no cavities in the shape
[12:01]	Saijanai Kuhn:	was just playing with H4 behavior and nested a follower sphere inside a hollow sphere and moved around
[12:01]	Andrew Linden:	an L shape is concave, but if made out of two boxes the pieces would be convex
[12:01]	Sorrelle Poutine:	Right, imagine those thought puzzles where you have 5 pieces and have to fit them into a square puzzle board
[12:02]	Saijanai Kuhn:	the follower would bounce through the hollow sphere on a regular basis
[12:02]	Rex Cronon:	a hollowed sphere is both convex and concave
[12:02]	Sorrelle Poutine:	iahhh, i see.
[12:02]	Andrew Linden:	No, a hollow spheres concave as far as Havok4 is concerned
[12:02]	Sorrelle Poutine:	well, we have a straight line, an L, a square, a short T, and an S type shape
[12:02]	Teravus Ousley:	hehe
[12:02]	Saijanai Kuhn:	so the force of the follower would force it past the collissin boundry and it would sometimes wriggle out
[12:03]	Sorrelle Poutine:	and they all do it, if that's any help
[12:03]	Rex Cronon:	how about if u cut it in hafl?
[12:03]	Andrew Linden:	concave prims use the MeshShape AND... in laggy situations they may be replaced with convex geometry to keep the sim from lagging too much
[12:03]	Lillie Yifu:	hmmm kk I knwo the bhavior you are talking about Saij... I saw it on a statute that rezzed small spehres ina a large one
[12:03]	Lillie Yifu:	some would leak...
[12:03]	Saijanai Kuhn:	happens quite offent if there's force ifnolved I guess.
[12:04]	Teravus Ousley:	Rex, if it isn't hollowed at all and not dimpled at all and not twisted, it's probably convex
[12:04]	Andrew Linden:	Sorrelle, I would recommend that you first make sure all sub-prims are convex. Use 4+ prims per object to do it if necessary
[12:04]	Teravus Ousley:	.. toruses are different though.. concave by default. :P
[12:05]	Gaius Goodliffe:	Convex prims don't hold water. ;)
[12:05]	Saijanai Kuhn:	I've seen inside out prims...
[12:05]	Sorrelle Poutine:	I'm pretty sure all the sub prims are just boxes. each one would be four squares if you were to count how much space they take (but I don't know if he programmed them to be 4 prims)
[12:05]	Sorrelle Poutine:	sorry, each puzzle piece.
[12:05]	Andrew Linden:	Well this office hour has just flown by. I'm going to have to run soon...
[12:05]	Teravus Ousley:	hehe, same here. Try Jhurlimans's test sculpties. Inside out.. the face normal points to the inside of the prim.
[12:05]	Rex Cronon:	i need a list that tells if a prim is concave, convex, or both
[12:05]	Lillie Yifu:	you can do it with sphere's too
[12:05]	Andrew Linden:	Re the question about basements under the land -- no that feature is not anywhere on the schedule yet.
[12:06]	Sorrelle Poutine:	the pieces also start drifting away once they encounter whatever is making them unstable
[12:06]	Saijanai Kuhn:	wondering if inside out spheres would allow better behavior when bouncing against the follower
[12:06]	Arawn Spitteler can see, why you'd want a cellar 50 meters deep, but a Kilometer is a bit much.
[12:06]	Saijanai Kuhn:	It makes for an amusign fake vehicle of you can keep the follower inside that sphere
[12:06]	Kitto Flora:	Drop the limit on prims below the height plane. then we can have tunnels!
[12:06]	Saijanai Kuhn:	independent thrusters so to speak
[12:07]	Morgaine Dinova:	Andrew: yes, I know, it never got that far. But it was the conclusion of a couple of years analysis of privacy solutions, and just needed knowledge of "Can havok work at negative altitude?"
[12:07]	Lillie Yifu:	I can see a reason to be a km down
[12:07]	Lillie Yifu:	in fact a use case came upthis morning for it.
[12:07]	Andrew Linden:	technically yes, it would be possible for Havok to collide below 0 on the z-axis
[12:07]	Gaius Goodliffe:	Oil prospecting in SL? ;)
[12:07]	Andrew Linden:	also, it would be possible to put holes in the heightfield's collision model
[12:07]	Sidewinder Linden:	i suppose the good part is that with a flat world model you don't have to worry about drilling through to the other side :)
[12:07]	Lillie Yifu:	draw distance clips at 1024
[12:07]	Morgaine Dinova:	Andrew: kk, tnx, good to know.
[12:08]	Lillie Yifu:	so if you want to be really invisible below gorund, you ahve to be at 1024M and lower
[12:08]	Teravus Ousley:	hehe, in ODE, things don't collide at below 0 on the z-axis :D..
[12:08]	Sorrelle Poutine:	andrew, is it possible for me to send you a movie as an example? I'll find out about the prims, but they're just squares linked together i'm pretty sure.
[12:08]	Lillie Yifu:	the idea was to do a simulation of a mine collapse, for an NGO.
[12:08]	Andrew Linden:	Yes, Sorrelle you can open a bug in JIRA and post the movie there
[12:08]	Sidewinder Linden:	i'm going to have to jump guys... good to see you - back again thursday (andrew's got to run too me thinks)
[12:09]	Morgaine Dinova:	Gaius: no, not tunnelling. The 1Km of bedrock was to ensure no possibility of information bleeding. The only way to get to the parcel basement was by a special teleport operation.
[12:09]	Saijanai Kuhn:	teravus, you could fake it by having the 0 of the simulator offset by a large factor
[12:09]	Teravus Ousley:	Yes, I could.
[12:09]	Andrew Linden:	alternatively, I could go look at the puzzle now, right after this office hour.
[12:09]	Rex Cronon:	bye sidewinder
[12:09]	Sorrelle Poutine:	fixing our puzzle isn't so important, but you might want to know the conditions under which this happens
[12:09]	Kiddoh Korobase:	/ao on
[12:09]	Hulaboom Voom:	Bye sidewinder... :)
[12:09]	Sorrelle Poutine:	since things start behaving oddly. :P
[12:09]	Andrew Linden:	ODE is open source. He could fix it by patching ODE
[12:09]	Teravus Ousley:	Take care :D
[12:09]	Morgaine Dinova:	Cyu Side
[12:09]	Teravus Ousley:	hehe, I could do that too.
[12:09]	Saijanai Kuhn:	laters all
[12:09]	Sorrelle Poutine:	we can do it now, absolutely
[12:09]	Yuu Nakamichi:	later Sai
[12:09]	JayR Cela:	byee all / byee Rexx
[12:09]	Rex Cronon:	bye sai
[12:10]	Hulaboom Voom:	nice boots
[12:10]	Saijanai Kuhn:	all who are leaving at least ;-)
[12:10]	Rex Cronon:	by jayr
[12:10]	Andrew Linden:	I've got to go...
[12:10]	Teravus Ousley:	I've got a meeting on OSGrid actually. Take care :D
[12:10]	Hulaboom Voom:	bye Andrew
[12:10]	Arawn Spitteler doesn't see any problem, with a km deep hole, if vision clips at 1024, since the avatar can sit 300 meters below the seat...
[12:10]	Rex Cronon:	bye andrew
[12:10]	Morgaine Dinova:	Cyu Andrew :-)
[12:10]	Yuu Nakamichi:	thanks Andrew
[12:10]	Rex Cronon:	bye everybody, i am going too
[12:10]	Andrew Linden:	Thanks for coming everyone.
[12:10]	Sorrelle Poutine:	thanks for your help andrew. if you wnat to come and see it now, let me know. although we've hacked it a bit but the behavior still happens
[12:11]	Kitto Flora:	Byebye all
[12:11]	Gaius Goodliffe:	Thank you, Andrew. :)
[12:11]	Teravus Ousley:	Thank you :D
[12:11]	Rex Cronon:	i wonder if andrew can be convinced to relax limits on prim linkage:)
[12:11]	Alyx Stoklitsky:	soup Kiddoh
[12:11]	Kiddoh Korobase:	lol hai :D
[12:11]	Andrew Linden:	I can certainly be convinced to relax link limits... after some other projects
[12:11]	Gaius Goodliffe:	/yay
[12:12]	Andrew Linden:	first of which is objects colliding across region boundaries