User:Andrew Linden/Office Hours/2008 04 17

Transcript of Andrew Linden's office hours:

[17:03]	Phantom Ninetails:	That's a neat chair script, and hello y'all
[17:03]	Andrew Linden:	Hi Phantom. Welcome and have a seat.
[17:03]	Phantom Ninetails:	:)
[17:04]	Andrew Linden:	Oof. I was just starting to get back into fixing bugs today and my alarm went off, reminding me to attend this meeting.
[17:04]	Andrew Linden:	I would have clean forgotten about it.
[17:04]	Phantom Ninetails:	Heh
[17:05]	Andrew Linden:	I've been complaining in previous office hours about some problems that were holding up some Havok4 updates that we had ready to go out.
[17:05]	Andrew Linden:	Today we did an emergency port of these fixes to a new clean branch (without the cruft that was holding us up) and I think the rollout for that will start this week.
[17:06]	Andrew Linden hopes.
[17:06]	Andrew Linden:	Meanwhile, we've got the second update in testing, and it looks like most of the bug fixes have passed. Just one more to verify.
[17:07]	Andrew Linden:	Which means we should have next week's update ready early.
[17:07]	Andrew Linden:	I expect some other projects to try to jump onto that release, so we'll see what happens.
[17:07]	Andrew Linden:	I'll just mention a few bugs that have been fixed but won't be out until next week at the earliest...
[17:08]	Andrew Linden:	"some vehicles hit prim seams hard"... there are several SVC-# bugs for this one and I'm not quite sure how many of them will prove to be fixed
[17:09]	Andrew Linden:	such as SVC-1698 which is specifically for some very small RC cars
[17:09]	Andrew Linden:	I haven't tested the fix with very small cars yet
[17:10]	Andrew Linden:	but the" #12 Pure Stock Brick (1976)" that I got from Les White (I think) and which suffered very badly from this bug, is fixed
[17:10]	Saijanai Kuhn:	whatkind of issues were causing the problems?
[17:10]	Andrew Linden:	SVC-1987 'severe camera hunting bug" is fixed
[17:11]	Andrew Linden:	again there are a bunch of related camera bugs... some I'm sure are fixed... suche as ...
[17:11]	Andrew Linden:	SVC-2049
[17:12]	Andrew Linden:	hrm... some of these I mentioned on Tuesday...
[17:12]	Andrew Linden:	but newly fixed would be SVC-1975 "avatar jump height is not right"
[17:13]	Andrew Linden:	and we hope the "llTargetOmega sometimes stops" bug is fixed, or greatly reduced. That one is hard to repro, but we think we've made some progress there.
[17:13]	Andrew Linden:	That's about it. There are others but most of those I haven't correlated with the public jira item #'s.
[17:14]	Andrew Linden:	Saijanai, you were asking about the source of the problems for the vehicle bounces on prim seams?
[17:14]	Saijanai Kuhn:	right, don't need massive details if its too complicated, but just curious as to the general cause or causes?
[17:15]	Andrew Linden:	Well, it turns out that ALL havok4 objects suffer from the prim seams problem
[17:15]	Gaius Goodliffe:	Just out of curiousity, did you end up using the "seam welder" thing you'd previously mentioned?
[17:15]	Saijanai Kuhn:	ah, so different versions have dfiferent issues?
[17:15]	Andrew Linden:	however there is special code in vehicles to try to reduce it
[17:15]	Andrew Linden:	turns out that a general solution is very hard, but vehicles happen to have a callback that is called every physics engine timestep
[17:16]	Andrew Linden:	so we can add some special magic code there
[17:16]	Andrew Linden:	I had reduced the problem for most vehicles
[17:16]	Andrew Linden:	but a few would still suffer
[17:16]	Saijanai Kuhn:	that might make it easier to move to H5 at some point too
[17:16]	Andrew Linden:	how much they suffered would usually depend on the vehicle params...
[17:17]	Andrew Linden:	vehicles that have a vertical attractor, and substantial inertia tensors tended to be spared
[17:17]	Arawn Spitteler wonders that collisions don't simply push the vehicle up, as seems to be a problem for some trains. Trains seem to be pushed up, when they press to the side of a prim, but stopped by a seam.
[17:17]	Andrew Linden:	I had to add some more... "history" for the vehicle... It now keeps around a notion about where it's "road surface" is
[17:18]	Saijanai Kuhn:	hmmm. Does that apply to things like mono rails and trains?
[17:18]	Andrew Linden:	and it will filter some new collision normals... it will actuall override them if they differ from what it expects
[17:19]	Sidewinder Linden:	andrew, is there a way to improve bumping of "passive" (non-scripted) physical objects over seams - the rolling ball example for instance?
[17:19]	Andrew Linden:	I'm not sure I understand your question Saijanai. Was that for me? Or for Arawn? And what exactly are you wondring if it applies to mono rails?
[17:19]	Andrew Linden:	Sidewinder, I don't know of a good general solution.
[17:19]	Saijanai Kuhn:	you say that vehicles keep track oftheir raod surface. Wondering if tracks count in that sense
[17:19]	Simon Linden:	Sorry folks - I have to duck out. Too many things happening in the office here
[17:20]	Andrew Linden:	It will require a lot of experimentation, and maybe some help from some Havok engineers.
[17:20]	Gaius Goodliffe:	Is it at all related (or possibly related) to objects getting collisions over greater distances now?
[17:21]	Andrew Linden:	Saijanai, if the "road surface" consists of several distinct planes... a large monorail for example, then the code I added might not help much
[17:21]	Saijanai Kuhn:	ah, OK.
[17:21]	Andrew Linden:	but I doubt it will make things worse for such vehicles
[17:21]	Saijanai Kuhn:	was just thinking of Squirrel's amusement parkrides
[17:22]	Andrew Linden:	once it is deployed I'll be interested in looking at further reports of problematic vehicles
[17:22]	Saijanai Kuhn:	She uses monorail type things a lot. Such as rocket chairs and. well, monorail rollercoasters
[17:23]	Andrew Linden:	roller coasters are problematic in Havok4 for a few other reasons as well
[17:23]	Andrew Linden:	there are two other reasons I can think of that are hurting some roller coasters that worked well in Havok1 but not in Havok4
[17:24]	Andrew Linden:	(1) a few objects have slightly different collision shapes, and content that had been tuned in Havok1 to be as tight as possible may be too tight in Havok4, and can jam, get stuck, or even fall through their tracks
[17:24]	Andrew Linden:	for instance... cylindrical objects now use the Havok4 implicit cylinder collision shape... if they have a symmetric cross section
[17:25]	Andrew Linden:	spheres use an implicit sphere shape
[17:25]	Andrew Linden:	(2) concave prims for dynamic objects can suffer a "level of detail" shift in their collision shape, if the physics engine gets severely loaded
[17:26]	Andrew Linden:	when that happens a "torus" will start to collide like a "box" and if the hole was being used to contain a rail or pivot axis, then the two objects will subsequently interpenetrat
[17:26]	Andrew Linden:	and will be forced apart
[17:26]	Sidewinder Linden:	and in some cases, um "rather dramatically" ;)
[17:27]	Saijanai Kuhn:	heh
[17:27]	Gaius Goodliffe:	hehe
[17:27]	Andrew Linden:	There have been a few reports of bugged roller coasters. I haven'
[17:27]	Phantom Ninetails:	lol
[17:27]	Andrew Linden:	I haven't yet surveyed them to see which problems they are suffering from.
[17:27]	Sidewinder Linden:	oh btw, another problem that i have seen in at least one roller coaster is a combination one
[17:28]	Gaius Goodliffe:	Mostly I've noticed that things collide in "near miss" scenarios over greater distances than they used to. Even with boxes, which I assume have a fairly stable shape.
[17:28]	Arawn Spitteler:	So, if a train pops off the rail, I should see what the Sim-Lag is?
[17:28]	Sidewinder Linden:	as a quick aside - it seems that more coaster users than others have no conact with th eoriginal scripters
[17:28]	Andrew Linden:	Arawn, it depends on if you are using concave parts
[17:28]	Sidewinder Linden:	so the other coaster issue seems to be the result of the increased motor speed limit - some coaster cars use the "max speed plus push" strategy
[17:28]	Andrew Linden:	I recommend only using convex parts for any new trains or roller coasters you might be building.
[17:29]	Saijanai Kuhn:	I think there are carnival equipment vendors who hired a scripter/builder but had nothing to do with the actual building
[17:29]	Sidewinder Linden:	so with havok4 the cars attempt to go much faster, which aggravates some of the other issues
[17:29]	Sidewinder Linden:	ahhh
[17:29]	Sidewinder Linden:	sai that would make sense... the "park owners" i've talked to seem as if they were dealing with a reseller for the products
[17:29]	Saijanai Kuhn:	you can actually rent carnival equipment Ithink
[17:29]	Andrew Linden:	yes, item (3) would be that perhaps greater speeds, or even Havok4's more relaxed stance on interpenetrations, and its built-in penetration resolver, could also help roller coaster cars leave their tracks
[17:29]	Saijanai Kuhn:	like for afundraiser
[17:30]	Sidewinder Linden:	i have wondered, actually, whether one of the easy solutions for some of these coasters might be to just slow them down 20% or so and see if they are more stable...
[17:31]	Andrew Linden:	Sidewinder, did you hear any final update on the status of that 1.20.1 simulator we built today?
[17:31]	Arawn Spitteler:	The Fast Hobo Train is using Collision to keep the rail underneath, and Primficiency would recommend using a hollow pathcut box, for this purpose.
[17:31]	Sidewinder Linden:	i've been trying to get a read on that but do not know for sure - i'll go poke around again
[17:31]	Sidewinder Linden:	be back in a few
[17:31]	Andrew Linden:	Arawn, if the concave prim is static you don't have to worry.
[17:32]	Arawn Spitteler:	Static? I think it's a vehicle
[17:32]	Andrew Linden:	If it is part of the dynamic train, then I suspect the dangers of collision LOD shifts would trump primficiency
[17:32]	Andrew Linden:	Unfortunately, I don't have a quick fix for the collision LOD shifting in Havok4
[17:33]	Phantom Ninetails:	Hmmm, would the prim limit be able to be raised from 32 at some point in the future?
[17:33]	Andrew Linden:	it happens to be one of the main things that stand between you and crippling physics lag
[17:33]	Andrew Linden:	Yes Phantom, however before we fix that we need to fix some other problems related to region border crossings
[17:33]	Arawn Spitteler:	So, my Primficient Rail rocket goes from one box to three cylinders?
[17:34]	Phantom Ninetails:	:> Hee, yay
[17:34]	Gaius Goodliffe:	It was actually 64 for a short time during the early beta. Ahh, good times...
[17:34]	Heather Goodliffe:	for some things it might be better if we can keep the physics consistent even if it's not as fine grained?
[17:34]	Andrew Linden:	Perhaps Arawn. I'd have to see your primficient Rail Rocket
[17:34]	Heather Goodliffe:	some bit we can flip in SLS perhaps?
[17:34]	Heather Goodliffe:	* LSL
[17:35]	Saijanai Kuhn:	collisions still work if a prim is 100% invis?
[17:35]	Sidewinder Linden:	it looks like there's another critical (non-havok4) bug that the release team wants to have in before deploying... so it is not clear if we'll be able to deploy this build before the weekend or not... i'd give it a "maybe" level of confidence.. in other words i won't be surprised if it gets held until next week...
[17:36]	Heather Goodliffe:	or, make all things at a low LOD by default
[17:36]	Andrew Linden:	Heather, it is difficult to make the RCCS (runtime collision control system) smart enough that it knows the difference between a grief bomb and some legit bit of concave content that just happens to be killing the simulator but must collide correctly at all costs.
[17:36]	Heather Goodliffe:	Sai, I assume you mean phantom?
[17:36]	Andrew Linden:	However... some progress may be possible there.
[17:36]	Heather Goodliffe:	I've seen some phantom object colliding with each other
[17:36]	Saijanai Kuhn:	well, actually invis. But my solution might nto be possible. You can't mix physics states in linksets, right?
[17:36]	Heather Goodliffe:	under Havok 4
[17:36]	Heather Goodliffe:	at least that's what I think is going on
[17:37]	Sidewinder Linden:	we do have a debug setting that we can set on the simulator for testing, but it is not exposed in any way that can be controlled by others... in most cases where we have turned off the rccs system, the overload characteristics are pretty harsh (that's why rccs was added - to soften that curve)
[17:37]	Andrew Linden:	I actually have an idea on how to fix the collision LOD shift problem for good... however it is a biggish project and I've punted it into the future when I have a great deal of copious spare time.
[17:37]	Sidewinder Linden:	@heather - re keeping lod consistent all the time
[17:38]	Saijanai Kuhn:	if you could, you'duse a phantom sculpty fo rhte appearance, and a simpler prim fofr the underlying collision
[17:38]	Gaius Goodliffe:	per-prim phantom would be a BIG plus for a lot of things.
[17:38]	Andrew Linden:	Right Saijanai, that is a feature idea we've talked about a lot in these office hours.
[17:38]	Andrew Linden:	per-prim phantom is one such feature that could help
[17:38]	Saijanai Kuhn:	nothing new (or I need to stay awake mreo)
[17:39]	Phantom Ninetails:	Hmmmm
[17:39]	Andrew Linden:	similar but distinct is the "replace all prims with a simple collision shape" idea
[17:39]	Phantom Ninetails:	Actually...
[17:39]	Phantom Ninetails:	If you set a primitive to flexible, and then link it to a solid prim, you have exactly what you want
[17:39]	Andrew Linden:	Also possible might be... "group these subset child prims and replace with a simpler collision sub-shape"
[17:39]	Phantom Ninetails:	Flexi prim stays flexi and phantom, solid prim stays phantom
[17:40]	Saijanai Kuhn:	so the physics engine would use the simpler shape for colllisions, but eh veiwer would get the drawn version to play with
[17:40]	Phantom Ninetails:	er solid
[17:40]	Andrew Linden:	right Saijanai
[17:40]	Heather Goodliffe:	if you made a complex object and say wraped in transparent box or sphere, that makes things easier on the physics engine correct?
[17:40]	Andrew Linden:	Yes Heather
[17:40]	Saijanai Kuhn:	sounds a better deal if its workable. Less prim count and com chatter
[17:41]	Sidewinder Linden:	heather, for surfing, do people depend on the wave shape physically for the way the wave "feels", or could it be a much simpler shape and still work right?
[17:41]	Andrew Linden:	Possibilities would be to boxify the object, sphereify it, or to convex-hullify
[17:41]	Phantom Ninetails:	Can sculpted prims be set flexi?
[17:41]	Saijanai Kuhn:	Phantom, nope
[17:41]	Heather Goodliffe:	I would do that more often, but then all my physical object fall to the ground
[17:41]	Phantom Ninetails:	Ah
[17:41]	Andrew Linden:	Phantom... "can they now"? Or "can they ever"?
[17:41]	Phantom Ninetails:	Well I was asking now, but I guess now I am curious about ever
[17:41]	Heather Goodliffe:	instead I have to use hollow objects, to keep the mass down
[17:41]	Heather Goodliffe:	.. not so good for the physics engine
[17:42]	Andrew Linden:	ah yes... that might be another more advanced feature
[17:42]	Heather Goodliffe:	not sure I was ever clear on the reasoning behind the "energy" parameters
[17:42]	Saijanai Kuhn:	H4 has a huge potential if you look at those PR movies
[17:42]	Heather Goodliffe:	which PR movies?
[17:42]	Andrew Linden:	Awww... nice Khona just gave me a martini
[17:42]	Saijanai Kuhn:	google youtube havok
[17:43]	Saijanai Kuhn:	or even youtube video search
[17:43]	Saijanai Kuhn:	havok*video search
[17:43]	Andrew Linden:	er... the more advanced feature I was about to talk about would be a custom "mass properties" setting, so you could override the mass, center of mass, and inertia tensor of the object
[17:44]	Heather Goodliffe:	that would be awesome
[17:44]	Heather Goodliffe:	:)
[17:44]	Andrew Linden:	then Heather could use solid objects without worrying about the extra mass
[17:44]	Heather Goodliffe:	yes! :)
[17:44]	Arawn Spitteler wonders if Andrew wants to see the One Prim Rail-Rocket. I think it'll unseat me, if I try to set the hollow back to Square, but this is what might have to become three prims.
[17:44]	Andrew Linden:	btw... the #12 Pure Stock Brick I was talking about earlier...
[17:44]	Sidewinder Linden:	i bet that would make a lot of people happy, and create a whole new class of interesting things in second life ;)
[17:44]	Andrew Linden:	it has ra really light weight body and a very small inertia tensor
[17:45]	Andrew Linden:	which was a main reason why it suffered from the prim seams problem so bad
[17:45]	Heather Goodliffe:	any reason why you can't just kill the "energy" thing?
[17:45]	Heather Goodliffe:	heh
[17:45]	Gaius Goodliffe:	Tradition. :)
[17:45]	Heather Goodliffe:	one less complexity in SL to worry about
[17:45]	Phantom Ninetails:	lol
[17:45]	Andrew Linden:	We're wandering into blue sky country... I can't work on all those wonderful features until we fix more bugs.
[17:45]	Gaius Goodliffe:	aww
[17:45]	Andrew Linden:	Another noteworthy announcement...
[17:46]	Andrew Linden:	Qarl Linden has expressed his interests in working on the "my neighbor's objects overlap into my parcel problem"
[17:46]	Heather Goodliffe:	seems like killing the "energy" thing would be straight forward?
[17:46]	Andrew Linden:	so I'm going to be helping him work on that... he may start as early as next week
[17:46]	Gaius Goodliffe:	Cool
[17:47]	Heather Goodliffe:	as long as you fix all the Havok4 bugs first! ;)
[17:47]	Sidewinder Linden:	hehe heather
[17:47]	Andrew Linden:	some of you know that the "prim encroachment" problem fix is a pre-requisite for megaprim liberation
[17:47]	Andrew Linden:	sorry, I didn't get that "killing the energy" reference
[17:47]	Sidewinder Linden:	actually while it might be workable to make script energy go away, i suspect it would break a huge amount of content if it were just removed... lots of things do a lot of manipulations around that behavior, for better or worse...
[17:48]	Andrew Linden:	oh, script energy
[17:48]	Sidewinder Linden:	heather had asked why we ahve script energy, and whether we might just remove the concept, or at least that's what i heard... ?
[17:48]	Heather Goodliffe:	yup
[17:48]	Phantom Ninetails:	You could just make it always tell the script that max energy is available?
[17:49]	Andrew Linden:	yeah arawn, that hollo monorail bottom is not going to work well
[17:49]	Gaius Goodliffe:	llDisableCrazyLegacyFeatures(TRUE);
[17:49]	Phantom Ninetails:	lol
[17:49]	Phantom Ninetails:	That'd work too
[17:49]	Arawn Spitteler:	I think a lot ofRail Content might work on this principle.
[17:49]	Andrew Linden:	"script energy" was added waaaaaay back in the very beginning
[17:50]	Andrew Linden:	the idea was to spend script energy for certain actions
[17:50]	Arawn Spitteler guesses that Script Energy was a limit of how much an object could demand of computations
[17:50]	Andrew Linden:	think of it as a pre-gray-goo protection system
[17:50]	Andrew Linden:	we were worried about scripted objects that could overwhelm the world before we could react
[17:51]	Sidewinder Linden:	and a way of making sure that "unlimited force" is not available to weapons
[17:51]	Andrew Linden:	so we thought we would throttle via some "resource" that the scripts could spend as long as they still had some
[17:51]	Andrew Linden:	It still works in this regard
[17:51]	Andrew Linden:	however, there are also other throttles in place now
[17:52]	Andrew Linden:	and it is not clear whether we could rip it out entirely -- probably not for legacy objects
[17:52]	Sidewinder Linden:	although if it made sense there could potentially be some sort of IGNORE_SCRIPT_ENERGY kind of flag... it's not clear what the full repercussions would be though
[17:53]	Gaius Goodliffe:	Less people putting ten scripts in something to do the work of one. :)
[17:53]	Andrew Linden:	Sidewinder... you mean a per-region setting? or a per-parcel setting?
[17:53]	Sidewinder Linden:	i was just thinking out loud... hmm
[17:53]	Sidewinder Linden:	applying it to geography might be interesting as a way to handle it )
[17:53]	Heather Goodliffe:	seems like reducing the complexity it adds might be a good things at this point :)
[17:53]	Saijanai Kuhn:	was talkingwith the Mono folk about that for the next gen of LSL --a transition between LSL and C#
[17:53]	Heather Goodliffe:	I'd be happy for being able to se tthe mass of my object
[17:53]	Andrew Linden:	Gaius, ten scripts on one object -- they all tap into the same script energy tank
[17:54]	Gaius Goodliffe:	Ah okay
[17:54]	Sidewinder Linden:	well this actually plays into something i'd been thinking, which is that, as we've said many times...
[17:54]	Andrew Linden:	however, the OTHER throttles in place for single scripts get circumvented
[17:54]	Saijanai Kuhn:	seems llike half of Babbages' prolems are maintaining LSL2 bugs
[17:54]	Sidewinder Linden:	this project was about maintaining as much compatibility as we could while getting to havok4
[17:54]	Andrew Linden:	such as per-script call sleeps
[17:54]	Sidewinder Linden:	to reduce crash rates
[17:54]	Heather Goodliffe:	you uped the max vehicle speed ;)
[17:54]	Sidewinder Linden:	perhaps some future phase could look at "rationalizing" some fo the history
[17:54]	Heather Goodliffe:	hurray for that :)
[17:54]	Sidewinder Linden:	now that we have a simulator that crashes so much more rarely
[17:54]	Gaius Goodliffe:	I use the multiple script option to get around script delays... but that's different from energy.
[17:54]	Andrew Linden:	Yeah, we had a few discussions about that
[17:54]	Andrew Linden:	the max vehicle speed
[17:54]	Heather Goodliffe:	just syaing, it's a similar thing ;)
[17:55]	Sidewinder Linden:	yup
[17:55]	Saijanai Kuhn:	with a new mono-only language, could there be a new H4 + mono-only physics model?
[17:55]	Sidewinder Linden:	<- has always wondered about the flight / altitude / flight feather etc models
[17:55]	Sidewinder Linden:	;)
[17:55]	Andrew Linden:	I'm not sure I understand that question Saijanai, re MONO + Havok4
[17:56]	Creem Pye:	more vehicle parameters, perhaps?
[17:56]	Saijanai Kuhn:	well, just as LSL3 has backwards compatiblity bugs that need suport in mono, H4 has to support bugs from H1
[17:56]	Saijanai Kuhn:	would it be possible to simplify by starting over in some sense?
[17:56]	Heather Goodliffe:	big ball of mud :)
[17:56]	Saijanai Kuhn:	LSL2*requires backwards support in mono
[17:57]	Andrew Linden:	Right, I had attended Babbage's office hours earlier this week, and tried to ask him about the idea of an LSL3
[17:57]	Saijanai Kuhn:	last I heard, he wants to just go whole hog to C#
[17:57]	Arawn Spitteler:	LSL2 doesn't require backwards compatibility, since it would only be MONO Compiled by the Script Creator.
[17:57]	Saijanai Kuhn:	I've been thinking a transitional LSL2-like subset of C#
[17:57]	Andrew Linden:	it didn't sound like it was what he really wanted to do... he has some ideas on how to greatly enhance LSL... loadble C# modules and user-created libs
[17:58]	Arawn Spitteler:	Old content simply wouldn't be compiled.
[17:58]	Andrew Linden:	but he's less interested in a major overhaul, or a second/better script engine that runs in parallel with the legacy one.
[17:58]	Heather Goodliffe:	might be getting a little OT ;)
[17:58]	Saijanai Kuhn:	eh, perhaps he's changed his mind again. Thats such a can ofworms there
[17:58]	Sidewinder Linden:	ya think? :)
[17:58]	Gaius Goodliffe:	C# will be nice, although I've kinda warmed to LSL's "state machine".
[17:59]	Saijanai Kuhn:	mono fakes it on the VM side anyway
[17:59]	Andrew Linden:	some of my main problems with LSL (after my Havok4 efforts) are to do with how some LSL calls are simple wrong/broken and cannot be fixed
[17:59]	Saijanai Kuhn:	so it can stay around as far as I can tell, no matter what language you use
[17:59]	Andrew Linden:	for example... llGetMas() returns a legacy mass
[17:59]	Andrew Linden:	where the density of your object is lighter than styrofoam
[18:00]	Creem Pye:	does it scale with "real" mass at least?
[18:00]	Andrew Linden:	it cannot be fixed... because so many LSL scripts rely on it returning legacy mass
[18:00]	Andrew Linden:	yes, it does scale with real mass
[18:00]	Saijanai Kuhn:	llNewMass() problem fixed ;-)
[18:00]	Andrew Linden:	so... we either need a llGetRealMass()
[18:00]	Andrew Linden:	or we need a new LSL3 engine where llGetMass() returns the real mass
[18:00]	Creem Pye:	I guess that's why llSetForce(<0,0,9.8>,FALSE)*llGetMass() doesn't quite make you hover...
[18:01]	Saijanai Kuhn:	I'd go with LSL3 + new api. Less confusion, fewer things to worry about
[18:01]	Andrew Linden:	no Creeme, I suspect a different problem there
[18:01]	Gaius Goodliffe:	llGetSunDirection is horribly borked now too, but that's a different topic...
[18:01]	Saijanai Kuhn:	Ted Maa is having fun with OSSL but thats another story
[18:01]	Andrew Linden:	Gaius... borked now that Havok4 is out? or borked for different reasons?
[18:01]	Gaius Goodliffe:	Borked by Windlight -- the sun is in a completely different place under 1.19+ vs. 1.18
[18:02]	Sidewinder Linden:	last i heard we're not calculating celestial mechanics using h4... ;)
[18:02]	Andrew Linden:	oh
[18:02]	Saijanai Kuhn:	ouch
[18:02]	Gaius Goodliffe:	Like I said, entirely different topic...
[18:02]	Gaius Goodliffe:	But yeah, lots of older LSL functions return meaningless values these days.
[18:02]	Heather Goodliffe:	so, perhaps on the plan for fixin gthis week, what are the priorities atm? :)
[18:03]	Saijanai Kuhn:	sai admires his stuck typikng animation
[18:03]	Creem Pye:	*cough* llGetOmega() always returns ZERO_VECTOR in attachments when your av sits on a vehicle
[18:03]	Sidewinder Linden:	gaius... it happens i'm umm pretty close to a guy who knows the innards of windlight... (boston office)
[18:03]	Andrew Linden:	Well, that about ends the hour. Any last minute questions or issues?
[18:03]	Sidewinder Linden:	and they know about the sun issue - it won't be broken forever ;)
[18:03]	Heather Goodliffe:	yes *see above*
[18:03]	Gaius Goodliffe:	cool
[18:03]	Heather Goodliffe:	" on the plan for fixin gthis week, what are the priorities atm? :)"
[18:04]	Umeko Kawanishi:	a quick question, andrew
[18:04]	Andrew Linden:	ah yes, my scroll had got stuck...
[18:04]	Umeko Kawanishi:	was wondering if anyone is doing performance tests/stress tests on both the server side and client side?
[18:04]	Andrew Linden:	Simon is working on some performance optimizations for some particularly laggy content
[18:05]	Andrew Linden:	Kelly is working on...
[18:05]	Andrew Linden:	I forget what kelly was working on
[18:05]	Sidewinder Linden:	in particular we are seeing some problems with collision event handling - not trigering well
[18:05]	Sidewinder Linden:	*triggering
[18:05]	Andrew Linden:	I've been going through public jira trying to find my next set of important bugs
[18:05]	Yuu Nakamichi:	svc-2014?
[18:06]	Scalar Tardis:	spinning objects seem to be freezing after a sim restart now. I've not checked to see if that has been JIRA'd
[18:06]	Umeko Kawanishi:	are there test suites or these are issues that just happen to come up?
[18:06]	Sidewinder Linden:	i suppose it's worth mentioning that we'd hinted there might be a change to boat behavior, and that's now off the table
[18:06]	Creem Pye:	it's JIRA'd I think...
[18:06]	Sidewinder Linden:	in terms of the basics
[18:06]	Scalar Tardis:	ok
[18:06]	Saijanai Kuhn:	Andrew a thought just occured to me. YOu know I have a python script that creates a ruthed avie in-world, right? What is needed to have an avie sit on a vehicle and send drive commands?
[18:06]	Andrew Linden:	Yes, I think we'll be able to take a pass at SVC-2014 over the next week
[18:06]	Sidewinder Linden:	the one thing that we've heard a lot about is that there are some boats that float at seemingly incorrect heights
[18:07]	Heather Goodliffe:	that's the most noticable and reproducable one :)
[18:07]	Sidewinder Linden:	if there are cases - such as the surfboards or others, we'll need samples to take a look at, as some float at the right place
[18:07]	Sidewinder Linden:	ando others don't
[18:07]	Andrew Linden:	Saijanai, the vehicle takes the normal control input that the avatar does.
[18:07]	Yuu Nakamichi:	cool
[18:07]	Creem Pye wonders if there's any relation to the geometric vs mass centers in the boats
[18:07]	Saijanai Kuhn:	hmm. thinking that it should be "trivial" to create a script to drive a car for QA purposes
[18:07]	Andrew Linden:	No creeme, I wouldn't expect center of mass changes to affect anything
[18:08]	Saijanai Kuhn:	I'll talk to Enus about it. He's looking to use the script on the beta grid
[18:08]	Yuu Nakamichi:	I might be able to thelp with the boats, Side
[18:08]	Andrew Linden:	Hrm... well maybe for some bobbing boats it might...
[18:08]	Gaius Goodliffe:	Aren't hover heights specified relative to the root prim?
[18:08]	Andrew Linden:	Heather, I had started to look at your trick sailboard
[18:08]	Andrew Linden:	however I couln't get it to flip in Havok1
[18:08]	Heather Goodliffe:	Ah, I was goign to mention I figured that one out
[18:09]	Andrew Linden:	I could get it to do some jumps
[18:09]	Heather Goodliffe:	seems to be related to the hover effieciency
[18:09]	Sidewinder Linden:	heather was there a "basic float height" problem with that sailboard? or is it mostly the tricks?
[18:09]	Andrew Linden:	and verified that the jumps don't happen in havok4
[18:09]	Heather Goodliffe:	I posted a comment on someone's jira that sounded related
[18:09]	Andrew Linden:	as far as I could tell the sailboard was floating at about the right height in Havok4
[18:09]	Heather Goodliffe:	if hover efficiency is anywhere close to 1.0 llApply impulse doesn't do anything in the verticle direction...
[18:09]	Heather Goodliffe:	which sounds logical
[18:09]	Creem Pye:	then does the hover altitude depend on object rotation? <gaius
[18:10]	Heather Goodliffe:	but havok 1 had different functionality
[18:10]	Heather Goodliffe:	for my purposes you probably don't have to worry about, probably just good to know if other people are having similar issues
[18:11]	Andrew Linden:	hrm... right the hover efficiency should override an llApplyImpulse() if the efficiency is high enough
[18:11]	Gaius Goodliffe:	Creem: It's always global Z -- otherwise you'd fall down if you tilted sideways
[18:11]	Heather Goodliffe:	apparently it didn't under H!
[18:11]	Heather Goodliffe:	H1
[18:11]	Sidewinder Linden:	does the basic hover height seem right to you heather, with your boards?
[18:11]	Heather Goodliffe:	perhaps another H1 bug
[18:11]	Andrew Linden:	Ok, I'll look at the hover stuff.
[18:12]	Sidewinder Linden:	it is funny - well to some degree expected, but interesting - that in this phase of the project we're starting to see a good number of things crawl out of the woodwork
[18:12]	Sidewinder Linden:	that are basically
[18:12]	Heather Goodliffe:	the boards now sink under water, people are more swimming under water than paddling on top of the water
[18:12]	Sidewinder Linden:	"hmm we fixed a bug for h4, but some old content relies on the bug being there"... and then the hard decision ;)
[18:12]	Creem Pye:	then the hover height is relative to your root prim when the vehicle is at zero rotation I guess..
[18:13]	Sidewinder Linden:	so andrew you weren't seeing the sink underwater that heather described?
[18:13]	Arawn Spitteler will have to look, whether concavity is the problem with SVC-1981 but, http://jira.secondlife.com/browse/SVC-2141 is something completely different: Any idea how the Fast Hobo Train is being picked up by its butt, and tossed into a corner?
[18:13]	Andrew Linden:	sidewinder, I was looking at her sailboard, not the paddle surfboards
[18:13]	Ellla McMahon:	yes Sidewinder, a lot of people don;t understand that ... relies on the bug being there".
[18:14]	Gaius Goodliffe:	Ugh... BMP
[18:14]	Andrew Linden:	Not yet Arawn, no idea, but I saw that bug in public jira, and was about to assign it to myself, or already did
[18:14]	Andrew Linden:	I'll doublecheck
[18:14]	Arawn Spitteler thinks, the place to loose the old bugs, is at LSL-MONO
[18:15]	Arawn Spitteler:	Okay, That Train may also have problems with the concavity issue, but that being picked up is weird
[18:15]	Andrew Linden:	I'm going to have to run soon
[18:15]	Sidewinder Linden:	thanks andrew
[18:15]	Arawn Spitteler has a class to teach, also: We're glad you could make it, as much as has been.
[18:15]	Phantom Ninetails:	Sayonara
[18:15]	Ellla McMahon:	thank you Andrew :)
[18:16]	Creem Pye:	thanks for your time
[18:16]	Heather Goodliffe:	thanks Andrew, gl with it :)
[18:16]	Andrew Linden:	thanks for coming