User:Andrew Linden/Office Hours/2008 04 03

Transcript of Andrew Linden's office hours:

[17:02]	JenzZa Misfit:	i have NOT ... taken the bait yet
[17:02]	JenzZa Misfit:	hello Simon :)
[17:02]	Rex Cronon:	i am using the rc and i can't seem able to reposition my HUDs:(
[17:02]	JenzZa Misfit:	I am skeeeered of the RC !
[17:02]	Andrew Linden:	wow, big crowd today. What's the occasion? ;-)
[17:02]	Simon Linden:	Hi everyone
[17:02]	RacerX Gullwing:	hi
[17:02]	JenzZa Misfit:	hey Andrew :)
[17:02]	Rex Cronon:	hi
[17:02]	JenzZa Misfit:	.........whats the occasion ? lol
[17:02]	Seven Shikami:	Oh boy, lindens! *throws pokeball*
[17:02]	JenzZa Misfit:	~*~ Hoooo !! ~*~
[17:02]	Scalar Tardis:	I am just here for the entertainment value.
[17:02]	Yuu Nakamichi:	hi andrew, simon
[17:02]	Rex Cronon:	i think it has something to do with h4 being relesed into the wild:)
[17:02]	Kitto Flora:	Hello hello
[17:03]	Simon Linden:	I think we're load testing this table pretty well
[17:03]	JenzZa Misfit:	mmmmmmmmmm.............could be !
[17:03]	Creem Pye wonders if this table can fit everybody
[17:03]	Cinthya Loveless:	move your feet lose your seat Simon
[17:03]	RacerX Gullwing:	my ati card hates crowds everything breaks up into triangles
[17:03]	Arawn Spitteler:	We should have had more track, before the release
[17:03]	Creem Pye:	hehe
[17:03]	Chair: Press Page Up to move chair up, or Page Down to move chair down
[17:03]	Kooky Jetaime:	Racer - thats not just crowds :)
[17:03]	Arawn Spitteler:	Imposters has to be turned of, for me to see people in Windlaugh
[17:03]	Kooky Jetaime:	I'll say this, I have 20FPS on my laptop now with an ATI.. but its crappy
[17:03]	Andrew Linden:	BTW everyone, I might not be able to attend for the whole hour. I've got some minor emergencies to deal with at home and am expecting someone to come by any minute.
[17:03]	Kitto Flora check the visitors for wet feet, and butts.
[17:03]	Harleen Gretzky:	*impostors
[17:04]	Cinthya Loveless:	Hey Andrew nice work on the feet issues you did really good on it
[17:04]	Andrew Linden:	Ok, I guess I'll just make any announcements I can remember before I have to go...
[17:05]	Andrew Linden:	We've got a new update that should have gone to QA today
[17:05]	Andrew Linden:	did that get out Simon?
[17:05]	JenzZa Misfit:	how will this work Andrew ? so we have order ... go round the table ?
[17:05]	JenzZa Misfit:	( when SW gets here I hope )
[17:05]	Andrew Linden:	No JenzZa, we'll make a few announcements and then we'll just open the table for topics
[17:05]	JenzZa Misfit:	groovey
[17:05]	Simon Linden:	I'm not sure of the exact status, Andrew ... I'll check ...
[17:06]	Douglas Rishmal:	where's a chair I can sit on?
[17:06]	Cinthya Loveless:	looks like the table is full
[17:06]	Kooky Jetaime:	Whats the update entail?
[17:06]	Andrew Linden:	There are a few bug fixes in this update... I can't remember them all,
[17:06]	Andrew Linden:	but if any occur to me I'll mention them
[17:06]	Andrew Linden:	Some bugs that are NOT in this update are:
[17:07]	Andrew Linden:	(1) some vehicles still hit prim seams
[17:07]	Kitto Flora:	HOVER height fix?
[17:07]	Simon Linden:	Nope, Kitto, not yet
[17:07]	Andrew Linden:	(2) buoyant vehicles with hover
[17:07]	Simon Linden:	Flying hover, that is ... it's not fixed yet
[17:07]	Andrew Linden:	(3) avatar sometimes floats down when in hover
[17:07]	Kooky Jetaime:	Andrew - something I mentioned to Jon/Jeremy is getting the KB containing a release notes for the various SL Server (and client) versions in one location, since if we have different sims we should have a way to see "at a glance" the differences.
[17:07]	Kooky Jetaime:	yea, I seem to sink while flying
[17:07]	Andrew Linden:	Some that should be fixed are:
[17:08]	Andrew Linden:	Avatar sometimes goes into "stride" animation state when walking up stairs and gets stuck
[17:08]	Andrew Linden:	Removing the vehicle action from an object sometimes leaves it with zero friction
[17:09]	Andrew Linden:	Simon fixed a bug with some concave objects not getting concave collision shapes
[17:09]	Andrew Linden:	And... some other stuff that I can't remember.
[17:09]	JenzZa Misfit:	cool !
[17:10]	Andrew Linden:	Oh yeah... I modified the settings on the "bomb squad" feature I've talked about before. It will be much more liberal in the future.
[17:10]	Douglas Rishmal:	what is the "bomb squad" feature?
[17:10]	Andrew Linden:	I did some legwork earlier this week tracking down incidents where the "bomb squad" had deleted content on region restart... and the main things affected were piles of very complex necklace attachments that were very close to each other.
[17:11]	Simon Linden:	One other fix ... some very thin prims were acting phantom (you wouldn't collide) and that's fixed
[17:11]	Scalar Tardis:	oh ah, that would be where too many collisions at sim startup can crash the whole server, I believe
[17:11]	Andrew Linden:	The "bomb squad" is a scanner that will delete objects if they are rezzed too many copies or prims in one small location.
[17:11]	Seven Shikami:	Delete, not return? Yikes.
[17:11]	Andrew Linden:	Basically, it will delete content if you make the prims per cubic meter too high.
[17:12]	Jen Shikami:	Eek.
[17:12]	Douglas Rishmal:	Sort of to prevent "grey goo" attacks, correct?
[17:12]	Andrew Linden:	However, in the future it has a smaller resolution in its binning.
[17:12]	Jen Shikami:	Ahhh, I see...
[17:12]	Gaius Goodliffe:	Not grey goo so much as physics bombs
[17:12]	Andrew Linden:	And it doesn't kick in until somthing over 1k prims per 1/8 cubic meter.
[17:12]	Seven Shikami:	Oh, that's reasonable, ok.
[17:12]	Andrew Linden:	yes, the bomb squad specifically targets some content that can crash Havok4
[17:12]	Scalar Tardis:	too many collisions consumers too much memory at sim startup, which makes the OS choke, if I recall right
[17:13]	Douglas Rishmal:	Would it help guard against some forms of griefing?
[17:13]	Andrew Linden:	once the bomb squad feature settles down we'll publish its thresholds
[17:13]	Simon Linden:	Douglas - yes, that's exactly why it's there
[17:13]	Andrew Linden:	yes sorta Douglas
[17:13]	Andrew Linden:	atm it does not do real-time scanning
[17:13]	Douglas Rishmal:	Are there any plans to implement Havok Cloth and Havok Destruction in SL? I'd think Cloth would make MUCH more realistic clothing than what we have now.
[17:14]	JenzZa Misfit has her hand up to get called on by Andrew when its her turn :)
[17:14]	Rex Cronon:	there are some sims that are still kind of laggy enven wiht h4 running
[17:14]	Douglas Rishmal:	(ie. fluttering shirts, etc.)
[17:14]	Andrew Linden:	but if you crash the region the bomb squad will scan on restart, find the offending objects, delete them, and log the incident
[17:14]	Andrew Linden:	Ok, any other announcements Sidewinder and Simon before we open up the discussion?
[17:14]	JenzZa Misfit:	hi Sidewinder :)
[17:14]	Seven Shikami:	Hi again, SW. :)
[17:14]	Douglas Rishmal:	hey SW :)
[17:15]	Simon Linden:	The new build just went to QA, so hopefully will get on the beta grid by the weekend
[17:16]	Andrew Linden:	okay, I guess we can open the floor for questions and issues.
[17:16]	JenzZa Misfit:	lol
[17:16]	Creem Pye:	/well, I'm still noticing that vehicle physics time is pretty high (like 0.8ms for small aircraft,~1.5s for ground vehicles driving on land, and ~2.2ms for ground vehicles driving on 10m cube road
[17:16]	JenzZa Misfit:	im goin !
[17:16]	JenzZa Misfit:	.. Ok Andrew -- we have worked with Sidewinder for months with our Rendezvous walking vehicle * and with you actually * and have some minor problems but not urgent. I am here tonight with RacerX and Seifert on behalf of the Giant Snail Races that go on each Sat for over 3 years now . So I will let Seifert and Racer explain the problem that has been on the Jira for sometime now -- and we got people to vote on it today. thanks ! Go Seifert or Racer .... lol
[17:16]	Andrew Linden:	Yes Creem, the cost for vehicles is still too high.
[17:16]	Creem Pye:	it's ~10x, although it's difficult to measure precisely in H1 because usually a single vehicle is ~0.1ms
[17:17]	Gaius Goodliffe:	I've got an airship that pumps the physics time up to 5-6ms!
[17:17]	Scalar Tardis:	I've seen physics "droop" for a second or two, where things all fall down, but then perk up again. haven't figured it out yet
[17:17]	Gaius Goodliffe:	Same scripts as other airships, just the shape seems to cause issues.
[17:17]	Kitto Flora:	I visited a problem Sim today, Phys time was 20mS!! LSL was running at 1/30th top speed.
[17:17]	Kitto Flora:	This matter needs attention.
[17:18]	Andrew Linden:	Yeah, we've had reports of more lag in Havok4 for some content.
[17:18]	Jen Shikami:	Our physics was 40ms earlier today till we removed some wave generators.
[17:18]	Andrew Linden:	I think one of the example bits of content was a vehicle+phantom+concave ocean wave.
[17:18]	Simon Linden:	We've found often there are very specific content that causes most of the lag ... like those wave generators
[17:19]	Scalar Tardis:	my spinning noncolliding llMoveToTarget spinning paddle wheels are burning 10ms not colliding with anything
[17:19]	Douglas Rishmal:	Are there any plans to implement Havok Cloth and Havok Destruction in SL? I'd think Cloth would make MUCH more realistic clothing than what we have now. (ie. Clothing that realistically flutters and moves, not just static tight clothing.)
[17:19]	Creem Pye:	Andrew, would it be wise to use our extra dynamic prim in H4 to make an invisible "bounding cube" to cover the vehicle?
[17:19]	Douglas Rishmal:	(also.. would Havok Behavior improve av realism?)
[17:19]	Seven Shikami:	A little wandering sand crab kicked our physics time to 30 ms. Removing it dropped it down to 0.3ms. Definitely check Top Colliders in estate tools, folks!
[17:19]	Andrew Linden:	Since vehicles are known to be high cost, in particular vehicles with complex shapes, this was exemplar content for punishing the H4 simulator.
[17:19]	Simon Linden:	Douglas - not concrete plans. We're still trying to make this much better and more stable before adding features
[17:19]	Andrew Linden:	That would be worth trying Creem.
[17:20]	Gaius Goodliffe:	Would you like copies of our worst offenders?
[17:20]	Douglas Rishmal:	Also, are there any plans for reducing/eliminating lag with high av counts in the scene?
[17:20]	Andrew Linden:	Douglas, no plans to use cloth or destruction features in SL right now
[17:20]	Andrew Linden:	.
[17:20]	Seifert Surface:	jenzZa was referring to this odd friction that affects avatars in freefall
[17:20]	Andrew Linden:	ATM we're working on bugs and trying to get Havok4 back up to H1 snuff in all areas.
[17:21]	RacerX Gullwing:	http://jira.secondlife.com/browse/SVC-1179
[17:21]	Simon Linden:	One trick for problem stairs you may have - put a transparent ramp on it for the AV to walk up
[17:21]	Andrew Linden:	Ah yes, I saw the PJIRA items and the comments. And got emails about that Seifert.
[17:21]	Seifert Surface:	the snail races use the same scripts as my stuff
[17:21]	JenzZa Misfit:	yes yes and next race is Sat :(
[17:22]	Andrew Linden:	I'm going to examine the stair problem and see if I should lift the limit for traversable obstacles.
[17:22]	Seven Shikami:	We've been having a problem with llTargetOmega in our sim. I realize it's mostly a client side effect but this only started with the Havok 4 push, sooo. Basically, many objects, notably wearable attachments, which had omega spins already going simply stop spinning simultaneously. It's happened 3-4 times since the H4 push, and it's noticable by multiple clients (not just an isolated single observer).
[17:23]	Jen Shikami:	When it's happening, spinning objects work fine rezzed on the ground, but wear the same item and it stops.
[17:23]	Andrew Linden:	Seven, do you have any hints on how to reproduce that bug? It is intermittent, right?
[17:23]	Renee Faulds:	I would like the stair limit up a tad I am having problems now on the deploy with some which would require total rebuilds and repackaging
[17:23]	Gaius Goodliffe:	I noticed yesterday that the new client cause spinning objects top stop spinning at a distance, and restart if I got closer.
[17:23]	Jen Shikami:	Yep, intermittent.
[17:23]	Gaius Goodliffe:	Switching back to the old client, they spun just fine.
[17:23]	Kooky Jetaime:	Jen - I've seen the llTargetOmega stop awhile ago
[17:23]	Kooky Jetaime:	like months ago
[17:23]	Andrew Linden:	If someone could recommend max stair heights I'd consider them when testing for the fixes for that.
[17:24]	Gaius Goodliffe:	(and by distance I mean a ridiculously short distance -- not seeing the radar dish spin while standing on the deck of a small carrier, for example)
[17:24]	Andrew Linden:	Some people, I hear, had "ladders" and their avatars could walk vertically up them.
[17:24]	Seven Shikami:	The massive omega stops only started with the H4 push, though. I've seen isolated stops in windlight before but this was exceptional.
[17:24]	Jen Shikami:	Huh, this is the first I've seen it (three times since H4 deploy, for an hour or two at a time.)
[17:24]	Kooky Jetaime:	Andrew - They did
[17:24]	Kooky Jetaime:	I've seen some
[17:24]	Kooky Jetaime:	Seven - I saw them before in 1.18
[17:24]	Renee Faulds:	I have a 1 degree ladder you can walk up on y water storage tower
[17:25]	Kooky Jetaime:	Andrew - perhaps a dual-threshhold for steps? Anything smaller than X or larger than X
[17:25]	Andrew Linden:	Ah, so that is how ladders work. They are not quite vertical -- slightly slanted.
[17:25]	Renee Faulds:	slight slant
[17:25]	Kooky Jetaime:	that way you still have long single-prim ladders (and sculpti-steps), and you can have standard stairs
[17:25]	JenzZa Misfit:	ie: ramp
[17:26]	Jen Shikami:	As for reproducing the spin-stop bug, unfortunately I don't know how to trigger it. I just sit in my fishing area and every so often, everyone's fish pets stop swimming. (They all see them stopped, too.)
[17:26]	Seven Shikami:	It seems to hit all spinning attachments, yes, and all at once.
[17:26]	Jen Shikami:	When I go to another region, they'll start again... I come back, and they stop.
[17:26]	Renee Faulds:	.75 I think would be max hieght for a stair
[17:26]	Kooky Jetaime:	.5 used to be "kick" height
[17:26]	Kooky Jetaime:	.5 or higher and you would kick the steps in front of you
[17:26]	Andrew Linden:	Ok Renee thanks. I'll see what I think about 0.75.
[17:27]	Andrew Linden:	We still have a few Havok1 regions running on the Beta grid
[17:27]	Andrew Linden:	and I can test what used to work there.
[17:27]	Jen Shikami:	When I talked to the concierge about it earlier, Leo, he said he'd email Sidewinder about it.
[17:27]	Andrew Linden:	No, it won't be possible to have a Havok1 region running indefinitely on the main grid.
[17:27]	Mitchmiester Slade:	I use the llTargetOmega to spin an object.. but it does not stop at all.. but it is not attached
[17:27]	Yuu Nakamichi:	I have a question related to collisions - water vehicles stop completely on contact with avatars (in h1, avatars would be pushed aside). Most people in the sailing community I have spoken to think that is not desirable. pjira:
[17:27]	Yuu Nakamichi:	https://jira.secondlife.com/browse/SVC-2014
[17:28]	Jezebella Desmoulins:	I had some weirdness happen with some scripted swimming fish... they no longer stay inside phantom water prim.
[17:28]	Kooky Jetaime:	I saw something about fish rezzers as an issue
[17:28]	Andrew Linden:	Jezebella, what used to keep them inside?
[17:28]	Rex Cronon:	we have flying fish now?
[17:28]	Andrew Linden:	How is the containment supposed to work?
[17:29]	Jezebella Desmoulins:	They stayed because the were placed at a z coordinate and not set to move.
[17:29]	Jezebella Desmoulins:	Swim x and y, not z.
[17:29]	Andrew Linden:	and when they escape they are changing their z component?
[17:29]	RacerX Gullwing:	i've been selling a trampoline a number of years and have one at my dock in fujin and it no longer stops you from falling into the water and only seems to sense your coliding with it half the time
[17:29]	Jezebella Desmoulins:	Yes, popping up to surface.
[17:30]	Jen Shikami:	Lindens, busy discussion here... if this isn't the appropriate forum for the sim-wide attachment omega stop bug, let us know who to bring it up to.
[17:30]	Andrew Linden:	This is a good place to bring up the TargetOmega bug
[17:30]	Jen Shikami:	Oh good. :D
[17:30]	Andrew Linden:	however, I worry that it really is a client-side bug
[17:31]	Jen Shikami:	Would everyone see it at once, then?
[17:31]	Seven Shikami:	It hit multiple clients simultaneously, though.
[17:31]	Simon Linden:	Jen - we do know about that bug ... the tough part is reproducing it. It's really hard to fix something like that, but we definitely want to
[17:31]	Andrew Linden:	it doesn't just happen in windlight right? It also happens with earlier clients?
[17:31]	Scalar Tardis:	I've never been completely clear if nonphysical has anything to do with havok. AFAIK it is totally client side and it doesn't really move as far as the sim is concerned
[17:31]	Jen Shikami:	Ah, so it's been reported from other sources soo? That's kind of reassuring.
[17:31]	Gaius Goodliffe:	Actually, it wouldn't surprise me if there are multiple bugs surrounding llTargetOmega. It's always been twitchy...
[17:31]	Jen Shikami:	We wouldn't know that one, Andrew -- as it only just started happening with H4 and I was using windlight and now the release client.
[17:31]	Mifune Thibaud:	TargetOmega is all clientside isn't it?
[17:31]	Andrew Linden:	Well, it is totaly client side, however the simulator has to properly communicate to the client that the prim is supposed to be rotating.
[17:32]	Kooky Jetaime:	Hmmm
[17:32]	Andrew Linden:	so the simulator is involved
[17:32]	Seven Shikami:	If it happens again we'll ask eveyrbody nearby what client they're using but odds are, they'll have the official (now windlight) client.
[17:32]	Andrew Linden:	I'
[17:32]	Jen Shikami:	When the attachments stop, everyone notices and says "Why did it stop?" so... sounds like the simulator stopped saying "spin"?
[17:32]	Kooky Jetaime:	but doesn't the sim only say that it is rotating, its not a constant update
[17:32]	Andrew Linden:	I'd be curious to know if anyone with a non-windlight client also sees the bug.
[17:32]	Kooky Jetaime:	It should only update if it stops in the sims eyes
[17:32]	Renee Faulds:	my rotating signs also are stopping
[17:33]	Jen Shikami:	I didn't think to ask the half-dozen people there what client they used, sorry...
[17:33]	Kooky Jetaime:	and if it does that, then why is it restarting.
[17:33]	Creem Pye:	perhaps it would stop if somebody selected it in edit mode?
[17:33]	Andrew Linden:	Uh oh... I'm going to have to go now.
[17:33]	Gaius Goodliffe:	The Windlight client seems to turn off spin at times, so no, apparently the sim doesn't need to tell it to stop.
[17:33]	Scalar Tardis:	heh if someone selects YOUR attachment that is rotating with omega, should it stop for you when THEY select it?
[17:33]	Kooky Jetaime:	Bye Andrew
[17:33]	Kooky Jetaime:	No Scalar
[17:33]	Renee Faulds:	ty Andrew
[17:33]	Creem Pye:	hey that's how it works with vehicles you're driving <Scalar
[17:33]	Andrew Linden:	Simone or Sidewinder... could you capture the chat logs from this meeting and save them for me?
[17:33]	Jen Shikami:	Clicking to edit, resetting scripts, etc. doesn't seem to restart it at all.
[17:33]	Mitchmiester Slade:	Now I thought that the llTargetOmega was client for non physical and server for physical.. it that correct?
[17:33]	Seven Shikami:	Scalar, I've seen that happen.
[17:33]	Andrew Linden:	So I can publish them on the wiki?
[17:33]	Andrew Linden:	Thanks.
[17:33]	Rex Cronon:	bye andrew
[17:33]	Scalar Tardis:	wave
[17:33]	Andrew Linden:	Sorry, but I've got an appointment to keep.
[17:34]	Jen Shikami:	Thank you Andrew!
[17:34]	Douglas Rishmal:	thanks andrew
[17:34]	RacerX Gullwing:	is sidewinder here?
[17:35]	Simon Linden:	So, if anyone does get some hints on how to make that llTargetOmega problem happen, definitely let us know. Reproducing that is the hardest part
[17:35]	RacerX Gullwing:	i don't see him or see his text
[17:35]	Simon Linden:	I don't see him ... I'm not sure who Andrew was talking to :)
[17:35]	RacerX Gullwing:	spooky
[17:35]	Seven Shikami:	If we figure out what causes the sim-wide, multiple-client omega stop, we'll let you know. Should we also file a bug report the moment we observe it, for tracking purposes?
[17:35]	Cinthya Loveless:	clueless
[17:35]	RacerX Gullwing:	oops
[17:36]	Simon Linden:	There's already a bug filed, so you can probably just add the steps to it. Drop me, Andrew or Sidewinder an IM as well so we're sure to notice it
[17:36]	Jen Shikami:	OK, will do!
[17:36]	Seven Shikami:	Got it. Thanks.
[17:37]	Creem Pye:	since we're all so quiet, I might as well add that sometimes high-speed vehicle dismounts are extremely violent
[17:37]	Simon Linden:	We also know there are some flying issues, such as a slow drift downwards. We put those at a lower priority than the big motion issues that had some content totally broken, but we'll revist them as the dust settles
[17:37]	Creem Pye:	(like being thrown 150m into the air and dying if you happen to be on damage-enabled land)
[17:37]	Seven Shikami:	Whee!
[17:37]	Simon Linden:	Ah, that is a bit rough ... it's from a very fast moving vehicle?
[17:37]	Mifune Thibaud:	it seems that when the agent is intersecting with a prim upon the unsit operation is when it occurs, from what i've noticed at least
[17:37]	Gaius Goodliffe:	Well, ejecting at that speed *should* be fatal, right? ;)
[17:38]	Kitto Flora:	Creem< Thats not anew problem. Happens in H1 too
[17:38]	Seven Shikami:	I've had unusually fast/high ejections when I get up from sculpted seats.
[17:38]	Mitchmiester Slade:	I have a simple swing set that if it is in the up position and moving you will get tossed when dismounting
[17:38]	Mifune Thibaud:	also seems to have a pretty big push from stationary too
[17:38]	Atashi Toshihiko:	I was thrown across a sim and a half when standing up from a sailboat that was *still* at the time
[17:38]	Jezebella Desmoulins:	It's not the speed that kills, it's the stopping.
[17:38]	Creem Pye:	well, I didn't notice it until H4 (so maybe it's more common)
[17:38]	Simon Linden:	That would make sense ... I know we give the AV a little push when they are unseated, but it sounds like something else is hitting them out of the park
[17:38]	Rex Cronon:	flying 4k up should also be fatal?
[17:38]	Gaius Goodliffe:	lol @ Jez
[17:38]	Scalar Tardis:	well, that could just be the collision solver. if you dismount embedded into a prim, it will forcefully push you out
[17:38]	Jen Shikami:	Isn't getting tossed when you leave a swingset at a top kind of like RL? ;)
[17:39]	Creem Pye:	yeah, I figure that's what it is, Scalar
[17:39]	Gaius Goodliffe:	Pushing you out would be nice. Firing you out is maybe a bit much...
[17:39]	Mifune Thibaud:	lol
[17:39]	Jen Shikami:	True. :)
[17:39]	Creem Pye:	but maybe there should be a for the push somewhere below what H4 has now =P
[17:39]	Scalar Tardis:	you may need to unsit and be moved away to not be on/in/embedded
[17:39]	RacerX Gullwing gave you denimtrampoline set the descriptoin to the power.
[17:39]	Yuu Nakamichi:	I guess the issue I brought up - SVC-2014 - is about mass? (An avatar should not stop a vehicle with much larger mass)
[17:39]	Mitchmiester Slade:	Not getting tossed 150m up
[17:39]	Simon Linden:	Very true ... in any case, we probably shouldn't be tossing an AV that much
[17:40]	Scalar Tardis:	the avata seems very lightweight interacting with real physics enabled objects
[17:40]	Scalar Tardis:	er, low mass
[17:41]	Simon Linden:	The mass values in SL are pretty wacky ... we've had to keep them as they were originally made to prevent breaking content
[17:41]	Scalar Tardis:	probably more of a feature request, but since hollow is so hated with H4, we need a way to control an objects mass SEPARATE from merely making it hollow
[17:41]	Simon Linden:	I'll look up SVC-2014 ... 1 sec...
[17:41]	Gaius Goodliffe nods.
[17:42]	Creem Pye:	H4 does offer the ability for different prim material densities...
[17:42]	Creem Pye:	(though it's not implemented)
[17:42]	Tengu Tachikawa:	=O
[17:42]	Gaius Goodliffe:	The physics engine encourages us to make difficult to process shapes.
[17:42]	Simon Linden:	I'm not sure what's going on there, but it definitely sounds like something worth checking out and fixing
[17:42]	Mitchmiester Slade:	Ok.. new issue.. I am new to physics so forgive me if I am just missing something... I have a ride that has a llTargetOmega in the center peice that just turns it... on the ends fo the bar are seats that ar phycial also. The seats keep falling off the bar weven when it is not moving
[17:43]	Simon Linden:	One other hint for you builders ... if you have a complex shape you don't really care about, wrap it in something simple and transparent. We're finding the biggest cost for CPU, memory and physics simulation is complex objects and collisions with them
[17:43]	Yuu Nakamichi:	I'll add a repro to it
[17:43]	Gaius Goodliffe:	I thought enclosing in a prim didn't simplify things any?
[17:44]	Mifune Thibaud:	a side note on the object masses in H4, sculpties still have higher mass than regular prims in H4 so as to not break content? Will this be expected to change?
[17:44]	Simon Linden:	Yuu ... that would be great. Really simple repo steps help us a LOT
[17:44]	Gaius Goodliffe:	(Suggested that at a previous meeting)
[17:44]	Mifune Thibaud:	(since sculpties are still relatively new)
[17:44]	Seven Shikami:	Yeah, someone had a 25 prim physical rubber duck lying around which was causing 20% of our sim's lag.
[17:44]	Jen Shikami:	>_<;
[17:44]	Jezebella Desmoulins:	Damn that Ernie
[17:44]	Scalar Tardis:	it sounds like linked objects in a physical item can now be phantom, so complex objects can be phantom within a simple enclosing prim
[17:44]	Simon Linden:	The one thing that did change is we are doing more accurate mass calculations based on the actual object mesh shape. In H1 there was a lot more fudging on the values, so some have changed and are more accurate now
[17:45]	Yuu Nakamichi:	sure thing. I'll IM you when I have it. I'll speak with the reporter
[17:45]	Scalar Tardis:	though I need to do some more testing of that
[17:45]	Simon Linden:	Thanks .... please add the steps to the Jira bug report too
[17:46]	Creem Pye:	wow you can really have a mixed phantom/nonphantom object now? that's pretty cool
[17:46]	Simon Linden:	Scalar - yes, that's exactly what I mean. Those wave generators mentioned before could have used something like that. I hacked a dev. version of the sim to just simulate them with a box, and I couldn't get it to slow down making lots of waves
[17:46]	Simon Linden:	... while the wave generator would slow it down with just 10 or 20 waves
[17:47]	Seven Shikami:	We only had 3 wave generators and each had one wave live at a time and it still did a bit of slowdown.
[17:47]	Scalar Tardis:	heh I've not tested my popcorn popper yet
[17:47]	Atashi Toshihiko:	I guess there's no chance of 48 or 64 prim dynamic objects any time soon?
[17:47]	Seven Shikami:	So yeah, I'm thinking those are gonna need to be rethought.
[17:47]	Creem Pye:	maybe an object with 32 nonphantom prims but 200 phantom prims could be made dynamic? :)
[17:48]	Simon Linden:	Atashi - you mean increasing the link count limit? We're not looking at that right now
[17:48]	Creem Pye:	since I imagine phantom prims are invisible to the physics engine, mostly...
[17:48]	Atashi Toshihiko:	yeah -- ok Simon, thanks :) Had to ask :)
[17:48]	Gaius Goodliffe:	Per prim phantom would solve a bazillion problems.
[17:48]	Creem Pye:	(maybe just a gravitational force, but nothing else)
[17:48]	Kooky Jetaime:	Lets get things stable before we start adding things.
[17:48]	Mifune Thibaud:	definitely
[17:48]	Popcorn Popper Havok 4 v2.0: Who wants popcorn?
[17:48]	Popcorn Popper Havok 4 v2.0: Awwww, no more popcorn.
[17:48]	Popcorn Popper Havok 4 v2.0: Who wants popcorn?
[17:48]	Simon Linden:	Kooky - that's definitel what we want to do
[17:48]	Popcorn Popper Havok 4 v2.0: Awwww, no more popcorn.
[17:49]	Popcorn Popper Havok 4 v2.0: Who wants popcorn?
[17:49]	Arawn Spitteler:	Fast Hobo Train is borked
[17:49]	Kooky Jetaime:	nice
[17:49]	Kitto Flora:	Sticky popcorn.
[17:49]	Scalar Tardis:	yeah and it's all glass
[17:49]	Simon Linden:	The butter in H4 is thicker
[17:49]	Renee Faulds:	lol
[17:49]	JenzZa Misfit:	... carmel corn ?
[17:50]	Scalar Tardis:	I'm not sure why it sticks as glass
[17:50]	JenzZa Misfit:	weeee !
[17:50]	Kitto Flora:	Too much static?
[17:50]	Gaius Goodliffe:	Cheap butter?
[17:50]	Arawn Spitteler runs to dinner: enjoy my fine photography.
[17:51]	Arawn Spitteler:	SVC-1981
[17:51]	Cinthya Loveless:	that thing is ouchies on physics
[17:51]	Scalar Tardis:	heh it's burning 5 ms
[17:51]	Jezebella Desmoulins:	Gonna take hours for us all to have popcorn.
[17:51]	JenzZa Misfit:	lol
[17:52]	Simon Linden:	There was an interesting effect ... when the sim drops below 20 fps, we drop the level of detail on objects. That changes their shape in the physics world, so they suddenly collide in a different way
[17:52]	JenzZa Misfit:	aww ... movies over ! forget the popcorn * j/k *
[17:52]	Popcorn Popper Havok 4 v2.0: Who wants popcorn?
[17:52]	Rex Cronon:	did u use vegetable oil:)
[17:52]	JenzZa Misfit:	lol
[17:52]	Gaius Goodliffe:	Ah I saw that earlier today. My spiral staircases turned impassible.
[17:52]	Mitchmiester Slade:	BUTTER!!
[17:52]	Rex Cronon:	i mean butter
[17:53]	JenzZa Misfit will serioulsy consider making RL popcorn now
[17:53]	JenzZa Misfit:	seriously*
[17:53]	Popcorn Popper Havok 4 v2.0: Who wants popcorn?
[17:53]	Jen Shikami:	Damn, now I want popcorn too. :D
[17:53]	Simon Linden:	Gaius - are the staircases lots of smaller prims linked together?
[17:53]	Renee Faulds:	mine are
[17:54]	Gaius Goodliffe:	No. Cut/twisted/hollowed tube.
[17:54]	JenzZa Misfit:	.....does sound good ! * unfolding map to my kitchen*
[17:54]	Simon Linden:	We did just fix SVC-2014, but that would have been impassible from the start
[17:54]	Mitchmiester Slade:	Can not use a scupty prim for stairs without using a ramp too...
[17:54]	Scalar Tardis:	well andrew says they don't like interior hollowing so that could be part of the problem
[17:55]	Jen Shikami:	Isn't that bound to happen sooner or later, given the stated goal of getting sculpty collisions closer to the shape rather than assuming a sphere?
[17:55]	Simon Linden:	Yes - it would be best to use the actual mesh shape
[17:55]	Seven Shikami:	Gotta go. Thanks for your assistance on the Omega stuff, lindens... hope you can figure out this weird one. I know how frustrating intermittent errors can be!
[17:56]	Jen Shikami:	Oh, it's time? Yes, thank you all! Bye!
[17:56]	Simon Linden:	We're discussing eventually adding something to LSL (absolutely not designed yet) to let builders give more hints about how they need collisions done for the objects
[17:56]	Simon Linden:	By Jen
[17:56]	Scalar Tardis:	whoah, that isn't me
[17:56]	Seifert Surface:	cheers all
[17:56]	Arawn Spitteler:	Spiral Stairs are borked? They were working when I taught that class last month
[17:57]	Gaius Goodliffe:	They work as long as there isn't too much lag.
[17:57]	Simon Linden:	Gaius - can you enter a jira for that, and drop me or SW a copy? It's probably worth looking at
[17:57]	Gaius Goodliffe:	I thought it was a feature, not a bug. :)
[17:57]	Simon Linden:	There is a size limit on some objects where if they are too large, we don't drop the LOD
[17:58]	Creem Pye:	am I the only one to get stuck between 2 flush 10m cubes? :)
[17:58]	Gaius Goodliffe:	Ah
[17:58]	Gaius Goodliffe:	Yes, these are like over 5m in size. Should they be immune?
[17:58]	Simon Linden:	Originally we had some people in a hollow-box house, and they were ejected rather violently when the sim fps dropped and we made it into a solid
[17:58]	Arawn Spitteler:	There's an entertaining wall between Horsime and Spini
[17:58]	Gaius Goodliffe:	hehe
[17:59]	Simon Linden:	Gaius - yes, they should be. I think the limit was roughly the AV size, to prevent it happening on something like that where you might climb inside
[17:59]	Gaius Goodliffe nods.
[17:59]	Arawn Spitteler:	Seperate prims for each wal will be the new Primficiency
[18:00]	Simon Linden:	OK, the hours pretty much up ... any other last-second questions?
[18:00]	Janet Rossini:	could i inquire about two items, to know whether to report or they are alreaedy in?
[18:00]	Creem Pye:	would a 1m wide prim with a 1% hollow setting always have a simplified detail that plugs the hole?
[18:00]	Rex Cronon:	why do some sims runs slower on h4?
[18:00]	Arawn Spitteler:	http://jira.secondlife.com/browse/SVC-1981
[18:00]	Douglas Rishmal:	Why am I sometimes seeing "sluggish" flight/motion in my home sim (Eduisland 4) when Time Dilation and other stats are normal.
[18:00]	Douglas Rishmal:	?
[18:01]	Creem Pye:	Rex, in the sims I've been to that were slower, the culprit seems to be physical objects
[18:01]	Mifune Thibaud:	yes, and they all happen to be OpenSpaces sims
[18:01]	Mifune Thibaud:	from what I've noticed as well
[18:01]	Simon Linden:	Creem - that sounds like the bug we just fixed (SVC-1531) where we weren't making concave shapes properly for hollowed (or some dimpled) shapes
[18:01]	Rex Cronon:	yes but now they r slower winth only a few thousand objects in them
[18:02]	Rex Cronon:	isn't h4 supposed to make them run faster?
[18:02]	Creem Pye:	ah, I'm just wondering if a very tiny hollow setting (for creating a new "face" for texture storage) would increase the physics lag much
[18:02]	Simon Linden:	We had a memory problem (fixed today) that affected the openspace sims. They were pre-allocating too much, and so fighting each other over the RAM on the server
[18:02]	Gaius Goodliffe:	Actually, I've seen completely empty sims running much slower yesterday and today.
[18:03]	Simon Linden:	Creem - yes, it would be more expensive. If you do that on a box, for example, you'd go from a simple shape to a complex mesh
[18:03]	Arawn Spitteler:	No way to see what else is on the server
[18:03]	Gaius Goodliffe:	Not sure if it's physics related, but it started happening right after the reboot into havok4.
[18:03]	Scalar Tardis:	h4 is supposed to cause less crashes. I don't know if it can improve speed of processing
[18:03]	Gaius Goodliffe:	Alas, SimNeighbors isn't very reliable so you can't check...
[18:03]	Janet Rossini:	is it known that if you rez a series of spheres from, say, 30m up, they will fall down and form a vertical stack?
[18:03]	Rex Cronon:	it think it was supposed to improve speed
[18:04]	Simon Linden:	We're going to work on performance first. We definitely don't want to be slower
[18:04]	Renee Faulds:	ty Simon
[18:04]	Mitchmiester Slade:	Just throw more processors and ram at it :)
[18:05]	JenzZa Misfit:	.. .Andrew ::: on an upnote for you .. my island is an old class 4 * 2 years old in April .. yay ** and ... monday .. had 3 hours of live music .. set it for 60 avatars ... PACKED the whole time .. WITH fireworks ..lol .. and pre H4 that would be crash city -- but it all held up great *** nice work !
[18:05]	Gaius Goodliffe:	Yeah where are those 20GHz Class 6 servers? :)
[18:05]	Douglas Rishmal:	do you think that things like Sun Enterprise, IBM pSeries, or other RISC servers would increase sim performance?
[18:05]	JenzZa Misfit:	er .. Simon
[18:05]	JenzZa Misfit:	*andrew left* sorry
[18:06]	Douglas Rishmal:	ie. stuff like these: http://www-03.ibm.com/systems/p/hardware/midrange_highend/index.html
[18:06]	Douglas Rishmal:	or is it more a code optimization than a raw CPU problem?
[18:06]	Mitchmiester Slade:	I read that IBM is putting some money into thier servers for 3D sims
[18:06]	Simon Linden:	Well, the faster hardware will run faster, and we're constantly adding new racks and retiring old stuff. It's not simple, though, when there are thousands of machines involved.
[18:06]	Douglas Rishmal:	ie. on the LL side... having a "supergrid" (ie. DB2/Oracle asset cluster, pSeries or Enterprise server farm for the sims)
[18:07]	Douglas Rishmal:	Is a lot of the asset server stuff caused by an infrastructure not able to handle the load?
[18:07]	Simon Linden:	In any case, I have to run ... thanks everyone for the feedback and help on this.
[18:07]	Douglas Rishmal:	(ie. MySQL not able to handle such high loads)
[18:07]	JenzZa Misfit:	me too .. good luck everyone :)
[18:07]	Rex Cronon:	bye simon
[18:07]	JenzZa Misfit:	nice !
[18:07]	Scalar Tardis:	it's mostly a money issue, and throwing special hardware at it may increase the cost per sim. That would mean more sims must run on the server to offset the cost, thereby giving no improvement in per-sim performance
[18:07]	Mifune Thibaud:	good luck, and lots of coffee
[18:08]	Kitto Flora:	Good night everyone
[18:08]	Douglas Rishmal:	good luck :)
[18:08]	JenzZa Misfit:	lately I am just eating the coffee beans
[18:08]	JenzZa Misfit:	faster
[18:08]	Arawn Spitteler:	Unless you want to buy by the server, and leave some empty sims
[18:08]	Simon Linden:	Thanks again, and will see you all at other office hours