User:Andrew Linden/Office Hours/2008 04 22

Transcript of Andrew Linden's office hours:

[11:07]	Rex Cronon:	hi everybody
[11:07]	Day Oh:	Heylo
[11:07]	Les White:	hiyas
[11:08]	Sidewinder Linden:	arawn - ou were looking for some viewer guru? or was that a joke i missed?
[11:08]	Arawn Spitteler:	Aybabtu was wondering about Viewer Hours
[11:08]	Sidewinder Linden:	ahh
[11:09]	Andrew Linden:	Sorry I'm late.
[11:09]	Squirrel Wood:	3 logins, 3 immediate crashes, 3 times the crash reporter failed to send the report... :(
[11:09]	Arawn Spitteler recalls that to be the algorhythm by which Windlight gets such a good reliability.
[11:09]	Andrew Linden:	you guys started without me right?
[11:10]	Arawn Spitteler:	Norp
[11:10]	Gaius Goodliffe:	We're lost without you.
[11:10]	Arawn Spitteler:	Is Bato here?
[11:10]	Andrew Linden:	Sidewinder, you were going to make some announcements, I think.
[11:10]	Sidewinder Linden:	we waited for yoooooo
[11:10]	Andrew Linden:	Why don't you start.
[11:10]	Sidewinder Linden:	kk
[11:10]	Sidewinder Linden:	well... we are hopeful to get a deploy of the build that's on the beta preview this week
[11:11]	Sidewinder Linden:	taht build, probably most notably has what we think is a targetomega fix in it
[11:11]	Sidewinder Linden:	it has been on hold for infrastructure issues... the latest is that
[11:11]	Sidewinder Linden:	later today or tomorrow we know if that's been cleared, and that there is a pretty strong likelihood it will get deployed by the end of the week
[11:12]	Sidewinder Linden:	i know there's been a lot of pain from product vendors and users about the targetomega problem,a nd we've been trying to get around some other issues to get this build out onto second life... hopefully this time around we'll have it out by end of week
[11:12]	Sidewinder Linden:	i'll find the url from the beta preview announcement for those who haven't seen the fix list
[11:13]	Day Oh:	Having object selection not be permanent is a good think ok :D
[11:13]	Day Oh:	thing*
[11:13]	Sidewinder Linden:	http://blog.secondlife.com/2008/04/19/second-life-simulator-v121085362-update-is-on-the-beta-preview-with-12-physics-fixes-2008-04-19/
[11:14]	Sidewinder Linden:	and i think that build also includes these fixes: http://blog.secondlife.com/2008/04/06/second-life-simulator-v120084105-pre-release-is-on-the-beta-preview/
[11:14]	Andrew Linden:	alright, I don't really have any announcements. Do you have anything Simon?
[11:15]	Sidewinder Linden:	btw jcool - feel free to take a seat, there's more where that one came from :)
[11:15]	Simon Linden:	No, nothing to annouce
[11:15]	Les White:	one of the sideeffects of non-crashing sims is ghost objects. they cant be seen, but they are solid. you can see them in debug/top scripts, but cant return them. only a sim restart cleans them up
[11:15]	Day Oh:	she's been afk there since I arrived 20 minutes before 11 ^^
[11:15]	Andrew Linden:	Ok then, let's open the discussion to whatever.
[11:16]	Arawn Spitteler:	Bato thought he might have the Prim Seam problem handled.
[11:16]	Andrew Linden:	We've heard reports of ghost prims.
[11:16]	Aybabtu Aabye:	i didnt get a chance to go check this one before i came here
[11:16]	Andrew Linden:	Anybody have clues on how to reproduce them?
[11:16]	Les White:	they have been happening for a LONG time, but crashy sims sorta autocleaned them
[11:16]	Les White:	yes. they are 99% of the time seen to me as vehicles
[11:16]	Gaius Goodliffe:	Deleting an object during heavy lag can often leave behind a ghost.
[11:16]	Aybabtu Aabye:	scripted objects that are flagged as vehicles and then as non-vehicles, do they still keep the zero friction state?
[11:17]	Andrew Linden:	Les, I thought ghost prims had been fixed in Havok1. You're saying that they were not?
[11:17]	Kitto Flora:	I've 'seen' ghost objects for years. Even had Lindens inspect them closely
[11:17]	Heather Goodliffe:	I think it's heavy sim lag combined with taking up or deleting an object
[11:17]	Andrew Linden:	Alright.
[11:17]	Les White:	makes an invis wall on the tracks. not sure how to produce one...only that they happen one or two a day at track locations
[11:17]	Day Oh:	Oh, curiously! I found using the estate tools to return all of a selected avatar's objects removed ghosted prims! But the functionality isn't in the viewer yet--there's a "scripted objects only" flag
[11:17]	Heather Goodliffe:	and/or returning an object
[11:17]	Andrew Linden:	Hrm... the object creation and cleanup code was totally redone in Havok4
[11:17]	Les White:	if you just do return scripted item from debug...they dont poof. i'll try that Day
[11:18]	Gaius Goodliffe:	I've seen it happen at least once since the Havok4 deploy.
[11:18]	Kitto Flora:	Yes, I'd agree - take or delete .. and the ghost stays behind
[11:18]	Heather Goodliffe:	I've seen it a number of times, my sim still crashes a lot though so...
[11:18]	Andrew Linden:	When the simulators are running debug we get some log notices on shutdown of orphaned Havok4 data, if any
[11:18]	Andrew Linden:	however we don't run the deployed simulators in debug mode
[11:19]	Andrew Linden:	and I haven't seen any reports of orphaned data in my own debug runs for a long time
[11:19]	Heather Goodliffe:	you mean sims on the beta grid are in debug and production grid not?
[11:19]	Heather Goodliffe:	or did I misread that?
[11:19]	Andrew Linden:	No, we never run simulators (beta or production) in debug mode
[11:20]	Andrew Linden:	however, when I'm running my own little test world I often run in debug
[11:20]	Andrew Linden:	I'm just thinking out loud...
[11:20]	Kitto Flora:	If ghosts are cause by packet loss, ADITI and internal development grids are not going to show up the problem.
[11:20]	Gaius Goodliffe nods.
[11:20]	Heather Goodliffe:	yeah, that's what I was thinking
[11:20]	Sidewinder Linden:	/@heather - at least so far, the havok folks have strongly discouraged us from using havok debug mode on any public simulator with any real load on it...
[11:20]	Les White:	i'm sure it's produced with a crash or logout/TP out. they happen with vehicles right on the road, seldom in the "parking" areas
[11:20]	Andrew Linden:	I really thought the Havok4 RigidBody cleanup in Havok4 was tight. Sigh...
[11:20]	Heather Goodliffe:	some things would only show up under "real world" setups
[11:21]	Heather Goodliffe:	it may correspond to asset server issues as well?
[11:21]	Arawn Spitteler:	Why not feature a debug mode party, with enough people, in attendance, to produce lag?
[11:22]	Andrew Linden:	No, it would be unrelated to asset. I think the problem is entirely simulator.
[11:22]	Kitto Flora:	I get ghosts of parked steam engines and carriages.
[11:22]	Rex Cronon:	if u want lag just come to swt:)
[11:22]	Arawn Spitteler:	Is that what's called a memory leak?
[11:22]	Andrew Linden:	A debug simulotor test might work... I'd like to have some idea on how to repro the problem first, though.
[11:22]	Aybabtu Aabye:	i had a question on motor decay
[11:22]	Heather Goodliffe:	object leak ;P
[11:22]	Simon Linden:	Kitto - so those are medium/large objects, correct? If so, that's a clue - it sounds like those are too big for the LOD rediction code
[11:22]	Andrew Linden:	Go ahead Aybabtu.
[11:23]	Gaius Goodliffe:	It definitely happens with large objects -- my last ghost was a 20m wide section of a prefab I had deleted.
[11:23]	Kitto Flora:	A steam engine is around 27 prims, 29 scripts. A carriage maybe 20 prims and 15 scripts
[11:23]	Les White:	car sized objects are what i notice Simon. it's cars i find as ghost (living at race tracks)
[11:24]	Aybabtu Aabye:	i noticed when i set motor direction to 0, and motor decay to anything, 0 low or high, that my apply impulse calls eventually get reduced effectiveness after about 10-20 seconds
[11:24]	Heather Goodliffe:	I've seen everythign from mega prims to single prim objects ghosted
[11:24]	Aybabtu Aabye:	i checked energy, its 1.0 as often as i poll for it
[11:24]	Simon Linden:	Kitto - what about size (meters?)
[11:25]	Heather Goodliffe:	you notice the bigger objects because you tend to walk into them
[11:25]	Heather Goodliffe:	but I've seen single prim tiny sphere ghosted too
[11:25]	Aybabtu Aabye:	i checked the strength of the impulse calls, and mass, they dont change either
[11:25]	Kitto Flora:	In the past, so have I - seen one odd little prim ghosted. But that was back in H1 days, year+ back. Such data is a bit stale now
[11:25]	Andrew Linden:	Aybabtu, are the vehicles "ground vehicles"? or hovercraft?
[11:25]	Simon Linden:	We don't change the level of detail on large objects. The LOD code is one place where we are adding and removing shapes from the physics engine, so if there's a bug, it could leave an orphan shape in-world
[11:26]	Day Oh:	The ghosted objects still have keys and an owner, etc, if that means anything
[11:26]	Aybabtu Aabye:	the solution i found so far is to reset the motor direction to 0, that gets me back to full effectiveness on the impulses
[11:26]	Aybabtu Aabye:	ground vehicles
[11:26]	Kitto Flora:	When steam engines ghost, the scripts are still there and functioning. Also 'llSensor()' can see them
[11:26]	Heather Goodliffe:	I've noticed alot of changes in the way impulses work with vehicles
[11:27]	Heather Goodliffe:	generally not so good :/
[11:27]	Gaius Goodliffe:	Yup -- ghost objects actually show up in prim counts in my experience.
[11:27]	Andrew Linden:	Aybabtu, the vehicles have a hidden "dynamic friction" feature.
[11:27]	Andrew Linden:	When the vehicles are trying to move around they get zero friction
[11:27]	Gaius Goodliffe:	You lose those prims until you reboot the simulator.
[11:27]	Aybabtu Aabye:	something beyond the documented friction timescales?
[11:27]	Andrew Linden:	and when they are trying to stop they get normal friction
[11:27]	Andrew Linden:	So my guess is that the friction is defeating your llApplyImpulse() calls
[11:28]	Aybabtu Aabye:	ah ok i think i see now, they slowly gain friction over time after the last call to a motor function?
[11:28]	Andrew Linden:	You might try testing that by setting the material of your vehicle to "glass" (and the road too)
[11:28]	Aybabtu Aabye:	the road is terrain the vehicles are glass
[11:28]	Andrew Linden:	and you should notice that llApplyImpulse() works ok (or better than if you set both to rubber, or concrete)
[11:28]	Andrew Linden:	Hrm... ok already all glass.
[11:29]	Andrew Linden:	well... what is the llApplyImpulse() doing that the vehicle can't do?
[11:29]	Aybabtu Aabye:	is there a way to get the zero friction effect of vehicles without being a vehicle?
[11:29]	Andrew Linden:	I think a workaround would be to not set the angular motor to zero
[11:30]	Gaius Goodliffe:	Set the friction timescale to over 300?
[11:30]	Andrew Linden:	or perhaps... to repeatedly set the linear motor to zero
[11:30]	Aybabtu Aabye:	give it a constant twist then
[11:30]	Andrew Linden:	(sorry, I meant "linear motor" in the earlier statement)
[11:30]	tx Oh:	would be good.
[11:31]	Aybabtu Aabye:	yeah for now im doing that, setting the motor to 0 every so often, but its overhead that im trying to limit
[11:31]	Andrew Linden:	although... angular motor would also do it...
[11:31]	Andrew Linden:	It is not possible to create an object with zero friction without a vehicle script
[11:31]	tx Oh:	how?
[11:31]	Andrew Linden:	and the friction of a vehicle will go back to normal when the vehicle "stops", or tries to stop
[11:31]	Heather Goodliffe:	does LL not expect llApplyImpulse to be used with Vehicles?
[11:31]	Sidewinder Linden:	/sry guys - i have to jump to another commitment... info asap about the deploy of these fixes... bye...
[11:32]	Aybabtu Aabye:	how about a vehicle that has no hidden dynamic friction?
[11:32]	Day Oh:	See you
[11:32]	Les White:	later sidewinder
[11:32]	Jim Brock:	bye
[11:32]	Andrew Linden:	that is, if you keep the vehicle moving around I think the friction never goes to zero
[11:32]	Aybabtu Aabye:	bye sidewinder
[11:32]	Andrew Linden:	I'll have to double check the code
[11:32]	Rex Cronon:	bye sidewinder
[11:32]	Aybabtu Aabye:	ah thats what im experiencing now but i think you explained why
[11:32]	Andrew Linden:	Aybabtu, no the hidden dynamic friction is built-in by default for all vehicle at the moment.
[11:32]	tx Oh:	i would like to see a more realistic friction on no scripted objects
[11:33]	Andrew Linden:	I could disable it with a flag... but I would have to add a new flag to the vehicle feature.
[11:33]	Aybabtu Aabye:	i have a vehicle that keeps moving but experiences mystery friction over time
[11:33]	Les White:	mmm, new flags
[11:33]	Aybabtu Aabye:	but it sounds like its because the last motor call was 30 seconds earlier
[11:33]	Andrew Linden:	tx, what do you mean by "more realistic friction"?
[11:33]	Andrew Linden:	that is, what exactly is unrealistic right now?
[11:34]	Aybabtu Aabye:	friction for non-vehicle objects is extremely high
[11:34]	tx Oh:	actually this stops moving to early. like on my pair of scales...
[11:34]	Aybabtu Aabye:	perhaps if we had a new material, more slippery than glass? or a few materials
[11:34]	Andrew Linden:	hrm...
[11:35]	Andrew Linden:	well, the materials have a global friction setting
[11:35]	Arawn Spitteler doesn't see his choo-choo on minimap.
[11:35]	Aybabtu Aabye:	i'm thinking thats a huge pain though
[11:35]	tx Oh:	they don't really come to a balance
[11:35]	Andrew Linden:	if they are too high they can be lowered
[11:35]	Andrew Linden:	do you know how to measure the friction coefficient beween two objects?
[11:35]	Rex Cronon:	shouldn't materials affect weight of an object
[11:35]	Aybabtu Aabye:	stick friction or slip friction?
[11:35]	Andrew Linden:	although... the balance is a little different...
[11:36]	tx Oh:	it would be good to give this materials a value.
[11:36]	Squirrel Wood:	problem with materials is you can only set single prims but not entire link sets to a specific material. so most things end up eing wood.
[11:36]	Aybabtu Aabye:	how difficult would it be to allow us to set the friction coefficients, within some reasonable bounds
[11:36]	Andrew Linden:	it isn't clear to me how much the coeffiients of friction affect it. Have you tried scales of different materials to see how the friction coefficients come into play tx?
[11:37]	Andrew Linden:	Aybabtu, customizable per-object, and maybe even per-prim, friction coefficients is something I'd like to add..
[11:37]	tx Oh:	not yet. i thought, till now, that the material settings doesn't mean anything.
[11:37]	Heather Goodliffe:	.. perhaps _after_ the havok 4 issues are worked though ;)
[11:38]	Andrew Linden:	however, it would be a pretty big project since it would be adding all sorts of new data to the data format for objects
[11:38]	Gaius Goodliffe:	Throw in a way to set density while you're at it. :)
[11:38]	Heather Goodliffe:	how are we doing on the floating issue?
[11:38]	tx Oh:	hm..
[11:38]	Heather Goodliffe senses scope creep
[11:38]	Aybabtu Aabye:	that wuld be great!
[11:38]	tx Oh:	yes, i see that.
[11:38]	Andrew Linden:	right, if I did friction I'd want to do all of the mass properties: mass, center_of_mass, and inertia_tensor
[11:39]	tx Oh:	but would be nice to have a ice surface, or mud
[11:39]	Andrew Linden:	Heather, you mean "floating avatar issues"?
[11:39]	Gaius Goodliffe votes for the new big project. :)
[11:39]	Les White:	me too!
[11:39]	Andrew Linden:	Oh, no you mean floating surfboard issues.
[11:39]	tx Oh:	i think it's time for consolidation
[11:39]	Arawn Spitteler:	Is there anything weird, about rezzing objects, here, or is the day a bit off?
[11:40]	Kitto Flora:	Asset server seems a bit slow
[11:40]	Andrew Linden:	I haven't started on the surfboard problems yet... I'm currently on the hovering avatar problems and the 768 cap on last_location.
[11:40]	Arawn Spitteler:	I just had two trains dissappear, while editing, and they don't seem to've returned to lost and found
[11:40]	Andrew Linden:	Sometimes I try to work on two or three bugs at once
[11:40]	tx Oh:	we should call a slim your inventory day
[11:40]	tx Oh:	lol
[11:41]	Gaius Goodliffe:	Andrew multitasks! Excellent. :)
[11:41]	Andrew Linden:	but I think I'm about to give up on the last_location problem... I can't find it, and the problem is not in the simulator or client as far as I can tell.
[11:41]	Heather Goodliffe:	focus ;)
[11:41]	Gaius Goodliffe needs a quad-core upgrade for his brain.
[11:41]	Simon Linden:	Heather - I did make some changes that affect collision() events ... those should be working better soon. I"m not sure if that made the upcoming release or not
[11:41]	Heather Goodliffe:	awesome
[11:41]	Jim Brock:	excuse me guys, have to leave... keep up the good work!
[11:41]	Kitto Flora:	I got rid of my float (HOVER_HEIGHT) problems by re-scripting, and adding a user-setable fudge factor.
[11:41]	Andrew Linden:	so that means the problem is in the webservices glue, which is pretty new and written in PHP. My eyes have been boggling on PHP code today
[11:41]	Heather Goodliffe:	I showed simon my colliding phantom objects
[11:42]	Day Oh:	May I bring up a bug? Someone I met has a problem with llLookAt not working in small objects, and I narrowed down her problem case
[11:42]	tx Oh:	urks
[11:42]	tx Oh is a perl evangelist
[11:42]	Gaius Goodliffe uses both.
[11:42]	Kitto Flora:	Simon, does your collision() fix cure the problems in the demo object I gave to Andrew L ?
[11:42]	Heather Goodliffe:	cool Simon :)
[11:42]	Aybabtu Aabye:	can you do perl poetry? :)
[11:42]	tx Oh:	hehe, we made a book about code art
[11:43]	Andrew Linden:	Kitto, I don't think Simon has seen the demo object.
[11:43]	Simon Linden:	Kitto - I don't know, I didn't check that. The bug was that collision_start() and collision_end() would cycle rapidly, with few collision() calls in between.
[11:43]	Heather Goodliffe:	Andrew, web services andphysics engine, I think you lost me?
[11:43]	Simon Linden:	Now it should get the start, collision and end events the way you expect
[11:43]	Kitto Flora:	Ah no - that problems has already been scripted out
[11:43]	Andrew Linden:	there is a 768 cap on your last_location when you log back in, so if you were higher it gets slammed down to 768
[11:43]	Gaius Goodliffe nods.
[11:44]	Andrew Linden:	that is only a "new bug" in Havok4 since we raised the default height limits of stuff in general
[11:44]	Kitto Flora:	Well - maybe can take out some LSL code next time then :)
[11:44]	tx Oh:	oh, thats still their?
[11:44]	Heather Goodliffe:	ah
[11:44]	Andrew Linden:	however... something is slamming the z value to 768 before it gets to the dataserver (a process that manages most of our blocking SQL queries), but I cannot track down where it is happening.
[11:45]	Kitto Flora:	You need a debug expert Andrew :)
[11:45]	Gaius Goodliffe:	Sanity check causing insanity. :)
[11:45]	Andrew Linden:	as to the hovering avatar problem... that is rather mysterious too and isn't a simple bug to fix
[11:45]	Heather Goodliffe:	sinking while hovering, or somelthing else?
[11:46]	Andrew Linden:	the havok engine is reporting that the avatar's velocity is zero... when it is not. Which causes our avatar control code to not push it (since it already has zero velocity in hover)
[11:46]	Rex Cronon:	that is what is happening to me right now. i am sinking. help:)
[11:46]	Simon Linden:	There's a similar bug where you rise slowly instead of hovering - it happens when you're touching a large object, like the side of a building. This goes up, not down
[11:47]	Andrew Linden:	that is the root cause of both the hover-down, and the hover-up when near a large dynamic object
[11:47]	Kitto Flora:	spurious type conversion?
[11:47]	Day Oh:	I've seen a thing where avatars go through something at a fast speed and llDetectedVel it reports 0 velocity
[11:48]	Andrew Linden:	in the rise-up case it is the avatar's anti-gravity that causes it to go up... the contact (or proximity) with the dynamic object cancels some of the real gravity.
[11:48]	Mastorian Kingsford:	couch
[11:48]	Aybabtu Aabye:	speaking of collision start,end,during event calls, is there any synchronization going on in LSL, any possibility of race conditions?
[11:48]	Andrew Linden:	There is a bug where an avatar can cuase a dynamic object to slow down just by proximity, but no touching.
[11:48]	Andrew Linden:	I think that is related to the hover-up problem.
[11:49]	Gaius Goodliffe:	I was noticing the other day that llGetVel was occasionally reporting weird values. Like dropping from 198m/s to 78m/s and back, during continuous flight at constant speed.
[11:49]	Les White:	there's a change with megaprims and server code 1.20.1. you can not sit on anything inside the bounding box. it thinks you are inside the megaprim. it also produces what seems like double physics time for things in the bounding box
[11:49]	Day Oh:	Sent you a thing showing llLookAt not working, Andrew
[11:49]	Andrew Linden:	Gaius, that fluctuation is in the avatar speed, right?
[11:49]	Gaius Goodliffe:	Yup.
[11:49]	Gaius Goodliffe:	This was in an attachment.
[11:49]	fr43k Paine:	Greetings everyone :)
[11:49]	Andrew Linden:	Yeah, there is rather strong damping for the avatar motion at high speeds.
[11:50]	tx Oh:	btw, i saw and object which change status to physic but don't act like it.
[11:50]	Mastorian Kingsford:	lol the things we do for a seat sorry about that
[11:50]	Rex Cronon:	hi
[11:50]	Andrew Linden:	Since the script you're using is running at LSL rates (async with physics engine) you get periods of sudden force application, and the strong damping
[11:50]	Kitto Flora:	tx: An object that physical, but will not move? Acts stuck?
[11:50]	Rex Cronon:	i have seen lots of physics thad don't want to move, doesn't matter what u do to them
[11:50]	Simon Linden:	Aybabtu - I'd guess anything is possible. The tricky parts of the collision code is that we throttle lots of collisions between 2 objects (one for every contact point) down to events that don't occur every frame (which would overload LSL) plus detecting the stop condition - which is figuring out when it's NOT touching, which we don't get events about from the physics engine
[11:50]	Gaius Goodliffe:	makes sense. Had never seen it before, but then I'd never flown 200m/s before either. :)
[11:51]	tx Oh:	kitto, exactly
[11:51]	Andrew Linden:	Rex, we belive the "dynamic objets won't move" problem is related to the "selection bit leak".
[11:51]	Heather Goodliffe:	is it possible the AV hover issue is related to the sinking vehicles?
[11:51]	Squirrel Wood:	1km+/s is a nice speed ^^
[11:51]	Heather Goodliffe:	seems to be a lot of issues around the hovering
[11:51]	Kitto Flora:	Rex : Those 'stuck' physicals - if you take them and re-rez, do they then work right?
[11:51]	Andrew Linden:	Heather, the vehicles always sink, right?
[11:52]	Rex Cronon:	is there anything a user can do to them using scripts or anything else that makes them physical again?
[11:52]	Heather Goodliffe:	someone said there were some that hovered too high, but I haven't seen any
[11:52]	Andrew Linden:	And by "sink" we mean "they sit a little low in water"?
[11:52]	Heather Goodliffe:	all the ones i've seen are sunk, yes
[11:52]	Mastorian Kingsford:	fearhers and other flight controls seem to stop the slow sinking for me in hover
[11:52]	Heather Goodliffe:	jet skis, boats, surf boards
[11:52]	Gaius Goodliffe:	I've seen boats that hover too high.
[11:52]	Heather Goodliffe:	(not just mine) ;P
[11:52]	Andrew Linden:	Rex, you may be able to verify the "selection bit leak" by checking the number of selected prims on the parcel.
[11:52]	Heather Goodliffe:	seems like more hover too low than too high
[11:52]	Rex Cronon:	usually those unmoving physical things are not mine
[11:52]	Gaius Goodliffe:	(0.218m too high in this case.)
[11:53]	Andrew Linden:	I don't know of a 100% reliable fix. Shift dragging will leave a copy behind... the original is the one you moved, so the copy should be selection free.
[11:53]	Andrew Linden:	Restarting the simulator works.
[11:53]	Rex Cronon:	question is: how can u make them move, if u are not the owner?
[11:53]	Kitto Flora:	Andrew : The stuck physical problem keeps showing up for me - What does one look for in 'selected prims'?
[11:53]	Day Oh:	Is there any way for avatars to see whether someone is selecting something?
[11:54]	Heather Goodliffe:	Day, I have a hack for that
[11:54]	Rex Cronon:	there is a particle trail day
[11:54]	Andrew Linden:	I think there is a way to look at the number of "selected prims" on the parcel. In the prim count breakdown.
[11:54]	tx Oh:	but i made the same observation, there is a triggered status change non-physic->physic but nothing happened
[11:54]	tx Oh:	(the flag goes on, thats for sure, but no action)
[11:54]	Andrew Linden:	Rex, I don't know of a fix for the case you're talking about, other than restarting the region.
[11:54]	Kitto Flora:	tx: This is turning into a real BUG.
[11:55]	Andrew Linden:	We hope that some fixes about to get deployed will fix the selection bit leak.
[11:55]	fr43k Paine:	Sorry to interrupt your convo, but may I ask a question about LSL even if it fits in in your current discussion? But I`d really need and like to know that.
[11:55]	fr43k Paine:	doesnt fit*^^#
[11:55]	Andrew Linden:	fr43k, go ahead and state the question.
[11:55]	tx Oh:	when we restart the region it works for a while...
[11:55]	fr43k Paine:	Thank you Andrew
[11:55]	Heather Goodliffe:	don't ask to ask... unless you want to be told no ;)
[11:56]	Andrew Linden scrolls back to see if he missed some questions.
[11:57]	Andrew Linden:	Best to re-ask if your question was not answered.
[11:57]	fr43k Paine:	I got a resident yesterday who asked me if it is possible to gain access to your comp in RL via LSL. Means someone hijacked his PC and deleted his OS (or parts of it). He said he was in a korean sim and got griefed. I told him it is not possbile to do such things via LSL...Am I right or is it possible via modfied client?
[11:57]	Les White:	missed this one... [11:49] You: there's a change with megaprims and server code 1.20.1. you can not sit on anything inside the bounding box. it thinks you are inside the megaprim. it also produces what seems like double physics time for things in the bounding box
[11:57]	Aybabtu Aabye:	is it possible, within a vehicle's script to detect the script load, or physics load of the occupant?
[11:57]	Kitto Flora:	I have had a number of steam trains reported 'stuck'. Only fix we forund was take&Re-rez. I had one that I owned go STUCK. I reset all the scripts in it, set physical by hand... nothing would fix it. Take&Re-rez instantly fixed it.
[11:57]	Les White:	not possible with lsl...possible with moded client
[11:58]	Andrew Linden:	I don't know of any way to exploit someone elses client via LSL.
[11:58]	Day Oh:	I agree, most likely not with LSL
[11:58]	Heather Goodliffe:	fr43k, you're correct, it's not possible and off topic ;)
[11:58]	Andrew Linden:	"he was in a korean sim and he got griefed..." that is a bit vague so it isn't clear how a comprimised viewer was deduced
[11:59]	Arawn Spitteler:	I've heard of a way it could be done, with Streaming video, rather than LSL, but this isn't hte place.
[11:59]	Simon Linden:	Les - I just fixed the 'can't sit inside a phantom megaprim' problem this morning. Won't be in the new version this week, but should be in the next one
[11:59]	Les White:	Hoooo!
[11:59]	Rex Cronon:	that is nice simon
[11:59]	Gaius Goodliffe:	...and there was much rejoicing.
[11:59]	Kitto Flora:	Aybabtu: I never found a way in LSL to check the scripting load of the sitting Av. Sould be nice to.
[11:59]	Les White:	nice work simon
[11:59]	fr43k Paine:	He just said he went there and then his comp started to delete stuff on his HDD...he thinks now it was a SL griefer...
[11:59]	Rex Cronon:	does it apply to regular prims too
[11:59]	Rex Cronon:	?
[12:00]	Gaius Goodliffe:	Coincidence.
[12:00]	Gaius Goodliffe:	Correlation != causality.
[12:00]	fr43k Paine:	Told him the same, but he did not believe me^^
[12:00]	Les White:	mute him!
[12:00]	Andrew Linden:	hrm... that sounds extraodinary, but I'll bounce the report off of a few LL dev's to see if they think it is possible.
[12:01]	fr43k Paine:	Thanks Andrew and everyone else... :)
[12:01]	Simon Linden:	No, it was a case where you were deep inside a phantom megaprim.
[12:01]	Heather Goodliffe:	fr43k, only if someone could've gotten him to download a modified client
[12:01]	Simon Linden:	I didn't try normal prims - you'd be ejected normally
[12:01]	Aybabtu Aabye:	are there office hours for LSL, non-physics?
[12:01]	Arawn Spitteler would like to know about http://jira.secondlife.com/browse/SVC-2141 : I was just trying to edit linked parts, on the fast hobo train, and it could have teleported.
[12:02]	Andrew Linden:	Aybabtu, babbage holds office hours and that would be a good place for LSL questions.
[12:02]	Simon Linden:	I think the Mono team probably has the most interest in LSL issues, and I believe they have office hours
[12:02]	Andrew Linden:	right now he's focused on MONO, but he would also be involved in any major overhaul of LSL itself
[12:02]	Arawn Spitteler suggests Fr43k and Aybabtu hang with the crowd in Grasmere, before Zero Hour.
[12:02]	Andrew Linden:	and most new LSL features are on hold until MONO completes
[12:02]	fr43k Paine:	Ok, thanks a lot, wil ldo that :)
[12:03]	Aybabtu Aabye:	when are babbage's hours, i don't see him on the list
[12:03]	Andrew Linden:	Zero Linden holds office hours, and may have more opinions/theories about how exploitable the SL client is via remote attacks.
[12:03]	Arawn Spitteler:	He also has a gang of techies waiting, before the hour.
[12:03]	tx Oh:	http://wiki.secondlife.com/wiki/Calendar_of_Scheduled_Meetings
[12:04]	Arawn Spitteler:	Said Korean Sim likely has streaming video
[12:04]	Andrew Linden:	I don't know when Babbage's hours are. I thought he had announced them somewhere. I'll ask him (tomorrow, since he in the UK and is probably wrapping up the day already)
[12:04]	fr43k Paine:	Great, appreciate your help...I`ll be quiete now^^
[12:04]	Aybabtu Aabye:	thanks
[12:04]	Mastorian Kingsford:	i would suspect most attacks will be aimed at the new browser in sl since it would allow xml code to run inside the sl client which runs ontop of os
[12:04]	tx Oh:	wrong link: http://wiki.secondlife.com/wiki/Office_Hours
[12:04]	Arawn Spitteler:	Anyhow, Have you looked at the weird flight patterns, of the Fast Hobo Train, Andrew?
[12:04]	Rex Cronon:	tommorow at 8am
[12:04]	Rex Cronon:	on beta
[12:05]	Andrew Linden:	Not yet Arawn, alas.
[12:05]	Aybabtu Aabye:	wasnt there a thing about a quicktime exploit running around SL?
[12:05]	Gaius Goodliffe:	XML "code" isn't executable.
[12:05]	Andrew Linden:	Perhaps I really should look at it... I'd love to find an easy bug.
[12:05]	Arawn Spitteler:	I was just trying it here, the edit parts behaviour, but the trains dissappeared, possibly by TP
[12:05]	Andrew Linden:	All the ones I've been tackling lately have been kicking my butt.
[12:06]	Arawn Spitteler:	It should be easy, or easily distinguished.
[12:06]	Andrew Linden:	Thanks for the reminder Arawn. I'll keep that bug in mind as a next candidate.
[12:06]	Heather Goodliffe:	tp trains?
[12:06]	Jcool410 Wildcat:	o.o
[12:07]	Arawn Spitteler:	I just remembered why my trains dissappeared, I have them set Temp. Temp shouldn't delete what's being edited, though, should it?
[12:07]	Day Oh:	lol welcome jcool
[12:07]	Kitto Flora:	Maybe trains that TP themselves?
[12:07]	Les White:	oops hehe
[12:07]	Gaius Goodliffe:	Aww, the table centerpiece just moved.
[12:07]	Heather Goodliffe:	only when you let go
[12:07]	Andrew Linden:	temp deletes after a minute or two
[12:07]	Arawn Spitteler:	They've been moving and vanishing while editing
[12:07]	Heather Goodliffe:	I get the sense Andrew is eager to get back to fixing bugs ;P
[12:07]	Andrew Linden:	I can't remember if it waits for them to NOT be selected.
[12:08]	Heather Goodliffe:	yeah, it waits
[12:08]	Arawn Spitteler:	It used to wait.
[12:08]	Les White:	i think they will poof while selected...not while sat on
[12:08]	Jcool410 Wildcat:	ty
[12:08]	tx Oh:	i'm just patient because i think most bugs are reported already.
[12:08]	Arawn Spitteler:	Oh, Hi, Jkool
[12:09]	Arawn Spitteler:	Did you see the Bato Solution, to Prim Seams?
[12:09]	Heather Goodliffe:	I'm eager to see the sim performance back up under havok 4
[12:09]	Mastorian Kingsford:	end
[12:09]	Andrew Linden:	I did notice some of Bato's comments on the prim seams problem.
[12:09]	Heather Goodliffe:	sounded like some of the performance issues would be improved in the next sim update?
[12:09]	Andrew Linden:	Some of them made sense, but others did not. I should go back and read the comments more carefully.
[12:09]	Arawn Spitteler:	If he solved it, Havok should be able to fix generally
[12:10]	Kitto Flora:	Temp Objects will vanish whilst being edited.
[12:10]	Gaius Goodliffe nods.
[12:10]	Andrew Linden:	Yeah, there were some very specific performance problems that may be faster... piles of volume_detect objects in particular
[12:10]	Arawn Spitteler:	Temp Objects aren't supposed to vanish while being edited, but they've started doing so.
[12:10]	tx Oh:	make the untemp when you select them for editing
[12:10]	Mastorian Kingsford:	after a hr or two seems seem more noticed now then before thought it was just my setings lol
[12:10]	Aybabtu Aabye:	does the prim seam collision fix affect physics load on the simulator?
[12:10]	Mastorian Kingsford:	seams
[12:11]	Andrew Linden:	Ok Arawn, deleting selected temp_on_rez sounds like a bug.
[12:11]	Arawn Spitteler:	Last I saw, Bato had a demo in Beta Periwinckle.
[12:11]	Andrew Linden:	Bato had a "prim seams collision workaround" for non-vehicles, I think.
[12:12]	Gaius Goodliffe:	They've vanished while editted for a long time. I think they've done so since they started disappearing after a minute instead of all disappearing in huge clumps less often (this change was over a year ago IIRC)
[12:12]	Heather Goodliffe:	would be cool if selecting them reset the temp-rez clean up time fo rthem
[12:12]	Heather Goodliffe:	I always forget and let go and it disappears right when I'm trying to take a copy
[12:12]	Arawn Spitteler:	I think it used to.
[12:12]	Andrew Linden:	Heather, can you think of any grief applications of a temp timer reset on selection?
[12:13]	Heather Goodliffe:	only if you had a bot that went around selecting everything that was temp-rez
[12:13]	Heather Goodliffe:	heh
[12:13]	Gaius Goodliffe:	How about if the OWNER selects.
[12:13]	Les White:	i'm not sure htey ever stuck around while selected
[12:13]	Heather Goodliffe:	seems like there are easier ways to grief that that ;P
[12:13]	Gaius Goodliffe:	There's a number of things that are a good idea to do when the owner selects, but not when just anyone selects (e.g. stop physics).
[12:13]	Arawn Spitteler:	Bato's solution is to use a script to set a Zen Taper, so it sounds a squashable bug, and Havok should have squashed it too long ago
[12:13]	Heather Goodliffe:	Les, I'm reptty sure they did
[12:13]	Andrew Linden:	Yeah, good point Heather.
[12:14]	Heather Goodliffe:	I work with alot of temp-rez stuff
[12:14]	Day Oh:	would be a great way to temp-rez a mansion
[12:14]	tx Oh:	i have a temp rez klub
[12:14]	Day Oh:	have your camping bot select it
[12:14]	Andrew Linden:	A bot, or a hacked viewer could keep their nearby army of temp-on-rez objects alive
[12:14]	Heather Goodliffe:	people already re-rez stuff
[12:15]	Jcool410 Wildcat:	if they had such aviewer they could just rerez the stuff
[12:15]	Rex Cronon:	would it be a good way to deny others the obility to rezz?
[12:15]	Heather Goodliffe:	I think rezzing is a bigger impact on the systme than keeping somethign temp-rez around
[12:15]	Les White:	they do that now with rezzers
[12:15]	Gaius Goodliffe:	It's already pretty easy to keep temp-rez objects alive.
[12:15]	Les White:	temp on rez was an evil i was glad to leave behind on the mainland o-o
[12:15]	Gaius Goodliffe:	Moving temp-rez objects can keep going pretty much indefinitely.
[12:15]	Gaius Goodliffe:	As long as they aren't trapped into a single sim.
[12:16]	tx Oh:	i love temp on rez
[12:16]	Les White:	to beat prim limits?
[12:16]	Andrew Linden:	Ah gaius, you mean moving temp-on-rez across region boundaries and back?
[12:16]	Gaius Goodliffe nods.
[12:16]	Andrew Linden:	that should be fixable, actually.
[12:17]	Gaius Goodliffe:	Les: To beat lag.
[12:17]	Heather Goodliffe:	there goes tthat hack ;P
[12:17]	Heather Goodliffe:	wtg ;P
[12:17]	Gaius Goodliffe:	There's no good reason for bullets, for example, to be created as permanent resources.
[12:17]	Day Oh:	there's a pretty prominant viewer bug that leaves things selected
[12:17]	Kitto Flora:	I've had complaint about poor sim performance, turned out to be due to these temp-on-rez hacks to beat prim limits.
[12:18]	Heather Goodliffe:	it's more the re-rezz-ing than the "temp-rezzing"
[12:18]	Arawn Spitteler:	Shut down scripts, for the return time
[12:18]	Andrew Linden:	ah yes, the everlasting temp-on-rez decorations
[12:18]	Gaius Goodliffe:	There is no useful feature that won't be abused. If it's useful, someone will find a use for it who has less than pure motives. That rarely justifies stripping out useful features.
[12:18]	Heather Goodliffe:	if I had a suggestion, it would be to make it so sim's cached a tmeplate for objects they just rezzed
[12:19]	Heather Goodliffe:	sims/viewers
[12:19]	Heather Goodliffe:	rezzing seems like it takes such a huge load on everything
[12:19]	Les White:	yes
[12:19]	Heather Goodliffe:	and more of the time it's just the same stuff being rezzed over and over again
[12:19]	Andrew Linden:	I'd first like to try to optimize rezzing.
[12:19]	Heather Goodliffe:	bullets, waves, etc
[12:19]	Heather Goodliffe:	yeah
[12:20]	Andrew Linden:	perhaps that what you meant by "template"? caching of info for rezzing?
[12:20]	Jcool410 Wildcat:	dont they already do that
[12:20]	Kitto Flora:	Andrew: Optimize rezzing on border crossing! :)
[12:20]	Heather Goodliffe:	I mean like, once you created the "model" of the object or whatever
[12:20]	Andrew Linden:	There is some caching for rezzing assets, but not enough, and not necessarily in the right places.
[12:20]	Heather Goodliffe:	without having a specific location or instance
[12:20]	Aybabtu Aabye:	is there a way to cap the number of objects on a sim owned by any one av, including attachments?
[12:20]	Arawn Spitteler recalls that Classes are Objects of Class Class, while Objects are Objects of Class Object
[12:20]	Heather Goodliffe:	cahcing comes with its own set of issues ofcourse
[12:20]	Andrew Linden:	Yeah I gotcha Heather. I think we're thinking the same thing,.
[12:20]	Heather Goodliffe:	* caching
[12:20]	Heather Goodliffe:	makes debugging a hell of a lot harder
[12:21]	Heather Goodliffe:	always got to add a switch to turn on/off caching ;)
[12:21]	Gaius Goodliffe:	Caching also increases memory footprint.
[12:21]	Gaius Goodliffe:	Especailly when you're not sure what's best to cache.
[12:21]	Arawn Spitteler:	Attachments aren't counted by the Physics Engine, but the Script Engine would notice soem.
[12:22]	Gaius Goodliffe:	temp-on-rez objects are a good bet, though. That's hint and a half that it's going to be rerezzed soon.
[12:22]	Andrew Linden:	Ok I'm going to head out. Thanks for comming.
[12:22]	Heather Goodliffe:	yeah, for sure temp-rez stuff
[12:22]	Les White:	thanks for the time Andrew. good stuff
[12:22]	Heather Goodliffe:	thanks for being here :)
[12:22]	Gaius Goodliffe:	Take care
[12:22]	Aybabtu Aabye:	thanks andrew
[12:22]	Day Oh:	Thanks Andrew
[12:22]	tx Oh:	ahoi
[12:23]	Rex Cronon:	we nee llGetAttachedPrimNr(key av), and llGetAttachedSripts(key av)
[12:23]	Arawn Spitteler:	Temp on Rez Objects could have Templates displayed as part of the Rezzer, as a characteristic, rather than a pre-load command.
[12:23]	Rex Cronon:	bye andrew
[12:23]	Les White:	sure do rex!
[12:23]	Gaius Goodliffe:	O