User:Andrew Linden/Office Hours/2008 06 17

Transcript of Andrew Linden's office hours:

[11:05]	Rex Cronon:	for one i am not downey:)
[11:05]	Rex Cronon:	and this is sl, anything is possible:)
[11:05]	Teravus Ousley:	hehe
[11:06]	Teravus Ousley:	yay for suspending disbelief in movies
[11:07]	Andrew Linden:	wow, I crashed out... I haven't crashed my SL client in a year at least...
[11:07]	Andrew Linden:	er... not the windows version anyway
[11:07]	Ilsa Munro:	Show off
[11:07]	Johan Laurasia:	lol
[11:07]	Ilsa Munro grins
[11:07]	Rex Cronon:	welcome to the real world:)
[11:07]	Johan Laurasia:	one would think working on all the latest builds, you'd crash alot
[11:08]	Andrew Linden:	I've been jumping between some "dazzle" versions of the SL client and the legacy client... and my graphics settings get blitzed each time I login
[11:08]	Ilsa Munro:	Well he did say "not the Windos version" - perhaps he's normally a Mac or Linux user
[11:08]	Andrew Linden:	I crashed the linux version yesterday
[11:09]	Gaius Goodliffe:	I've long suspected I would't be crashing constantly if I didn't alt-cam a lot. It always seems to crash while I'm alt-camming. But tht could just be because I'm always alt-camming
[11:09]	Johan Laurasia:	when the next release comes out, will it be skinnable? or will it be 'dazzle' only?
[11:09]	Andrew Linden:	When it comes to announcements... first of all I'll be going on vacation soon..
[11:09]	Andrew Linden:	and won't be able to attend the Thursday hour this week
[11:10]	Andrew Linden:	or any of the hours next week
[11:10]	Andrew Linden:	I'll ask Simon if he will attend...
[11:10]	Ilsa Munro:	Hope you're going someplace great
[11:11]	Teravus Ousley:	yes, though I suspect time away from the avatar control code will you you great :D
[11:11]	Rex Cronon:	i guess u do need some vacation time, after spending so much time in havok:)
[11:11]	Andrew Linden:	Last hour someone was talking about the bug where bullets were passing through the avatar...
[11:12]	Andrew Linden:	I tried to make some headway on that but last week and yesterday but failed to figure it out
[11:12]	Andrew Linden:	so that bug probably won't be fixed until I get back
[11:12]	Rex Cronon:	lag might play part in it
[11:12]	Andrew Linden:	No... I don't think it is lag
[11:12]	Teravus Ousley:	bullet with too much velocity! :D
[11:13]	Andrew Linden:	but what I need to do is regress and find a version where the bugs don't happen
[11:13]	Johan Laurasia:	full metal jacket
[11:13]	Andrew Linden:	well... could be that the bullet has too much velocity... but I'm according to the reports something changed in 1.22
[11:14]	Teravus Ousley:	frame 1: 2m in front of the avatar, frame 2 two meters behind the avatar..
[11:14]	Andrew Linden:	Lessee... RacerX, you gave me a notecard complaining about bugs that were affecting the snail races
[11:15]	RacerX Gullwing:	right you said you'd look into it howecd that go?
[11:15]	Andrew Linden:	SVC-2511 is fixed in an internal branch
[11:15]	Andrew Linden:	however the other two bugs to do with llPushObject are not fixed, as far as I know
[11:16]	Andrew Linden:	That is... https://jira.secondlife.com/browse/SVC-1179
[11:16]	Andrew Linden:	and https://jira.secondlife.com/browse/SVC-875
[11:16]	RacerX Gullwing:	right the unacined 6 month o.ld with 120 votes
[11:17]	Andrew Linden:	Hrm... I thought maybe Kelly had made some changes that would improve SVC-875... but that was done a while ago... a month or more
[11:17]	Andrew Linden:	should be deployed already
[11:18]	RacerX Gullwing:	did you get a chance to look at the films?
[11:18]	Andrew Linden:	It would be nice to have a simple demo script that doesn't work in Havok4 but does in Havok1
[11:18]	Andrew Linden:	Films? Video clips? No not yet.
[11:18]	RacerX Gullwing:	ok pass you one
[11:19]	Andrew Linden:	RacerX, by simple script I mean a simple "touch this object and it pushes the avatar so-hard" script that could be pasted into the SVC item
[11:20]	Andrew Linden:	Not a full-fledged object in my inventory
[11:20]	Andrew Linden:	but the object in inventory will work in a pinch
[11:20]	Andrew Linden:	Ilsa, you had some questions about a vehicle you were trying to make
[11:21]	Andrew Linden:	a balloon of some sort?
[11:21]	Ilsa Munro:	Yep, though I think I jusdst need a nudge in the right direction figuring out all the myriad vehicle variables
[11:21]	Ilsa Munro:	The one that's been killing me lately is doing away with pitch and roll.
[11:22]	Ilsa Munro:	I'd prefer none or just a little but I get nearly 90%
[11:22]	Andrew Linden:	Ok... your balloon was rolling too much... you probably want to use a little bit of "vertical attractor" to keep the top up
[11:22]	RacerX Gullwing gave you cannon1.
[11:22]	Andrew Linden:	the vertical attactor can be made weak by increasing its "timescale"
[11:22]	Ilsa Munro nods
[11:22]	Andrew Linden:	the "timescale" of vehicle features are basically how long it taks the feature to accomplish its
[11:23]	Andrew Linden:	speed or whatever
[11:23]	Creem Pye:	also don't forget to remove the VEHICLE_FLAG_LIMIT_ROLL_ONLY flag
[11:23]	Andrew Linden:	when it comes to damping angular motion...
[11:23]	Andrew Linden:	the vehicle has an angular friction feature...
[11:23]	Andrew Linden:	I actually have to look up the name
[11:24]	Andrew Linden:	I think it is VEHICLE_ANGULAR_FRICTION_TIMESCALE
[11:24]	Ilsa Munro:	No worries, I can find it - I've got I think every web site with the vehicle stuff mentioned bookmarked
[11:24]	Andrew Linden:	but it is a vector, and each component represents how strong the friction is around the relevant axis
[11:25]	Andrew Linden:	so... for lots of friction about the roll, but non about pitch and yaw, set it to something like <1, 1000, 1000>
[11:25]	Ilsa Munro:	Ahhh, so a high Y and Z would help out?
[11:25]	Ilsa Munro:	Ahh, what you said :)
[11:25]	Andrew Linden:	so the friction takes about 1 second to kick in for roll, but 1000 seconds to kick in for pitch (Y) and yaw (Z)
[11:25]	Andrew Linden:	yup
[11:25]	Ilsa Munro:	Excellent, thanks
[11:25]	Andrew Linden:	any other questions that weren't in the notecard?
[11:25]	Creem Pye:	http://lslwiki.net/lslwiki/wakka.php?wakka=LindenVehicleTutorial this tutorial is pretty dece3nt
[11:26]	Johan Laurasia:	I just started a balloon myself recently, glad this all came up
[11:26]	Andrew Linden:	Ilsa said she read the tutorial but it wasn't clear
[11:26]	Ilsa Munro:	Umm.. All of them is too broad an answer isn't it :)
[11:26]	Andrew Linden:	I wrote the original tutorial since I wrote the SL vehicle feature
[11:26]	Ilsa Munro:	I'm not swell at math and physics
[11:26]	Ilsa Munro:	Normally I get by by just changing things and seeing what happens
[11:27]	Andrew Linden:	so the tutorial makes sens to me... but not everybody else ;-)
[11:27]	Ilsa Munro:	But there are elevent zillion vehicle settings so my usual method hasn't worked too well
[11:27]	Johan Laurasia:	lol
[11:27]	Andrew Linden:	right, and some vehicle features can "defeat" others
[11:27]	Ilsa Munro:	I've made it do some interesting things :)
[11:27]	RacerX Gullwing:	so how long till a rollout that will fix SVC-2511?
[11:28]	Andrew Linden:	if you put too much linear friction it can prevent the motor from working right
[11:28]	Teravus Ousley:	yay for options:D
[11:28]	Andrew Linden:	RacerX, the fixes are still in QA now, but are wrapping up
[11:28]	Johan Laurasia:	you think a balloon might be better done with controlling bouyancy and just pushing it around?
[11:29]	Andrew Linden:	so the fixes would make the deadline for a next week release
[11:29]	Andrew Linden:	unfortunately... rumor has it that the deploy group in LL doesn't want to push out a new server next seek
[11:29]	Andrew Linden:	er... week
[11:29]	Creem Pye:	hmm the vertical attractor would probably be very useful for you, Johan
[11:29]	Andrew Linden:	No Johan, the vehicle has a buoyancy feature that should work
[11:29]	Andrew Linden:	a balloon vehicle is doable
[11:30]	Creem Pye:	you could always use linear impulses to move, but still have vehicle dynamics to affect flight
[11:30]	RacerX Gullwing:	ok well soon would be good 2 saturdays till we use the track again next week were on linden roads doing the cross country race
[11:30]	Andrew Linden:	So RacerX, the fix for SVC-2511 might not be deployed for two weaks
[11:30]	Teravus Ousley pictures an LSL 'desired position----> llApplyImpulse' PID :D
[11:30]	Andrew Linden:	er... weeks
[11:31]	RacerX Gullwing:	ok good nuff
[11:31]	RacerX Gullwing:	i gota go
[11:31]	Ilsa Munro:	Other than my pitch and roll issues I've done well with a Cubey based script Johan
[11:31]	Johan Laurasia:	there's been talk in the forums about a new llMoveToTarget() but personally I havent had any troubles getting it to do what I want
[11:31]	Andrew Linden:	new llMoveToTarget()?
[11:31]	Johan Laurasia:	their complain is no control over damping, but I just set a decent target, like 1.5 meters or so, and there's little damping
[11:32]	Andrew Linden:	as in... a new feature coming out? or a new feature that has been released?
[11:32]	Jillian Callahan:	As in a change they'd like
[11:32]	Johan Laurasia:	well, they're asking for it, but I dont see the need personally
[11:33]	Andrew Linden:	oh... llMoveToTarget() uses a "critically damped" velocity model...
[11:33]	Johan Laurasia:	yes
[11:33]	Andrew Linden:	which has an exponential decay shape
[11:33]	Johan Laurasia:	right, but if you just set a target of 1.5 meters or so the at_target triggers before too much damping occurs
[11:33]	Andrew Linden:	I can see the need
[11:33]	Creem Pye:	do they want an option for slow-timescale movement that still has a fairly strong force?
[11:33]	Andrew Linden:	That is a feature I'd like to work on
[11:34]	Jillian Callahan:	Not so much need as it's be convenient and easier.
[11:34]	Andrew Linden:	It would be nice to have a slow-fast-slow approach model, and also a linear speed move model
[11:34]	Gaius Goodliffe nods.
[11:35]	Johan Laurasia:	the linear speed would be nice, but I've more or less coded my way around it
[11:35]	Andrew Linden:	Ok, anybody have new issues or questions for the table?
[11:36]	Johan Laurasia:	any idea on how long before shadows get released? The demo torley showed was really neat
[11:36]	Andrew Linden:	Shadows do look great, but that is totally experimental.
[11:37]	Andrew Linden:	I'd suspect it would be a month at least before they were an option in a release-candidate SL client
[11:37]	Andrew Linden:	The right person to ask is Runitai Linden
[11:37]	Ilsa Munro:	Out of curiosity will the shadows be detectable by a script?
[11:37]	Andrew Linden:	but I'll try to ask him should I see him
[11:37]	Andrew Linden:	No they would be totally client-side
[11:38]	Ilsa Munro nods
[11:38]	Andrew Linden:	Ilsa, do you mean... script can detect when hit by a shadow? or script can detact when some avatar is using shadows in their client?
[11:38]	Creem Pye:	Andrew, last week you mentioned somebody's request for lower latency on avatar movement. I thought about it a bit, and maybe he's asking for more avatar movement prediction in the client. That would be possible, I think, if the client had a crude physics engine working to predict how sim-side physics would react
[11:39]	Andrew Linden:	Sigh Ceem...
[11:39]	Ilsa Munro:	I guess either though thinking about it I can see where the former would be really tought to do
[11:39]	Creem Pye:	of course with H4 you probably can't copy the physics engine to the viewer
[11:39]	Andrew Linden:	Creem.. that is exactly what I was talking about... yes he was asking for client-side prediction
[11:40]	Teravus Ousley looks at CPU usage with no physics engine..
[11:40]	Andrew Linden:	however my *point* was that client-side prediction doesn't make sense unless you first tighten the avatar response on the server
[11:40]	Andrew Linden:	everybody asks for client-side prediction
[11:40]	Teravus Ousley:	not that it's *that* cpu intensive to simulate a tori, a heightfield and a few surrounding objects.
[11:41]	Creem Pye:	I'd expect collisions between the avatar and static prims would be difficuolt to replicate 100% as well, unles the physics engines used almost the exact same code
[11:41]	Andrew Linden:	here's a little tidbit... Qarl said he was working on some ray-trace vs LLVolume collision query code in the SL client
[11:42]	Teravus Ousley:	ray ---> triangle derived plane?
[11:43]	Andrew Linden:	if he were to complete that then it might be useful for a few features... including, perhaps, helping with client-side avatar motion prediction when hitting obstacles and stairs
[11:43]	Andrew Linden:	LLVolume is the C++ class that holds the geometry info for the various primitives in SL clietn
[11:43]	Creem Pye:	that would be nice as an LSL function too, something like a narrow sensor that tells you the UUID of the first object the beam "hits"
[11:44]	Andrew Linden:	well, server side we use the Havok engine for ray-trace tests
[11:44]	Teravus Ousley:	yeah, I'll bet it's a lot faster :D.
[11:44]	Jillian Callahan:	Is there a chance we'll get a little of that in LSL?
[11:44]	Teravus Ousley:	I wrote an OBB raycaster for rezzing things in OpenSimulator.
[11:45]	Teravus Ousley:	.. it's fast, except in our case.. I re-get the OBB each time.. instead of caching it.
[11:45]	Teravus Ousley:	.. so that means decomposing the object over and over again.
[11:45]	Andrew Linden:	OBB = Object Bounding Box
[11:45]	Teravus Ousley:	Oriented Bounding Box
[11:45]	Andrew Linden:	oh right
[11:46]	Andrew Linden:	as opposed to Axis Aligned Bounding Box (AABB)
[11:46]	Teravus Ousley:	right.
[11:46]	Andrew Linden:	what are you using the OBB ray-cast test for?
[11:46]	Teravus Ousley:	rezzing and DuplicateOnRay... (Copy Selection)
[11:47]	Teravus Ousley:	.. when you rez, you have to offset the prim based on the raycast
[11:47]	Andrew Linden:	Ah. I think we use GLPick for most rez casting
[11:47]	Andrew Linden:	but I think there are inacuracies with the methods we use
[11:48]	Gaius Goodliffe:	Half the time it tells me "unable to rez".
[11:48]	Andrew Linden:	And Qarl was talking about GLPick vs his ray-cast, so perhaps the improved ray cast will replace some of the GLPick cases
[11:48]	Creem Pye:	yeah, it's worse when you try to rez on megaprims too - usually in that case I find myself needing to reposition the camera to face the floor megaprim I'm trying to rez an object on
[11:48]	Andrew Linden:	but yes, his cat hits the various triangles of the shape...
[11:49]	Andrew Linden:	he transforms the ray into the local space of the primitive geometry
[11:49]	Teravus Ousley:	.cool :D
[11:49]	Gaius Goodliffe:	I often find I can create a new box just fine, then drag the object I want to rez onto the box.
[11:49]	Johan Laurasia:	I do the same
[11:49]	Johan Laurasia:	my platfom is made from 20x20's so I have a 10x10 just to rez onto
[11:49]	Andrew Linden:	Did someone say megaprims don't work quite right?
[11:49]	Gaius Goodliffe:	Rezzing onto.
[11:50]	Creem Pye:	well, rezzing objects on top of 20x20m megaprim platforms
[11:50]	Dahlia Trimble:	seems easier since H4
[11:50]	Andrew Linden:	hrm... I need more sarcasm sprinkles for my chat
[11:50]	Gaius Goodliffe:	I get the same errors, though, even when there are no megaprims involved.
[11:50]	Teravus Ousley:	hehe
[11:50]	Johan Laurasia:	heh
[11:50]	Gaius Goodliffe:	hehe
[11:51]	Teravus Ousley:	</sarcasm>
[11:51]	Andrew Linden:	The hour was going so sell and then someone brought up megaprims
[11:51]	Johan Laurasia:	lol
[11:51]	Ilsa Munro laughs
[11:51]	Andrew Linden:	just kidding... I wish I could work on megaprims
[11:52]	Andrew Linden:	but all these little emergencies keep showing up
[11:52]	Johan Laurasia:	running around putting out first eh Andrew?
[11:52]	Teravus Ousley:	hehe
[11:52]	Dahlia Trimble:	delegate ;)
[11:52]	Johan Laurasia:	fires*
[11:52]	Teravus Ousley:	fireman!
[11:52]	Andrew Linden:	wow, my typo rate is terrible with this keyboard
[11:52]	Gaius Goodliffe:	Andrew needs his own fire brigade. :)
[11:52]	Johan Laurasia:	lol
[11:52]	Teravus Ousley has a friend.. who says.. Physics is hardn' stuff.
[11:53]	Johan Laurasia:	Physics is Phun
[11:53]	Gaius Goodliffe:	Physics is a very dynamic field. :)
[11:53]	Teravus Ousley:	Phun is only 2D physics though .....
[11:53]	Simon Linden:	Sorry to cut and run ... gotta go early. Bye everyone
[11:53]	Johan Laurasia:	adios Simon
[11:53]	Ilsa Munro:	By eSimon
[11:53]	Teravus Ousley:	ok, take care Simon
[11:53]	Creem Pye:	seeya Simon
[11:53]	Dahlia Trimble:	bye :)
[11:53]	Andrew Linden:	So Simon, you going to run the office hours on Thursday and next week?
[11:54]	Andrew Linden:	Ooff, he logged off
[11:54]	Creem Pye:	there's your answer, maybe =P
[11:54]	Andrew Linden:	I mentioned earlier, but should re-iterate for those who may have showed up late...
[11:54]	Rex Cronon:	what happened?
[11:54]	Gaius Goodliffe:	Hopefully he can show up. We need *someone* to harass. ;)
[11:54]	Andrew Linden:	I won't be able to attend these office hours on Thursday, and all of next week.
[11:55]	Andrew Linden:	However Simon is going to try to remember to show up.
[11:55]	Rex Cronon:	i had silent crash:(
[11:55]	Teravus Ousley:	:D yes.. taking a needed break from avatar control.
[11:55]	Teravus Ousley:	:D
[11:55]	Andrew Linden:	Someone should send him an IM about one hour before the meetings
[11:55]	Andrew Linden:	or maybe at the beginning of the meetings
[11:55]	Jillian Callahan:	Remind him google calendar can send reminder e-mails =^.^=
[11:56]	Creem Pye:	Andrew, out of curiosity, do you and Simon work from an office?
[11:56]	Teravus Ousley finds it irritating that Linux/windows/max need different Angular Motor tensors to keep the avatar capsule standing up in oDE..
[11:56]	Andrew Linden:	Creem, I work in the SF office. Simon works out of Davis
[11:56]	Teravus Ousley:	max-mac
[11:57]	Andrew Linden:	Every two weeks or so Simon shows up in SF
[11:57]	Andrew Linden:	and I go to Davies once in a while... once every few months.
[11:57]	Andrew Linden:	Davis
[11:57]	Johan Laurasia:	Davis is where?
[11:57]	Creem Pye:	ah I htought you worked from Seattle for some reason
[11:57]	Andrew Linden:	Davis, CA
[11:58]	Creem Pye:	but SF is where most of the real work gets done, eh? =)
[11:58]	Andrew Linden:	SF is the biggest office, however we are focusing hiring in the remote offices
[11:58]	Andrew Linden:	SF is out of room
[11:58]	Ilsa Munro:	Ugh, speaking of which I have to get back to reality myself
[11:58]	Andrew Linden:	Boston, UK, and Mountain View are where we're hiring
[11:58]	Gaius Goodliffe:	Any plans for a Minneapolis office? :)
[11:59]	Ilsa Munro:	Thanks so much Andrew, and have a great vacation
[11:59]	Andrew Linden:	and Davis.
[11:59]	Johan Laurasia:	I'm all about the Cleveland office :)
[11:59]	Teravus Ousley:	:D Yes have a great vacation.
[11:59]	Andrew Linden:	You're welcome Ilsa, good luck.
[11:59]	Rex Cronon:	have fun:)
[11:59]	Ilsa Munro waves
[11:59]	Jillian Callahan:	Ciao
[11:59]	Rex Cronon:	bye everybody
[11:59]	Andrew Linden:	Heh... we might have a Chicago office in another year or two
[12:00]	Gaius Goodliffe:	Cool
[12:00]	Andrew Linden:	I think we've got two people in Chicago
[12:00]	Johan Laurasia:	argh, Cleveland is better :)
[12:00]	Johan Laurasia:	great location
[12:00]	Teravus Ousley:	:D
[12:00]	Gaius Goodliffe:	Minneapolis would be best. People want to stay inside and work on code more. :)
[12:00]	Dahlia Trimble:	lol
[12:01]	Teravus Ousley:	k, time for me to head out. Have a good time, Andrew. See you in ~2 weeks :D
[12:01]	Andrew Linden:	If we hired anotheh two or three people in Chicago then it would make sense to provide them with a central office
[12:01]	Dahlia Trimble:	bye :)
[12:01]	Andrew Linden:	We'd have to find a few people in Cleveland before we opened the office there.
[12:01]	Johan Laurasia:	uh huh
[12:01]	Andrew Linden:	laters Teravus
[12:01]	Dahlia Trimble:	bye all :)
[12:02]	Johan Laurasia:	Well, enjoy your vacation Andrew, hard earned I'm sure
[12:02]	Andrew Linden:	any last minute issues or questions before we adjourn?
[12:02]	Andrew Linden:	Thanks Johan.
[12:02]	Gaius Goodliffe:	Thank you Andrew. Have fun :)
[12:03]	Creem Pye:	thanks for your time!\
[12:03]	Andrew Linden:	alright, thanks for coming everyone.
[12:03]	Johan Laurasia:	k, see ya when u get back :)
[12:03]	Andrew Linden:	yup
[12:03]	Johan Laurasia:	we'll hold the fort down while you're gone :)