User:Andrew Linden/Office Hours/2008 05 03

Transcript of Andrew Linden's office hours:

[10:59]	Arawn Spitteler isn't clinically Agoraphobic, but considers it a worthy spiritual development: I could probably cure anything, if I could just find someone willing to tolerate my presence.
[10:59]	Sidewinder Linden gave you Personal Hugger (90sec).
[10:59]	Mathieu Basiat:	lol
[10:59]	Arawn Spitteler:	Have you been to Mystic Academy?
[10:59]	Rex Cronon:	helloe everybody
[10:59]	Mathieu Basiat:	nope
[11:00]	Mathieu Basiat:	hey Andrew
[11:00]	Rex Cronon:	hey andrew
[11:00]	Andrew Linden:	hello
[11:00]	Rex Cronon:	btw, to u have any announcements today, andrew?
[11:00]	Andrew Linden:	I have a few. Nothing impressive.
[11:01]	Andrew Linden:	BTW Seifert, I published your and Lex's llPushObjectLinear() proposal on the wiki finally
[11:01]	Arawn Spitteler was just mentioning, there's an SL Success Story, at the Gazebo outside Mystic Academy, in Spirit City: Hi Andrew. Ah, Seifert, have you been past the GCofSL, lately?
[11:01]	Seifert Surface:	gc?
[11:01]	Andrew Linden:	there were a few bugs people entered about llPushObject() not working right, so the discussion of alternatives came up a few times.
[11:01]	Seifert Surface:	cool andrew
[11:02]	Andrew Linden:	SL Success Story?
[11:02]	Andrew Linden:	Arawn, perhaps you could summarize for me.
[11:02]	Andrew Linden:	I probably won't be able to find the time to explore it.
[11:03]	Arawn Spitteler:	It was Baji Bourjade's Piece, on some network morning show. How her socializing in SL helped her come off Medications, and go shopping in Supermarkets.
[11:03]	Wolfhaven Teleportation HUD v3.0 is Ready
[11:03]	Andrew Linden:	Oh right, I think I saw that show.
[11:04]	Rex Cronon:	just crashed:(
[11:04]	Arawn Spitteler:	Gives me an idea for a HUD; I've just been trying my first llLoadURL
[11:04]	Heather Goodliffe:	sry, perhaps we can focus ;)
[11:04]	Andrew Linden:	It is impressive how the vicarious experience of moving your avatar around can, in some cases, translate to "true experience".
[11:04]	Heather Goodliffe:	what's the status of the physics updates?
[11:04]	Andrew Linden:	Heather, I was really just waiting a bit for a few more people to show up.
[11:04]	Mathieu Basiat:	yes, i was onvacation for a month and everything is broken again, lol
[11:05]	Andrew Linden:	I'm typically a few minutes late anyway, and go over a few minutes too.
[11:05]	Andrew Linden:	Alight... announcements...
[11:05]	Andrew Linden:	First of all, I may have announced earlier (I can't recall) that the "slowly descend while hovering" bug was fixed. But turns out it is not.
[11:06]	Andrew Linden:	However, the "slowly ascending while hovering near large dynamic objects" bug appears to still be fixed.
[11:06]	Andrew Linden:	Sigh. You can see how fun my job is.
[11:06]	Mathieu Basiat:	lol
[11:07]	Andrew Linden:	We currently have some fixes that are in QA (Quality Assurance) being tested.
[11:07]	Andrew Linden:	Whatever we've got ready needs to be "done" by the end of today in order to get into next week's update (if there even is an update next week).
[11:08]	Andrew Linden:	Wed morning is the new deadline for "fixed and tested candidates for the next update"
[11:08]	Andrew Linden:	I think they want to have pending simulator updates on a beta grid over the weekend before they even try to deploy to the main grid
[11:08]	Rex Cronon:	i have to cut my time here short. bye everybody
[11:08]	Andrew Linden:	Later REx
[11:09]	Andrew Linden:	So, let me check the list of fixes that are in testing now...
[11:09]	Andrew Linden:	They are:
[11:09]	Kitto Flora:	It would be good to know WHEN updates hit beta grid, and WHAT the updates are.
[11:09]	Andrew Linden:	SVC -2216 Unsit Behavior Changed
[11:10]	Andrew Linden:	SVC-1648 push applied to avatar by another avatar's scripted attachment does nothing in Havok4
[11:11]	Andrew Linden:	VWR-6267 can not log into "last location" when last location was > 768
[11:11]	Andrew Linden:	and a crash mode that is not in public jira
[11:11]	Andrew Linden:	a few other things that are not in public jira
[11:12]	Andrew Linden:	Simon, did you get any performance/optimizations into this version?
[11:12]	Mathieu Basiat:	ok
[11:12]	Andrew Linden:	Perhaps you could list any interesting ones.
[11:12]	Simon Linden:	Are you talking about the build we just did today, or the earlier one?
[11:12]	Simon Linden:	(and if I seem more confused than normal, it's due to multitasking :)
[11:14]	Simon Linden:	The one interesting thing we found was some regions running with the "disable collisions" admin setting always turned on. This ends up being a performance hit, as it adds some extra object scanning on every frame
[11:14]	Arawn Spitteler doesn't multitask well, but focuses on Chaos, Confusion, Depair, a Fanatical Devotion to the Pope, and these nifty red uniforms.
[11:15]	Gaius Goodliffe slaps Arawn with a herring.
[11:15]	Mathieu Basiat:	speaking of collisions, LAND_COLLISION_START not working for child prims
[11:15]	Simon Linden:	So far no solutions to the 'periodic lag' some people have been reporting
[11:15]	Andrew Linden:	ok I'm back... was AFK for a second.
[11:15]	Mathieu Basiat:	Andrew i think i sent you a test object
[11:16]	Andrew Linden:	Mathieu, just now or a while ago?
[11:16]	Mathieu Basiat:	awhile ago, last week
[11:16]	Heather Goodliffe:	I think I've noticed that one too
[11:16]	Mathieu Basiat:	http://jira.secondlife.com/browse/SVC-2248
[11:16]	tx Oh:	HI
[11:16]	Andrew Linden:	That is probably true Mathieu. I get a lot of test object submissions
[11:16]	Heather Goodliffe:	seems to work intermitently
[11:16]	Heather Goodliffe:	more not working than working though
[11:16]	Andrew Linden:	and my memory is vauge on those...
[11:17]	Andrew Linden:	oh right... the land_collision_start problem
[11:17]	Andrew Linden:	wasn't some progress made on that Simon?
[11:17]	Heather Goodliffe:	I don't think I ever found land collisions to be 100% reliable
[11:17]	Heather Goodliffe:	even before H4
[11:17]	Andrew Linden:	I think Kelly Linden was working on that.
[11:17]	Gaius Goodliffe:	The bug says land_collision_start, but the descript uses a script with land_collision (which works fine from child prims for me)
[11:17]	Mathieu Basiat gave you land_collision_start in child object broken.
[11:17]	Simon Linden:	I did a bunch of work on the regular collision events, but not land collisions. The others were definitely wacked - cycled between start and end a lot
[11:18]	Mathieu Basiat:	i might have mistyped
[11:18]	Andrew Linden:	Oh right, now I remember. yeah Simon was working on that. It may be working better Mathieu
[11:18]	Gaius Goodliffe nods.
[11:18]	Andrew Linden:	but I haven't yet been able to test your object
[11:18]	Mathieu Basiat:	i will edit the jira, ooops
[11:18]	Andrew Linden:	I'm currently working on huggers, and another bug where avatars can fall through stuff (if they are tall avatars)
[11:19]	Andrew Linden:	I hope to fix these two bugs before the end of today and in time for testing.
[11:19]	Mathieu Basiat:	great
[11:19]	tx Oh:	i once raised the thing with the catapult where the avatar didn't get released from. is this in the pipeline?
[11:19]	Andrew Linden:	ah yes tx.
[11:20]	Gaius Goodliffe:	Andrew, if you can fix the huggers, we'll all give you a... pat on the back. :)
[11:20]	Andrew Linden:	I think I know what is happening there. I'm not sure how best to fix it.
[11:20]	Cinthya Loveless:	yes the huggers are all jumpy
[11:20]	Simon Linden:	I'm working today a bit outside of Havok, but have found that our object parsing code can be really slow. it may make rezzing new objects a performance hit
[11:20]	Mathieu Basiat:	that i have noticed, lol
[11:20]	Andrew Linden:	The problem with the catapult, I think, is that there is (new in havok4) code that tries to help avatars stand on moving platforms
[11:20]	Andrew Linden:	even when the moving platforms are going down
[11:21]	tx Oh:	i just ask because my client (one of your majors) is unhappy and like to switch their sim back to H1, is that possible?
[11:21]	Andrew Linden:	so the avatar is sticking to the dynamic catapult
[11:21]	Heather Goodliffe:	Simon, that would be awesome to have any sort of performance improvmenets in that area :)
[11:21]	Andrew Linden:	there may be a workaround... setting the catapult phantom momentarily
[11:21]	Heather Goodliffe:	* improvements
[11:21]	Andrew Linden:	if that works, then it would add credence to my theory
[11:22]	Andrew Linden:	however... much like Mathieu's object I haven't been able to get to it yet :-(
[11:22]	tx Oh:	i can try it but when i do that the release time vary.. but i will try. can i post you a notecard with the result?
[11:22]	Andrew Linden:	What I would like to work on very soon are a number of motorcycle/vehicles that I've been getting reports about.
[11:23]	Gaius Goodliffe:	Could this stick also cause sticking to walls? I seem to find myself getting stuck on railing for spiral stairs and stuff a lot. I thought it was the LOD going down, but I understand that doesn't happen for large objects, so it appears I'm just getting stuck inside, or at least stuck to prims a lot more than under H1.
[11:23]	Mathieu Basiat:	that i would love
[11:23]	Andrew Linden:	Sure tx, I'd like to know how that works out
[11:23]	Heather Goodliffe:	Andrew, did you guys decide one way or another on what to do about objects that no longer float at water height?
[11:23]	Andrew Linden:	oh yeah, another work around idea tx...
[11:23]	Mathieu Basiat:	Gaius, I have noticed that also
[11:24]	Andrew Linden:	the "avatar is standing on a moving object" code relies on collisions near the avatar's feet (below the "knees" really)
[11:24]	Gaius Goodliffe:	Hmm. That's probably unrelated then.
[11:24]	tx Oh:	aha, this brings me to another idea..
[11:24]	tx Oh:	ok. noted that
[11:25]	Andrew Linden:	so if your catapult were to have a flat surfaceto stand on (instead of a cup) then once the surface became vertical then the avatar should stop colliding near the feet and would collide higher up (waist or higher) and the stickyness should stop
[11:25]	Gaius Goodliffe:	In any case, the inner walls of towers (hollowed cylinders) and spiral stairs (hollowed twisted tubes) seem very easy to get stuck in.
[11:25]	Andrew Linden:	antother thing to test perhaps
[11:25]	tx Oh:	ok, i will try and report back
[11:25]	Andrew Linden:	Gaius, could you elaborate on that "easy to get stuck in" statement?
[11:25]	Andrew Linden:	stuck just when walking?
[11:26]	Andrew Linden:	when teleporting in?
[11:26]	Mathieu Basiat:	basically you walk into something and are stuck on it
[11:26]	Gaius Goodliffe:	I frequently find myself immobalized after just walking up to the wall, or trying to walk up the stairs.
[11:26]	Mathieu Basiat:	SLuperglue
[11:26]	Gaius Goodliffe nods.
[11:26]	Andrew Linden:	oh, I think I know what you're talking about. Your avatar goes into a flailing (falling) animation?
[11:26]	Mathieu Basiat:	no
[11:26]	Simon Linden:	Heather - we definitely want objects to float correctly, but unfortunately there's a lot of different ways to make stuff float. Are your objects consistently high or low, moving around, or what?
[11:27]	Andrew Linden:	Mathieu, then the avatar is stuck partially inside the object?
[11:27]	Gaius Goodliffe:	Nope. Doesn't usually do anything to the animation, it just stops responding to keyboard movement. If you fight with it a bit, you can usually get unstuck, but it's annoying.
[11:27]	Mathieu Basiat:	you just walk into it, and are attached, you can sometimes walk next to it, but you are stuck to it
[11:27]	Heather Goodliffe:	consistently low
[11:28]	Heather Goodliffe:	I think the majority of object are consistently low
[11:28]	Andrew Linden:	hrm...
[11:28]	Mathieu Basiat:	flying gets yyou out
[11:28]	Heather Goodliffe:	though there are some consistently high
[11:28]	Kitto Flora:	Heather: Using vehicle code, and HOVER_HEIGHT?
[11:28]	Heather Goodliffe:	yes
[11:28]	Andrew Linden:	Heather, are you talking about vehicles hovering near water?
[11:28]	Heather Goodliffe:	yes
[11:28]	Gaius Goodliffe:	Yup, hitting fly -- I should try running sometime.
[11:28]	Mathieu Basiat:	strangely i am not having the vehicle too high or low problems
[11:29]	Simon Linden:	Are these the semi-submerged surfboards you showed me a while ago?
[11:29]	Andrew Linden:	You know, I tried the surfboards and the three or four that I tried seemed to float right.
[11:29]	Heather Goodliffe:	those are the ones
[11:29]	Andrew Linden:	Nothing like the picture reported in the jira item.
[11:29]	Kitto Flora:	Heather - may be same problem as I have - Does the code set VEHICLE_HOVER_EFFICIENCY and VEHICLE_HOVER_TIMESCALE ?
[11:29]	Heather Goodliffe:	interesting, you tried them on the beta grid?
[11:29]	Simon Linden:	I remember seeing them on the sim and they were definitely low inthe water
[11:30]	Heather Goodliffe:	yes, it's definately related to the efficiency setting
[11:30]	Mathieu Basiat:	hover_timescale is definatly broke
[11:30]	Andrew Linden:	No Heather, I tried them here on the main grid.
[11:30]	Andrew Linden:	I visited two regions...
[11:30]	Andrew Linden:	Tsunami Beach and another.
[11:30]	Mathieu Basiat:	it is like timescale is just ignored
[11:31]	Heather Goodliffe:	I've since adjusted the script, but still have a significant number of people who bought stuff pre-update who have boards underwater
[11:31]	Heather Goodliffe:	not-so-happy people ;P
[11:31]	Mathieu Basiat:	yup, me too
[11:31]	Kitto Flora:	VEHICLE_HOVER_TIMESCALE does do things - but not as documented. It changes the HEIGHT, not the time-constant to acheive the height.
[11:31]	Andrew Linden:	hover_timescale is broken...
[11:31]	Cinthya Loveless:	maybe you should do an auto update for the boards Heather
[11:31]	Cinthya Loveless:	so everyone who has bought will auto maticly get a new one
[11:32]	Andrew Linden:	ah, perhaps I aquired a fixed surfboard Heather.
[11:32]	Kitto Flora:	But I suspect it was like that in H1 - so fixing it in H4 may break even more content
[11:32]	Mathieu Basiat:	auto update would be nice, but for stuff made 2 years ago?
[11:33]	Kitto Flora:	I do updates for everything I make - even 3 year old stuff.
[11:33]	Mathieu Basiat:	no, i mean auto update in the script
[11:33]	Heather Goodliffe:	auto-updating, yet another hoop to jump through
[11:33]	Mathieu Basiat:	can't retrofit that
[11:33]	Kitto Flora:	Oh - remote load script pin thing - thats hopeless.
[11:33]	Heather Goodliffe:	also make selling non-copy items pointless
[11:34]	Kitto Flora:	All my stuff is non copy. The 'update' consists of deleting the old item and giving the owner a new fixed item
[11:34]	Heather Goodliffe:	anyhow, perhaps some attention on the timescale issue then :)
[11:34]	Gaius Goodliffe:	You couldn't pay me enough to deal with the hassles of supporting no-copy products.
[11:35]	Mathieu Basiat:	yes timescale
[11:35]	Mathieu Basiat:	thats why all my stuff is copy
[11:35]	Gaius Goodliffe:	So nice to just be able to send someone a fresh copy for whatever reason.
[11:35]	Kitto Flora:	How about $3000 US/year + ?
[11:35]	Andrew Linden:	Heather, I need a copy of an old bugged surfboard. Perhaps we could go get one right after this meeting.
[11:35]	Gaius Goodliffe:	Meh -- nope, not worth it.
[11:36]	Andrew Linden:	I'm wondering if I'll be able to fix the old surfboards.
[11:36]	Mathieu Basiat:	there are issues witth hover and setforce, no matter ho high you setforce, the hover object does not react proerly
[11:36]	Heather Goodliffe:	the one I just sent shoudl display the issue
[11:36]	Andrew Linden:	Hether, what if the fix for the old surfboard were to raise the newly patched ones above the water?
[11:37]	Heather Goodliffe:	that would suck ;P
[11:37]	Andrew Linden:	What would your preference be? Fix the old ones or keep the new ones working?
[11:37]	Heather Goodliffe:	keep the new ones working for sure
[11:37]	Andrew Linden:	Not that such a choice must happen -- I haven't looked at it yet -- but it is a possible outcome.
[11:37]	Heather Goodliffe gave you Classic Wood Surfboard.
[11:37]	Mathieu Basiat:	well, better to have 10000 working boards and 1000 broken ones, though
[11:37]	Heather Goodliffe gave you Classic Wood Surfboard H4.
[11:38]	Kitto Flora:	There's the dilemma - but it does not worry me so much - all my fixed new HOVER items have a user settable frig-factor to adjust the float height.
[11:38]	Heather Goodliffe:	gets pretty messy
[11:38]	Andrew Linden:	Heather, you just gave me two surfboards. Both broken in Havok4? or the second one is patched to work right?
[11:38]	Heather Goodliffe:	the H4 one shoudl work under Havok 4
[11:38]	Andrew Linden:	ok thanks
[11:38]	Heather Goodliffe:	the first one shoudl sink under havok 4
[11:39]	Andrew Linden:	got it
[11:39]	Andrew Linden:	Saijanai, you make motorcycles right? I forget...
[11:39]	Mathieu Basiat:	andrew, can i give you 2 broken vehicles? hangglider and windsurfer...?
[11:40]	Andrew Linden:	sure Mathieu. I'm going to try to get back to vehicles before the end of this week.
[11:40]	Mathieu Basiat:	great
[11:40]	Andrew Linden:	Lessee... I thought I had some topics to raise but I forgot to write them down.
[11:41]	Mathieu Basiat gave you Bora Hangglider Red.
[11:41]	tx Oh:	does it still happen that switchable physics stop acting physic?
[11:41]	Andrew Linden:	I've been trying to write down a list of features that I want to work on once Havok4 is "done + mostly fixed"
[11:41]	Andrew Linden:	and I was going to submit a list here to see what you all thought about how they should get prioritized
[11:42]	Andrew Linden:	a sort of "What should Andrew work on next?" excercise
[11:42]	Mathieu Basiat gave you Bora Windsurfer Red.
[11:42]	tx Oh:	thats cool
[11:42]	Heather Goodliffe:	could have someone post them on Jira and we could vote on them ;P
[11:42]	Andrew Linden:	But I don't have the list fully compiled yet. Maybe on Thursday.
[11:42]	Andrew Linden:	Mathieu, what is wrong with the hangglider?
[11:42]	Heather Goodliffe:	I personally vote for the adjusting mass, center of mass, etc :)
[11:43]	Gaius Goodliffe:	You should put the list on the wiki, and let us add to the wishlist. :)
[11:43]	tx Oh:	that voting thing in jira is a good meassure, i think
[11:43]	Andrew Linden:	The problem with that is... I've already got more stuff to work on than time in the year.
[11:43]	Gaius Goodliffe:	That's okay, we can come up with work for you for the next century. ;)
[11:43]	Andrew Linden:	Especially when something like "reimplement joints" is on the list.
[11:44]	Seifert Surface:	heh
[11:44]	Mathieu Basiat:	a few things....the timescale for hovering back to the ground is set at 60, well, now it drops in about 5 seconds
[11:44]	Kitto Flora:	Its a pity LL hired a bunch of QA guys and no bug hunters.
[11:44]	tx Oh:	andrew, thats good. congrat
[11:44]	Heather Goodliffe:	perhaps you should talk LL into letting you guys hire a couple of interns ;P
[11:44]	tx Oh:	joints, that would be nice
[11:45]	Mathieu Basiat:	also, when you fly down, the force applied is too strong, and fly up is too weak....
[11:45]	Mathieu Basiat:	using setforce for both of those
[11:45]	Andrew Linden:	interns are a tough proposition. So much training involved
[11:45]	Mathieu Basiat:	i volunteer!
[11:45]	Heather Goodliffe:	heh
[11:45]	Andrew Linden:	interns are probably a net win, but it is hard to get people to commit to mentoring them
[11:45]	Andrew Linden:	we were on the Google Summer of Code project list
[11:46]	Andrew Linden:	and we had some submissions
[11:46]	Andrew Linden:	but I don't know what came of that
[11:46]	Andrew Linden:	it was all viewer-side, since that is the only open-source project we have
[11:46]	Simon Linden:	We've taken on some other interns (outside the Google SoC project) as well. And we're recruiting programmers as well
[11:46]	Andrew Linden:	and I've been to "heads down" to pay much attention to the GSoC efforts
[11:47]	tx Oh:	hehe, yes. linden lab is mentor of summer of code, right?
[11:47]	Kitto Flora:	You guys need bug hunters
[11:47]	Seifert Surface:	thats us kitto...
[11:47]	Andrew Linden:	Kitto, explain what you mean by "bug hunter"
[11:48]	Simon Linden:	QA is tough with this kind of system ... our people tend to focus on testing where we are making changes, as "test the world" takes loooong time :)
[11:48]	Andrew Linden:	that is how is "bug hunter" distinct from "QA engineer"
[11:48]	Mathieu Basiat:	the problem with the windsurfer is that i use setforce to make the windsurfer jump out of the water, well, it doesn't jump andmore no matter what setforce setting i use
[11:48]	Kitto Flora:	Bug hunter is someone who specialises in discovering where existing code does 'the wrong thing'
[11:49]	Mathieu Basiat:	setforce and hover are not playing well together
[11:49]	Arawn Spitteler:	SetForce? I'd expect ApplyImpulse
[11:49]	Mathieu Basiat:	i mean that, lol
[11:49]	tx Oh:	they should go to the bugs
[11:49]	Heather Goodliffe:	hover sems to be having a range of issues
[11:50]	Andrew Linden:	Mathieu, it is "dangerous" to combine two actions, such as llSetForce() and llSetHover()
[11:50]	tx Oh:	i mean the QA people.
[11:50]	Heather Goodliffe:	it also has llApplyImpulse issues
[11:50]	Mathieu Basiat:	haven't had my coffee yet
[11:50]	Gaius Goodliffe:	:o
[11:50]	Gaius Goodliffe:	Life without coffee barely scares contemplating...
[11:50]	Mathieu Basiat:	i meant setforce to mean llApplyImpulse
[11:50]	Andrew Linden:	the two actions can "fight", however their effects are computed one at a time, and it is not defined which one gets processed first
[11:50]	Arawn Spitteler wants some coffee, for Hovering QA People to hunt bugs in...
[11:50]	Mathieu Basiat:	well, it worked before, lol
[11:51]	Mathieu Basiat:	consistently
[11:52]	Kitto Flora:	I get the impression that QA at LL does not 'find the cause of bugs' - they simply concentrate on if the product works in a way that is a risk. Its a yes/no thing, not a 'heres where its wrong' thing.
[11:52]	Andrew Linden:	in Havok1, from my casual examination it appeared that the actions were sorted by creation time
[11:52]	Andrew Linden:	so the first action created went into the list and the second underneath
[11:53]	Andrew Linden:	I think they were processede first-comes-first
[11:53]	Mathieu Basiat:	but the second action was negated, you are saying, in H1?
[11:53]	Andrew Linden:	which means the last action created gets the final "say" before the integration step
[11:53]	tx Oh:	i mean, isn't it possible that QA lindens pick up issues from jira, make in-world appointments with reporters to get the bug explained. sounds like a good way to get a more detailed report, more qualified for internal work processes...
[11:53]	Mathieu Basiat:	ahh
[11:53]	Andrew Linden:	how they are sorted in Havok4 I do not know yet
[11:54]	Heather Goodliffe:	tx, perhaps something they could have interns do ;P
[11:54]	Andrew Linden:	it could be they are sorted in a std::map and processed in order by pointer value -- just don't know
[11:54]	Cinthya Loveless:	Kitto why not make your own bug hunting group and go hunt bugs as best as you can indepentantly, your group.
[11:54]	Mathieu Basiat:	hummm
[11:54]	Andrew Linden:	in any case... the order is undefined, especially should we ever swap in a different physics engine again
[11:54]	Andrew Linden:	best to avoid undefined behavior when possible
[11:55]	Mathieu Basiat:	it just seems to me that if you apply impulse 100000000 that something should happen
[11:55]	tx Oh:	he, yeah. god don't dice..
[11:55]	Kitto Flora:	Cintya: I dont work for LL.
[11:55]	Cinthya Loveless:	indepentantly not through LL
[11:55]	Andrew Linden:	Well Kitto, I'm a little unclear on the distinction between "bug hunter" and "QA engineer". Some of our QA engineers are good at knowing when the feature is "wrong" and how.
[11:56]	Andrew Linden:	but they also do other stuff
[11:56]	Kitto Flora:	Cinthya: The Sim code is not open. Thats where a large number of the bugs are.
[11:56]	Arawn Spitteler thinks the physics engine shouls be a modular product, with different pricing as appropo for basic sim activity: Isn't the simulation event driven?
[11:56]	Andrew Linden:	and we get more bug reports faster from the residents than the developers can typically handle
[11:56]	Andrew Linden:	the real help would be to get better bug-reports-with-repro-recipies
[11:57]	Mathieu Basiat:	i try, lol
[11:57]	Kitto Flora:	Andrew: You are overloaded wit hexamples...
[11:57]	Andrew Linden:	Arawn, a modular physics engine is a good idea.
[11:57]	tx Oh:	anyway. it's also a bit offtopic.. but yes, the quality of the main grid in whole is in question. /me sighs
[11:57]	Kitto Flora:	Andrew: Problem is you are mixing bug hunting with bug fixing.
[11:57]	Andrew Linden:	The simulation is probably not very "event driven"... I'm unclear on what that means exactly.
[11:58]	Andrew Linden:	The simualtion is processed every 1/45th of a second (or as fast as possible in cases of lag)
[11:58]	Andrew Linden:	when events happen in the physics engine (such as collisions) they may or may not trigger certain callback functions to execute
[11:58]	Andrew Linden:	so some code is very event driven
[11:58]	tx Oh:	i love callbacks..
[11:59]	Andrew Linden:	Havok4 has lots of little hooks where we can insert special callbacks to happen
[11:59]	tx Oh:	make interfaces to them :-)
[12:00]	Andrew Linden:	Some have interfaces... sorta
[12:00]	Andrew Linden:	Havok's collisions trigger our callbacks which can trigger LSL event callbacks
[12:00]	Andrew Linden:	similarly the volume_detect feature is a callback chain
[12:00]	Simon Linden:	Some can be performance bottlenecks ... I've seen 30k+ collision events per frame, so we have to filter them a lot
[12:01]	tx Oh:	:-) way more hooks to go... a region with changed gravity would be nice, for instance
[12:01]	Heather Goodliffe:	space sims ;P
[12:01]	Arawn Spitteler:	I've started to wonder, if walking might burden the physics engine, more than Flying, by a large margin.
[12:01]	Andrew Linden:	ah tx, that would not be a hook so much as a settable parameter.
[12:02]	Andrew Linden:	Arawn, it is hard to say. There is evidence that in certain cases flying loads the simualtor more. There is a bug related to flying (hovering) avatars slowing down nearyb objects.
[12:03]	Kitto Flora:	I saw a post somewhere about collisions: It claimed that if a physical hit a prim that was overlapping many other prims, then all the overlapping prims would also check for collisions - causing a lot of CPU load. Is that (still) true?
[12:03]	Andrew Linden:	(I need to test that to make sure it still happens... there is a small chance that some other changes I made fixed it)
[12:03]	Andrew Linden:	yes Kitto that is true
[12:03]	Andrew Linden:	that is the Havok4 "simulation island" that is waking up all potentially touching objects.
[12:04]	Simon Linden:	If they have settled down, Havok knows it doesn't need to do more simulation
[12:04]	Simon Linden:	Then something comes along and hits it, and it becomes active
[12:04]	Gaius Goodliffe:	I was on the beta grid this weekend and was able to make a blimp on autopilot stop flying while just hovering nearby.
[12:04]	Kitto Flora:	That is a pity. In many/most builds there are hundreds of overlapping prims.
[12:04]	Andrew Linden:	Gaius... I can't remember the state of the beta grid.
[12:05]	Andrew Linden:	I suspect that my hovering avatar changes are already there, so that bodes ill for my hopes that I fixed that bug.
[12:05]	Gaius Goodliffe nods.
[12:06]	Andrew Linden:	Kitto, overlapping static objects don't collapse the simulation islands, I think.
[12:07]	Andrew Linden:	I was noticing in the Havok4 "visual debugger" that static objects (not dynamic) only contribute to the simulation island when touching (or nearly touching) dynamic objects
[12:07]	Andrew Linden:	I'll have to double-check
[12:08]	Andrew Linden:	Well, I've got lots of bugs to fix so I think I'll close the hour.
[12:08]	Seifert Surface:	good luck
[12:08]	Mathieu Basiat:	thx andrew
[12:08]	Gaius Goodliffe:	Good hunting :)
[12:08]	Andrew Linden:	Thanks for showing up everyone.
[12:08]	Kitto Flora:	TY Andrew... See you next time, I hope
[12:08]	tx Oh:	ty
[12:08]	tx Oh:	and bye
[12:08]	tx Oh:	tata
[12:08]	Andrew Linden:	Kitto doesn't want me to mix hunting with fixing, I think.
[12:09]	Gaius Goodliffe:	lol
[12:09]	Kitto Flora:	I want you to get more Help Andrew :)
[12:09]	Gaius Goodliffe:	You can do whatever you like as long as things get fixed. :)
[12:09]	Andrew Linden:	Yeah I know ;-)
[12:10]	Kitto Flora:	It's apparent that LL is very under-powered in some areas
[12:10]	Andrew Linden:	Yes that is true.
[12:10]	Simon Linden:	Thanks everyone for coming
[12:10]	Saijanai Kuhn:	which is why office hours and AWG exist
[12:10]	Saijanai Kuhn:	not to mention the open source community
[12:10]	Andrew Linden:	There are a number of discussions going on internally as to why we're so understaffed, and how to fix it.
[12:11]	Gaius Goodliffe:	To keep coders away from the code another hour a day? :)
[12:11]	Andrew Linden:	Some people think we're just too picky about who we hire.
[12:11]	Saijanai Kuhn drinks to that LOL
[12:11]	Andrew Linden:	Others think there are some fixable problems with our process.
[12:11]	Saijanai Kuhn:	though honestly, you get enough applciations I'm sure
[12:12]	Gaius Goodliffe:	You definitely don't want to hire just anyone -- some people introduce more problems than they fix. :p
[12:12]	Cinthya Loveless:	I wish they were as picky on the concierge team