User:Andrew Linden/Office Hours/2007 10 18

Transcript of Andrew Linden's office hours:

[07:56]	Andrew Linden:	Hello Ryozu
[07:56]	Andrew Linden:	You know... dynamic chains work best when using linked boxes instead of tori.
[07:58]	Ryozu Yamamoto:	Heya =D
[07:58]	Ryozu Yamamoto:	Yeah, I was just goofing around with something quick
[07:59]	Andrew Linden:	I was telling one of the Havok developers about what wonderful residents we have...
[07:59]	Ryozu Yamamoto:	=D
[07:59]	Andrew Linden:	how they diligently test the Havok4 preview and update/close bugs in such a timely manner.
[08:00]	Ryozu Yamamoto:	I actually feel kind of bad not having had time to do more lately
[08:00]	Andrew Linden:	Btw, I've been thinking about ways to nerf the "push in no push" objects.
[08:00]	Ryozu Yamamoto:	There is one thing of interest
[08:00]	Ryozu Yamamoto:	I managed to crash "Bug Island" a few times this morning
[08:01]	Andrew Linden:	Bug Island in Second Life? Or in the Havok4 Preview?
[08:01]	Ryozu Yamamoto:	Havok4
[08:01]	Andrew Linden:	Ok. We have a known crash mode that one of our Havok engineers is working on.
[08:01]	Ryozu Yamamoto:	Some time just before 6 am, and again at exactly 6:17 am
[08:02]	Ryozu Yamamoto:	/me nods.
[08:02]	Andrew Linden:	It looked like he might have been close to a fix yesterday when I had to leave.
[08:02]	Andrew Linden:	But, he hasn't sent me an update about it yet.
[08:02]	Ryozu Yamamoto:	Hopefully that'll be it then
[08:02]	Andrew Linden:	I'll take a quick look at the callstack of the Bug Island crash
[08:02]	Ryozu Yamamoto:	/me nods.
[08:03]	Ryozu Yamamoto:	Has there been much discussion about temp on rez objects?
[08:04]	Andrew Linden:	No, no discussion about temp on rez
[08:04]	Andrew Linden:	we can talk about them now
[08:04]	Andrew Linden:	I know more than most LL dev's about temp on rez
[08:04]	Andrew Linden:	I wrote the latest temp on rez total count management on parcels
[08:05]	Ryozu Yamamoto:	I just wanted to voice some concern. There's an example of good use up on the beta grid, Ben's vehicle lot
[08:05]	Nock Forager:	hey thanks for notice, Ryozu.
[08:05]	Ryozu Yamamoto:	No problem ;)
[08:05]	Ryozu Yamamoto:	Unfortunatly, it's also broken right now on Beta grid
[08:05]	Ryozu Yamamoto:	Most of the vendors there use "Temp on rez" previews for the vehicles
[08:05]	Andrew Linden:	Ok, the crash in Bug Island this morning was a type #1 crash in the physics engine.
[08:06]	Ryozu Yamamoto:	So, one of the known types
[08:06]	Andrew Linden:	The Havok engineer was working on type #2 yesterday, but his suspicion is that the two are related or just two versions of the same problem.
[08:06]	Ryozu Yamamoto:	/me nods.
[08:06]	Ryozu Yamamoto:	Is there any more info you need from me on it?
[08:06]	Andrew Linden:	yes, I say #1 because it was found first
[08:07]	Andrew Linden:	No, I've got a repro for #1 already.
[08:07]	Ryozu Yamamoto:	/me nods.
[08:07]	Andrew Linden:	but thanks.
[08:07]	Ryozu Yamamoto:	I didn't really want to write a repro for this one since it affects main grid still
[08:08]	Andrew Linden:	What were you thinking about temp-on-rez?
[08:08]	Ryozu Yamamoto:	I'm not sure really what should or could be done with it
[08:08]	Andrew Linden:	What are some of its "problems"?
[08:08]	Ryozu Yamamoto:	One could say things like the "0prim rezzers" are an exploit
[08:09]	Ryozu Yamamoto:	But it's hard to combat something like that, and still allow it's other use
[08:09]	Andrew Linden:	I'm unfamiliar with "oprim rezzers"
[08:09]	Ryozu Yamamoto:	One takes their construction, usually a high prim statue or building
[08:09]	Ryozu Yamamoto:	Makes it temporary, and places it in a box
[08:09]	Ryozu Yamamoto:	The box just re-rezzes it every 30-40 seconds
[08:10]	Andrew Linden:	Ah yes. That is a problem.
[08:10]	Ryozu Yamamoto:	used by quite a few to get around parcel prim counts
[08:10]	Andrew Linden:	constantly recreating objects such as that can cause simulator lag
[08:10]	Ryozu Yamamoto:	I imagine it's not very asset server friendly either
[08:10]	Andrew Linden:	I haven't thought about the temp-on-rez problems recently
[08:11]	Andrew Linden:	as I recall... the last time I did I didn't have many bright ideas on how to solve the "oprim rezzers" problem.
[08:11]	Ryozu Yamamoto:	On the other hand
[08:11]	Ryozu Yamamoto:	Are you familiar with "Holographic" vendor scripts?
[08:11]	Andrew Linden:	hrm... I think the asset-server load is either non existent, or could easily be optimized away
[08:12]	Andrew Linden:	we currently protect our asset system with an army of squids
[08:12]	Ryozu Yamamoto:	/me chuckles
[08:12]	Andrew Linden:	in fact, each simulator node is running a squid
[08:12]	Ryozu Yamamoto:	Ah, didn't know
[08:12]	Ryozu Yamamoto:	Oh, heya Soft!
[08:12]	Andrew Linden:	the cost of constantly rebuilding a particular object could also be optimized
[08:12]	Soft Linden:	Hey hey
[08:12]	Nock Forager:	hey
[08:13]	Andrew Linden:	I've had a few ideas on how to do that... mostly to solve the rapid rez of things like bullets
[08:13]	Ryozu Yamamoto:	/me nods.
[08:13]	Andrew Linden:	Hello Soft.
[08:13]	Soft Linden:	Hi hi - okay if I sit in?
[08:13]	Andrew Linden:	No problem, welcome.
[08:13]	Ryozu Yamamoto:	There's a lot of vendors that rez high prims objects as temp on rez, just to preview the item when requested by a shopper
[08:13]	Andrew Linden:	We were just talking about temp-on-rez objects and the "oprim rezzer" problem
[08:13]	Ryozu Yamamoto:	Also known as "holovendors"
[08:14]	Ryozu Yamamoto:	The difference between that and a 0prim rezzer is the whole "On request" instead of on a timer
[08:14]	Andrew Linden:	where builds that exceed the prim limit are maintained by constantly replacing them with temp-on-rez facads
[08:14]	Andrew Linden:	oh... holovendors are kinda cool
[08:14]	Ryozu Yamamoto:	They're all kind of busted on Havok though
[08:15]	Andrew Linden:	busted... how so?
[08:15]	Ryozu Yamamoto:	Every holovendor in the Havok4 Vehicle yard reports insufficiant prims to rez
[08:15]	Andrew Linden:	that sounds like a bug
[08:16]	Andrew Linden:	I guess I'll have to investigate temp-on-rez objects and make sure their bit is set.
[08:16]	Tmyclyk Dmytryk:	Hello
[08:16]	Andrew Linden:	Welcome Tmyclyk
[08:16]	Nock Forager:	Hi Tmy
[08:16]	Tmyclyk Dmytryk:	I had a bit of a question that's been rummaging through the confines of my head
[08:16]	Tmyclyk Dmytryk:	ty :D
[08:16]	Ryozu Yamamoto:	Well, if you need a place to see how it's working, there's Busy Ben's, right next to the vehicle sandboxes
[08:17]	Andrew Linden:	what is the question Tmyclyk?
[08:17]	Tmyclyk Dmytryk:	I had read a long time ago sims are limited to 256x256 due to the current havok implementation
[08:17]	Tmyclyk Dmytryk:	is that true?
[08:18]	Andrew Linden:	it isn't so much the Havok implementation as much as the simulator implementation itself
[08:18]	Andrew Linden:	to overhaul the simulator to beable to handle larger simulations would be quite a bit of work
[08:18]	Ryozu Yamamoto:	That's something I'd love to see done some day
[08:18]	Andrew Linden:	assumptions about 256 are riddled throughout the code,
[08:18]	Ryozu Yamamoto:	barring smoother region hand offs
[08:19]	Andrew Linden:	and in the communication with the client
[08:19]	Ryozu Yamamoto:	Ouch =(
[08:19]	Andrew Linden:	and the client also makes some assumptions about 256x256
[08:19]	Ryozu Yamamoto:	Heh, well
[08:19]	Ryozu Yamamoto:	So do some vehicles actually
[08:19]	Andrew Linden:	it could be abstracted out and different simulator sizes could be done... but it would be a lot of work
[08:19]	Andrew Linden:	right, I'm sure there are instances of content that make such assumptions too
[08:19]	Ryozu Yamamoto:	I think it's an important task, maybe not now, but eventually
[08:19]	Tmyclyk Dmytryk:	Understood, the reason I asked is I from what I understand the server source code is going OS one day as well, granted it's not going to be for quite a while I'm just making sure i have things in check. The larger sim size would be awesome for a project i have in mind.
[08:20]	Andrew Linden:	yes, we would like to make the simulator open source eventually
[08:20]	Andrew Linden:	once we did then serious work could start on abstracting the region size out
[08:20]	Ryozu Yamamoto:	I'm guessing a lot of the Havok4 work was abstracting it so it wouldn't have to be tangled into the sim source
[08:20]	Soft Linden:	llGetObjectDetails had a bug allowing it to return the position of items recently moved far off sim, which could return very large/large negative numbers. That alone has tons of scripts hard-coding checks on 0-256.
[08:21]	Andrew Linden:	but that project would reveal areas in the code where fundamental cleanup is required
[08:21]	Ryozu Yamamoto:	Soft: Heh, so that IS a bug eh?
[08:21]	Tmyclyk Dmytryk:	yeah, the idea for mine is all of the content residing of the simulator in question would be developed specifically for that application. Thanks for clarifying that :)
[08:21]	Ryozu Yamamoto:	I hadn't been keeping up on the details too much, but that "bug" is actually becoming used widespread
[08:22]	Andrew Linden:	could someone explain that bug to me?
[08:22]	Sidewinder Linden:	/sry to be late guys... held up in traffic.... (rl)
[08:22]	Soft Linden:	Sure...
[08:22]	Ryozu Yamamoto:	Heya Sidewinder =D
[08:22]	Andrew Linden:	I don't quite grok how it is buggy.
[08:22]	Tmyclyk Dmytryk:	welcome sidewinder
[08:22]	Nock Forager:	Hi Sidewinder
[08:22]	Ryozu Yamamoto:	Long story short: llGetObjectDetails is supposed to return details on objects in the same sim
[08:23]	Ryozu Yamamoto:	It's also returning the position of avatars that just teleported to a different sim (So technically, not in the same sim)
[08:23]	Soft Linden:	The problem was that when you initiate a teleport, the sim knows where you were TPing to. So until an agent's task was garbage collected, you could read the agent's TP destination with llGetObjectDetails.
[08:23]	Ryozu Yamamoto:	People have been using it to make "TP out trackers
[08:23]	Ryozu Yamamoto:	Whenever someone would teleport, it would tell you where they went
[08:23]	Soft Linden:	I've changed it to fail the function if getting the position on an avatar who's not in the region, or who has traveled >96m off any sim edge.
[08:23]	Andrew Linden:	ah, I see
[08:23]	Andrew Linden:	and did someone fix that already?
[08:24]	Ryozu Yamamoto:	Fair warning Soft, you'll get an earful ont hat
[08:24]	Ryozu Yamamoto:	Me, I'll laugh
[08:24]	Sidewinder Linden:	soft - is that fix in the mainline or in havok4?
[08:24]	Ryozu Yamamoto:	But that's just me knowing when NOT to use something -_-;;
[08:24]	Soft Linden:	It will go in mainline. Should be part of the next rolling restart (not today's)
[08:24]	Andrew Linden:	It's not in Havok4
[08:24]	Sidewinder Linden:	ok... so it will be on the next merge... cool
[08:25]	Ryozu Yamamoto:	I actually had a conversation a couple hours ago with some girl who was angry the blitz sphere is getting fixed
[08:25]	Sidewinder Linden:	really...?
[08:25]	Ryozu Yamamoto:	I offered to crash her sims with it, she didn't take me up on the offer, =P
[08:25]	Andrew Linden:	Speaking of the blitz sphere getting fixed...
[08:25]	Ryozu Yamamoto:	/me chuckles
[08:25]	Andrew Linden:	I'm thinking I may also fix "push in no push". Anyone here know how that works?
[08:25]	Tmyclyk Dmytryk:	yeah
[08:26]	Tmyclyk Dmytryk:	well i know how i've made it work
[08:26]	Tmyclyk Dmytryk:	Flooding llPushObject @ MAX_INT
[08:26]	Ryozu Yamamoto:	There's only 2 push in no_push that I know about
[08:26]	Ryozu Yamamoto:	blitz and TWO_PI target omega
[08:26]	Andrew Linden:	Oh, I already capped llPushObject()
[08:26]	Andrew Linden:	no, there is another way...
[08:26]	Ryozu Yamamoto:	=/
[08:26]	Tmyclyk Dmytryk:	what's it capped at?
[08:27]	Nock Forager:	so many way... ough.
[08:27]	Andrew Linden:	you make an object that interpenetrates the avatar, and follows the avatar constantly setting its position inside the avatar
[08:27]	Andrew Linden:	the penetration resolution code will make the avatar go up
[08:27]	Ryozu Yamamoto:	Ah, there's the megaprim issue of course
[08:27]	Soft Linden:	You can also use twisted objects that reverse the surface normals. I've meant to ask if havok4 normalizes those.
[08:28]	Andrew Linden:	yes, and you can do it with megaprims... but that hammer is pretty big
[08:28]	Ryozu Yamamoto:	Yeah =/
[08:28]	Ryozu Yamamoto:	There's no good answer to that one
[08:28]	Andrew Linden:	I don't know about the twisted objects and Havok4 yet Soft.
[08:28]	Andrew Linden:	I haven't investigated yet.
[08:28]	Ryozu Yamamoto:	They seem mostly fixed
[08:28]	Ryozu Yamamoto:	The normals that is
[08:29]	Andrew Linden:	I think that is primarily how the blitz sphere worked.
[08:29]	Ryozu Yamamoto:	But I can't relaly tell
[08:29]	Ryozu Yamamoto:	Err, Yeah
[08:29]	Ryozu Yamamoto:	Oh yeah, there's a seat for yah Sidewinder =D
[08:29]	Andrew Linden:	anyway... I'm looking into how it might be possible to "bypass" certain collision/penetration events
[08:30]	Nock Forager:	Nexttime, should wrote in notice "with your favorites chair" :)
[08:30]	Andrew Linden:	such as... ignore penetrations if one object is an avatar and another object is using llSetPos() to put itself into penetration.
[08:30]	Soft Linden:	I believe the two objects I just handed you take advanage of that to bounce about wildly, if you wanted to compare. (I can JIRA it if you prefer)
[08:30]	Ryozu Yamamoto:	Seifert!
[08:30]	Tmyclyk Dmytryk:	what about code which instead of pushing the avatar up, it pushes it the quickest way out? I.E. if you're on the soutside of the prim it pushes you out that side
[08:30]	Ryozu Yamamoto:	Have I ever told you that I love you Seifert? XD
[08:30]	Seifert Surface:	hi all
[08:30]	Andrew Linden:	Welcome Seifert.
[08:31]	Seifert Surface:	ty
[08:31]	Tmyclyk Dmytryk:	would that be a little more effective?
[08:31]	Ryozu Yamamoto:	Oh Soft
[08:31]	Ryozu Yamamoto:	Have you ever come across the sphere that doesn't bounce
[08:31]	Andrew Linden:	Tmyclyk, that is tough code to write correctly.
[08:31]	Ryozu Yamamoto:	But the surface normals make it spin wildly?
[08:31]	Andrew Linden:	That is, it is difficult to find the fastest route out of penetration
[08:32]	Seifert Surface:	im happy to hear it ryozu...
[08:32]	Soft Linden:	Haven't!
[08:32]	Tmyclyk Dmytryk:	understood
[08:32]	Ryozu Yamamoto:	I can't find it in my inventory anymore!
[08:32]	Ryozu Yamamoto:	I found it for sale in the air somewhere, lol
[08:32]	Tmyclyk Dmytryk:	I'm not sure how the backend works, I've only done something in a similar manner with lsl lol
[08:32]	Ryozu Yamamoto:	Seifert, Andrew, can I make a suggestion?
[08:32]	Andrew Linden:	Yes please.
[08:33]	Ryozu Yamamoto:	Can we get a copy of "The Future" up on beta?
[08:33]	Seifert Surface:	last i hcecked volumedetect wasnt in
[08:33]	Seifert Surface:	so it wouldnt work
[08:33]	Andrew Linden:	Oh, I was thinking about putting "The Future" up on the preview anyway, but had not got around to it.
[08:33]	Ryozu Yamamoto:	I think it'd be a good sim to have up
[08:33]	Andrew Linden:	The Future uses some scripts to push the avatar around, and I suspect they are broken in Havok4
[08:33]	Seifert Surface:	i did check if a phantom prim will push you correctly if youre inside it
[08:33]	Ryozu Yamamoto:	I suspect so as well
[08:33]	Seifert Surface:	and it seems to
[08:34]	Seifert Surface:	velocities are right
[08:34]	Seifert Surface:	its just volumedetect
[08:34]	Ryozu Yamamoto:	When did you check velocities Seifert?
[08:34]	Andrew Linden:	well, that is a good sign Seifert.
[08:34]	Seifert Surface:	few days ago
[08:34]	Ryozu Yamamoto:	Ah
[08:34]	Seifert Surface:	the scripts call llgetmass
[08:34]	Ryozu Yamamoto:	Have the PushObject and ApplyImpulse caps on avatars been relaxed enough?
[08:34]	Seifert Surface:	so any fiddling with relative weights should be fine
[08:35]	Seifert Surface:	it seemed great, i didnt design any of it to push at greater than 50m/s
[08:35]	Seifert Surface:	and the cap seems higher than that
[08:35]	Ryozu Yamamoto:	Oh, and your "Vote" object in the The Future hub blocks the push to the Q transport
[08:35]	Andrew Linden:	the llApplyImpulse() for attachments has been fixed. There was a bug there.
[08:35]	Ryozu Yamamoto:	hehe
[08:35]	Ryozu Yamamoto:	It felt fixed last I checked, but I didn't do any real testing
[08:35]	Seifert Surface:	thanks, ill check out the blockage :)
[08:36]	Andrew Linden:	The "cap" on llPushObject that Tmyclyk asked about...
[08:36]	Brampton Region:	CRASH
[08:36]	Andrew Linden:	what I did was to clamp incoming arguments to some sane values
[08:38]	Ryozu Yamamoto:	What the crap
[08:38]	Seifert Surface:	crashed out
[08:38]	Ryozu Yamamoto:	Sim restarted
[08:38]	Ryozu Yamamoto:	For no apparent reason
[08:38]	Nock Forager:	quite suddenly.
[08:38]	Sidewinder Linden:	how appropos ;)
[08:39]	Tmyclyk Dmytryk:	that was random :D
[08:39]	Ryozu Yamamoto:	Rolled me back as well
[08:39]	Andrew Linden:	/me looks at the logs...
[08:39]	Ryozu Yamamoto:	So where were we?
[08:40]	Ryozu Yamamoto:	llPushObject caps
[08:40]	Ryozu Yamamoto:	and sane values
[08:40]	Tmyclyk Dmytryk:	yeah
[08:40]	Seifert Surface:	volumedetect is still nonfunctional right?
[08:40]	Nock Forager:	Hi there
[08:41]	Andrew Linden:	The last bit that happened right before the crash was a user detached an attachment.
[08:41]	Ryozu Yamamoto:	Doh
[08:41]	Andrew Linden:	I'll investigate more later.
[08:41]	Nock Forager:	detaching make sim crash...
[08:41]	Ryozu Yamamoto:	Asset server blockage maybe
[08:41]	Seifert Surface:	if only every sim could have a linden troubleshooter :)
[08:41]	Ryozu Yamamoto:	Heh
[08:42]	Nock Forager:	Live supppot enhanched. :)
[08:42]	Andrew Linden:	Oh right... llPushObject caps...
[08:42]	Ryozu Yamamoto:	I'd like to ask, what is considered a sane value
[08:42]	Andrew Linden:	there was a bug in llPushObject that allowed you to push very hard... but you couldn't do it with accuracy
[08:42]	Andrew Linden:	but when pushing very hard you don't really care
[08:43]	Andrew Linden:	and llPushObject has this odd 1/r^3 addative falloff factor... you can't use it with accuracy anyway
[08:43]	Seifert Surface:	people have
[08:43]	Andrew Linden:	so I just capped the really huge values
[08:43]	Seifert Surface:	taking into account the 1/r^3
[08:43]	Andrew Linden:	and force it to actually fall off for "large" distances
[08:44]	Ryozu Yamamoto:	/me nods.
[08:44]	Ryozu Yamamoto:	Seems reasonable
[08:44]	Andrew Linden:	lemme check the Havok4 code...
[08:44]	Seifert Surface:	i dont really understand where the dropoff starts - it doesnt seem to start when inside the prim itself
[08:45]	Ryozu Yamamoto:	It's a logarithmic distance thing I think
[08:45]	Seifert Surface:	i asked phoenix about that a year or so ago, and he didnt see it in a quick scan of the code
[08:45]	Ryozu Yamamoto:	Or something like that
[08:46]	Andrew Linden:	yeah so... llPushObject() calls from more than 17 meters away will actually start to fall off... within that distance it will behave as it used to... with an additional cap
[08:46]	Seifert Surface:	how did it used to behave? :)
[08:46]	Ryozu Yamamoto:	17m That's a good number to take note of
[08:46]	Seifert Surface:	with no fall off?
[08:47]	Andrew Linden:	I wonder why I made it 17. 16 seems like such a better number.
[08:47]	Ryozu Yamamoto:	Heh
[08:47]	Andrew Linden:	Meh... I'll probably drop it to 16. Easier to remember.
[08:47]	Seifert Surface:	hmm
[08:47]	Andrew Linden:	This is how I capped it...
[08:48]	Ryozu Yamamoto:	Wow, that's uh
[08:48]	Andrew Linden:	each object has an odd "script energy" budget
[08:48]	Ryozu Yamamoto:	little excessive
[08:48]	Andrew Linden:	calling llPushObject() taps into that budget
[08:48]	Ryozu Yamamoto:	The farther away it is, the more energy it takes to push, right?
[08:49]	Andrew Linden:	so what I do is... If you try to push harder than the object could push with max energy... I cap it to the effective max energy push...
[08:49]	Andrew Linden:	this is done before the object's actual energy reserves are looked at
[08:49]	Ryozu Yamamoto:	Max, can you perhaps stop the pumpkin?
[08:49]	Ryozu Yamamoto:	Heh
[08:49]	Seifert Surface:	is energy proportional to m (change in vel)^2?
[08:49]	Andrew Linden:	the push will then be attenuated by the fraction of energy reserves on hand
[08:50]	Andrew Linden:	the energy consumption rate is odd... I'm not even sure what it is proportional to, at least not by memory
[08:50]	MAX Brenham:	no Ryozu
[08:50]	Andrew Linden:	I'de have to study the code
[08:50]	Ryozu Yamamoto:	You keep dropping the SimFPS to 0 =/
[08:50]	Andrew Linden:	I didn't write the energy consumption curve... it was a "seat of the pants" hack done early
[08:51]	Seifert Surface:	yeah, ive never understood that stuff
[08:51]	MAX Brenham:	if the lindin asks i will :) but your not my boss
[08:51]	Andrew Linden:	unfortunately it cannot be changed now... it would break content if I did
[08:51]	Ryozu Yamamoto:	Righteo
[08:52]	MAX Brenham:	ya i have been trying to crash the sims in the beta
[08:52]	MAX Brenham:	no luck
[08:52]	Ryozu Yamamoto:	;)
[08:52]	Sidewinder Linden:	thanks for trying max...
[08:52]	Andrew Linden:	anyway, very large llPushObject()'s will not orbit people from arbitrarily far away.
[08:52]	MAX Brenham:	i tryed my best
[08:52]	MAX Brenham:	:(
[08:53]	Andrew Linden:	Well, my hope is that we'll soon make it much harder.
[08:53]	Ryozu Yamamoto:	/me nods.
[08:53]	Andrew Linden:	I had a havok engineer in yesterday trying to fix the two crash modes we know about.
[08:53]	MAX Brenham:	ya they need to get the lag out of it to
[08:53]	MAX Brenham:	the only way you can crash it is with a math crash
[08:53]	Andrew Linden:	Looked like he was onto something when I had to leave, but I haven't got an update yet.
[08:55]	Ryozu Yamamoto:	I don't suppose we hae a czar of vehicle bugs do we?
[08:55]	Seifert Surface:	are there still crashes without repro?
[08:55]	MAX Brenham:	when is Havok4 coming to the normal grid
[08:55]	Ryozu Yamamoto:	"When it's ready" I'd imagine ;)
[08:55]	Sidewinder Linden:	max... that will depend on how long it takes to clean up the things that we are now talking about... eliminating crashes and being sure that as little existing content would "break" as possible
[08:55]	Andrew Linden:	I suspect there are crash modes without repro's
[08:56]	Sidewinder Linden:	we do not have a specific date, but are making strong progress...
[08:56]	Andrew Linden:	however, all the repro's I've found fall into two categories
[08:56]	Andrew Linden:	both of which are probably faces of the same bug
[08:56]	Andrew Linden:	Ryozu, what do you mean by "czar of vehicle bugs"?
[08:57]	Sidewinder Linden:	max... please do not crash the sim... this is not havok4...
[08:57]	MAX Brenham:	sorry
[08:57]	Ryozu Yamamoto:	Someone how's keeping on top of how vehicles work, and how they should work
[08:57]	Ryozu Yamamoto:	Sort of like how I"m on everything else scripting related ;)
[08:57]	MAX Brenham:	you need like 10 of them to crash a sim that was only 3
[08:57]	Sidewinder Linden:	well, what we are doing is using the jira (issue tracker) bug listings as the work list...
[08:58]	Andrew Linden:	No, not yet that I know of Ryozu. I think we've got several vehicle bugs to fix
[08:58]	Sidewinder Linden:	hopefully the bug reports we get will be representative of many other cases...
[08:58]	Andrew Linden:	I'm just trying to fix what is found, and at the moment I've got plenty to keep me busy.
[08:58]	Ryozu Yamamoto:	I'd be doing what I could with vehicle scripting as well,but... I'm not at all familiar with vehicle code
[08:58]	Sidewinder Linden:	i would expect that once we think "vehicles seem to be pretty close to right", then i might blog asking folks to specifically "go to town" with vehicles to see what else is left
[08:59]	Sidewinder Linden:	but we're not there yet
[08:59]	Ryozu Yamamoto:	/me nods.
[08:59]	Andrew Linden:	As to delivery of Havok4... I suspect we'll make that transition lowly...
[08:59]	Andrew Linden:	We'll probably open a Havok4 sandbox in SL first
[08:59]	Andrew Linden:	or two sandboxen
[09:00]	Andrew Linden:	and then allow estate owners who want to move to Hvok4 early to do so
[09:00]	MAX Brenham:	aaawww:(
[09:00]	Seifert Surface:	might be some who want to just try it for a day
[09:00]	Seifert Surface:	see what breaks
[09:00]	Sidewinder Linden:	the good news is that we can "flip" sims from havok1 to havok4 and vice versa...
[09:01]	Seifert Surface:	and then fix parts in h4 sims without having their sim be broken until they can fix it
[09:01]	Nock Forager:	Any sideeffect?
[09:01]	Tmyclyk Dmytryk:	Pretty cool
[09:01]	Sidewinder Linden:	although i wouldn't want to set that up as a standard mode... the hetgrid project created the infrastructure to be able to change a sim from one to the other
[09:01]	Tmyclyk Dmytryk:	i'd imagine not, just loading different libraries
[09:01]	Nock Forager:	hmhm
[09:01]	Andrew Linden:	Hrm... that is a good point Seifert. We wouldn't want to flip the estate simulator version constantly.
[09:01]	Ryozu Yamamoto:	Good for a worse case fallback though
[09:02]	Andrew Linden:	Ideally we would make it easy to flip the Havok4 bit.
[09:02]	Seifert Surface:	there will be lots of gnashing of teeth if its a one way street
[09:02]	Sidewinder Linden:	however... if something gets broken, in terms of content, with havok4, there are cases where we'd need to rollback as well when "going back"
[09:02]	Andrew Linden:	I'm not sure how that is going to work out.
[09:02]	Sidewinder Linden:	realistically, i think we need to have "early adopters" who are willing to take those early risks in the first conversion set
[09:02]	Ryozu Yamamoto:	Aye, if estate owners can flip to Havok4, I'm guessing a fair number would want to flip back just in case
[09:02]	Seifert Surface:	would it make sense to have some office hours in the beta grid?
[09:03]	Andrew Linden:	No, I'm inclined to keep the office hours here in Second Life proper.
[09:03]	Tmyclyk Dmytryk:	What'd be interesting is, (and to take load off of you guys in particular) is for a while that h4 is on the main grid, instead of performing havok 4 on sims that owners request it, just add a small dropdown list that restarts the sim if they wish to 'flip' back and forth
[09:03]	Andrew Linden:	It might make sense to have special office hours in the Havok4 preview.
[09:03]	Tmyclyk Dmytryk:	it would take the load off of you and allow more time for development/bug finding
[09:03]	Seifert Surface:	thats what i meant
[09:03]	Sidewinder Linden:	tmyclyck it is not that simple
[09:03]	Seifert Surface:	i think...
[09:03]	Ryozu Yamamoto:	/me chuckles
[09:03]	Tmyclyk Dmytryk:	Ah alright
[09:04]	Andrew Linden:	Here's a question for you all... currently I hold office hours twice a week.
[09:04]	Sidewinder Linden:	running havok4 is *not* just flipping the havok4 binaries
[09:04]	Andrew Linden:	I'm thinking maybe that is too often... perhaps it should only be once a week.
[09:04]	Andrew Linden:	What do you all think?
[09:04]	Seifert Surface:	there doesnt seem to be a huge crowd, although that could be timing
[09:04]	Tmyclyk Dmytryk:	I keep missing them, I think it'd be useful for some sort of notification group, unless there is and i'm completely oblivious
[09:04]	Ryozu Yamamoto:	Personally, I like to talk about new code rollout after it happens
[09:04]	Ryozu Yamamoto:	But if that's only once a week, the only one office hour is really needed
[09:05]	Andrew Linden:	Right... that is sorta my thinking Ryozu
[09:05]	Andrew Linden:	with office hours and internal meetings I'm finding my time eaten up and am having difficulty getting into the "state" of fixing bugs.
[09:06]	Seifert Surface:	yeah, id say go to 1 per week then
[09:06]	Sidewinder Linden:	well, unless there is a strong preference otherwise, i would be tempted to push for one day per week
[09:06]	Andrew Linden:	However... I kinda like the idea of office hours in the Havok4 preview
[09:06]	Sidewinder Linden:	yes... that would allow folks to demonstrate issues...
[09:06]	Ryozu Yamamoto:	/me nods.
[09:06]	Andrew Linden:	what I worry about in preview office hours is that...
[09:06]	Soft Linden:	If you do, you might want to have it on the edge of a region so people can demonstrate sim-crashing/lagging issues over the edge. :)
[09:06]	Ryozu Yamamoto:	How often do you expect to get new code live on the beta grid?
[09:07]	Ryozu Yamamoto:	Good idea Soft =D
[09:07]	MAX Brenham:	how would you contact linden labs? .. phone .. email .. ?
[09:07]	Andrew Linden:	(1) It'll drive me nuts to see all these problems and not be able to work on them while in the office hour
[09:07]	Ryozu Yamamoto:	/me chuckles
[09:07]	Sidewinder Linden:	heh ;)
[09:07]	Seifert Surface:	:D
[09:07]	Ryozu Yamamoto:	Tmy crash?
[09:07]	Andrew Linden:	(2) Residents will probably be able to show me an issue and say "I showed you this last week, why haven't you fixed it yet?"
[09:07]	Ryozu Yamamoto:	Yep
[09:07]	Seifert Surface:	but would we?
[09:07]	Nock Forager:	wb Tmy
[09:08]	Ryozu Yamamoto:	I'd expect a lot of that =/
[09:08]	MAX Brenham:	cool
[09:08]	Tmyclyk Dmytryk:	Interesting
[09:08]	Seifert Surface:	the residents are always complaining that lindens dont actually use the product they work on...
[09:08]	Tmyclyk Dmytryk:	Thanks
[09:08]	Tmyclyk Dmytryk:	wasn't so much of a crash, as an application hang
[09:08]	Andrew Linden:	That is mostly true in my case seifert, unfortunately.
[09:08]	Andrew Linden:	I work to hard to be able to play on SL.
[09:09]	Seifert Surface:	seems like this would be an opportunity, if painful...
[09:09]	Ryozu Yamamoto:	Aww
[09:09]	Andrew Linden:	Ok well let's try this...
[09:09]	Sidewinder Linden:	perhaps after a new code drop, run the office hours in the beta preview - to show new behaviors ?
[09:09]	Andrew Linden:	I'll hold next Tuesday office hours here...
[09:09]	Sidewinder Linden:	and otherwise on second life proper?
[09:09]	Andrew Linden:	and we'll hold next Thursday's office hours in the Havok4 preview.
[09:09]	Seifert Surface:	cool
[09:10]	Ryozu Yamamoto:	Sounds good
[09:10]	Tmyclyk Dmytryk:	sounds appropriate :)
[09:10]	Andrew Linden:	Then... when Havok4 is done maybe we'll retire the Thursday hours.
[09:10]	Ryozu Yamamoto:	/me nods.
[09:10]	Seifert Surface:	burn that bridge when we get there...
[09:10]	Ryozu Yamamoto:	Who's interested in notifications of meetings
[09:10]	Seifert Surface:	/me
[09:10]	Ryozu Yamamoto:	And how would you like to recieve them?
[09:10]	Tmyclyk Dmytryk:	me :)
[09:10]	Ryozu Yamamoto:	Actually,
[09:11]	Tmyclyk Dmytryk:	I seem to be horrible at timekeeping
[09:11]	Ryozu Yamamoto:	Anyone interested in maybe forming a "Havok4" group?
[09:11]	Sidewinder Linden:	sure we could do that...
[09:11]	Tmyclyk Dmytryk:	that or an office hours group
[09:11]	Ryozu Yamamoto:	I'm prone to say just a Havok4 group
[09:11]	Tmyclyk Dmytryk:	yeah
[09:12]	Seifert Surface:	groups... another scarce resource
[09:12]	Ryozu Yamamoto:	I get tired of hearing about the rest sometimes =|
[09:12]	Sidewinder Linden:	i am going to have to run at the moment, but will get that set up this afternoon if you want
[09:12]	Seifert Surface:	but yeah, ive missed a couple through forgetting the time
[09:12]	Ryozu Yamamoto:	No offense to the other LIndens
[09:12]	Seifert Surface:	thatd be great sw
[09:12]	Soft Linden:	Or SL Physics. No use in codifying havok4 specifically.
[09:12]	Ryozu Yamamoto:	True, very true
[09:12]	Sidewinder Linden:	ok i'll do that this afternoon - good point soft... SL Physics it is... check for it later today
[09:12]	Ryozu Yamamoto:	Anyone interested in physics in general would like that
[09:13]	Ryozu Yamamoto:	Alright =D
[09:13]	Andrew Linden:	let's make it a Havok4 group with the intention of disbanding the group when it no longer makes sense
[09:13]	Andrew Linden:	then we can migrate to a physics engine group or something.
[09:13]	Nock Forager:	Some students will join too :)
[09:13]	Ryozu Yamamoto:	Up to you guys
[09:13]	Sidewinder Linden:	ok... tell you what... i'll check with andrew and soft... i have to run... sry bye - another commitment...
[09:13]	Ryozu Yamamoto:	Alright
[09:13]	Seifert Surface:	groups with finite lifespans are good...
[09:13]	Ryozu Yamamoto:	We'll find out next office hour meeting then ;)
[09:13]	Andrew Linden:	Yeah, I need to get to work, so I'm going to run too.
[09:14]	Seifert Surface:	good luck
[09:14]	MAX Brenham:	bye
[09:14]	Ryozu Yamamoto:	Kept you guys here long enough
[09:14]	MAX Brenham:	:)
[09:14]	Andrew Linden:	We'll talk more about the group next week.
[09:14]	MAX Brenham:	have a great day
[09:14]	Ryozu Yamamoto:	Thanks for the hard work guys
[09:14]	Andrew Linden:	er... if we decide anything or make progress on that I'll talk about it next week.
[09:14]	Andrew Linden:	Thanks for the chairs Ryozu
[09:14]	Tmyclyk Dmytryk:	You guys do an excellent job on the platform, I enjoy it!
[09:14]	Ryozu Yamamoto:	I'll just manually annoy you all to show up Tuesday ;)
[09:14]	Andrew Linden:	I need to get one of those dynamic conference tables.
[09:15]	Ryozu Yamamoto:	I'll script one up for yah
[09:15]	Ryozu Yamamoto:	I've been meaning to anyway
[09:15]	Andrew Linden:	Keep an eye out for one for me. I'll go buy it if you show me where the vendor is.
[09:15]	Ryozu Yamamoto:	/me chuckles
[09:15]	Andrew Linden:	That would be great Ryozu.
[09:15]	Ryozu Yamamoto:	Alright, seeya later =D
[09:15]	Andrew Linden:	see you all later
[09:15]	Nock Forager:	ciao Andrew.