User:Andrew Linden/Office Hours/2007 11 01

Transcript of Andrew Linden's office hours:

[8:00]	Andrew Linden:	Good morning Ryozu.
[8:01]	Ryozu Yamamoto:	Good morning =)
[8:01]	Andrew Linden:	I changed the autoreturn of this parcel to be 60 minutes, so the table won't be returned too soon
[8:01]	Ryozu Yamamoto chuckles
[8:02]	Andrew Linden:	I'll have to remember to set it back to 10 min at the end of the hour
[8:02]	Ryozu Yamamoto:	I didn't even look, and here I was working on an anti-auto-return script just now
[8:02]	Andrew Linden:	I should move the office hours to a new place
[8:02]	Ryozu Yamamoto:	Good mornin Seifert!
[8:02]	Andrew Linden:	I will eventually.
[8:02]	Seifert Surface:	hi hi
[8:03]	Andrew Linden:	I'll find some Linden Land and take it for my own
[8:03]	Andrew Linden:	and we'll hold it there.
[8:03]	Ryozu Yamamoto grins.
[8:03]	Andrew Linden:	Hello Seifert
[8:03]	Andrew Linden:	I was working on your llPushObject() on the avatar bug yesterday
[8:04]	Andrew Linden:	I know what is going wrong, but had to leave before I could actually fix it
[8:04]	Seifert Surface:	is it the falloff?
[8:04]	Andrew Linden:	No, the problem is in the "script energy" consumption rate.
[8:04]	Ryozu Yamamoto:	That would have been my guess.
[8:04]	Seifert Surface:	even on the first hit?
[8:04]	Ryozu Yamamoto:	When I was testing various things on an avatar, a single call with hardly any force ate the energy up
[8:05]	Andrew Linden:	llPushObject() taps into the mysterious "script energy" feature and is attenuated when not enough energy exists.
[8:05]	Seifert Surface:	ah, so at sufficient distance it was getting nerfed
[8:05]	Seifert Surface:	because the needed energy was higher than it could supply?
[8:05]	Andrew Linden:	Yeah, I've got the energy consumption off, by a factor of 100, I think.
[8:05]	Seifert Surface:	ok
[8:06]	Andrew Linden:	The reason this bug cropped up is that I've changed the internal density/mass values that the physics engine uses
[8:06]	Seifert Surface:	ah, right
[8:06]	Andrew Linden:	mostly to try to achieve more stability for motion of small objects
[8:06]	Andrew Linden:	I had to keep the llGetMass() unchanged
[8:07]	Seifert Surface:	and, well, does script energy have the effect it was originally intended to have?
[8:07]	Andrew Linden:	and the script energy uses the legacy mass units so I have to do some magic to get the new energy consmption rates the same
[8:07]	Andrew Linden:	and my magic is broken in at least one spot
[8:07]	Seifert Surface:	it seems like something that one could potentially fiddle with without breaking content, if it were made more lenient
[8:08]	Seifert Surface:	or are we hoping that there wont be ways to get around stupidly high pushes etc even with the energy constraints now?
[8:08]	Andrew Linden:	I think I've fixed the arbitrarily large push problem
[8:08]	Andrew Linden:	as I mentioned in that SLViews talk way back in Sept.
[8:09]	Seifert Surface:	right
[8:09]	Seifert Surface:	and energy is tied to the prim rather than the script right?
[8:09]	Ryozu Yamamoto:	So things like Hand Of God should be nerfed more or less
[8:09]	Andrew Linden:	Also... there is a distance fall off on llPushObject() now... but it doesn't kick in until about 17 meters
[8:09]	Andrew Linden:	17 meters ~= 10 * sqrt(3) / 2
[8:10]	Seifert Surface:	k
[8:10]	Andrew Linden:	er no... 10*sqrt(3)
[8:10]	Seifert Surface:	no/2?
[8:10]	Seifert Surface:	so twice the size of a max normal prim now
[8:11]	Andrew Linden:	Yeah, twice the size of the diameter of the bounding box of the max prim.
[8:11]	Andrew Linden:	er... bounding sphere
[8:11]	Seifert Surface:	ok that makes sense
[8:11]	Seifert Surface:	and so outside of that range, its doign a 1/r^3?
[8:12]	Andrew Linden:	Well, the falloff is still up in the air... I recently changed it to fall off 1/ ( (r-R)/Q)
[8:12]	Andrew Linden:	where r = radius, R = 17, Q = 5
[8:12]	Andrew Linden:	er no, that formula isn't quite right...
[8:12]	Seifert Surface:	that may break some stuff, if people were accounting for the falloff already
[8:13]	Seifert Surface:	although the only thing ive seen for that are these star wars telekinesis type things
[8:13]	Andrew Linden:	There used to be NO falloff. However there was (and still is) a HUGE script energy consumption the farther away you try to push
[8:13]	Seifert Surface:	which might be just updating on the fly
[8:13]	Seifert Surface:	ah interesting
[8:13]	Ryozu Yamamoto:	Haven't even looked at the script in this thing
[8:14]	Seifert Surface:	wow, classy prims
[8:14]	Andrew Linden:	yeah, so now I have a falloff where the script energy consumption should be absolutely tremendous anyway
[8:14]	Ryozu Yamamoto:	Heh
[8:14]	Ryozu Yamamoto gave you GGun.
[8:14]	Ryozu Yamamoto:	Only works in a push enabled area, but let's you "Grab" physics objects
[8:14]	Seifert Surface:	llPow(llVecDist(llGetPos(), pos), 3),
[8:14]	Andrew Linden:	What does the GGun do Ryozu?
[8:14]	Andrew Linden:	Ah
[8:15]	Seifert Surface:	likely it works reasonably well no matter what the falloff is
[8:15]	Andrew Linden:	Well, here is the update on Havok4...
[8:16]	Andrew Linden:	I think there is a database update happening right now... or else it has completed
[8:16]	Andrew Linden:	however I suspect the preview is down at the moment
[8:16]	Andrew Linden:	I haven't had time to check it this morning.
[8:16]	Seifert Surface:	so the blog post said
[8:16]	Ryozu Yamamoto nods.
[8:16]	Andrew Linden:	We had actually updated the simulator with a few fixes yesteday.
[8:17]	Andrew Linden:	Did either of you get to play with stuff on that new binary?
[8:17]	Ryozu Yamamoto:	Not yet =(
[8:17]	Seifert Surface:	nope
[8:17]	Ryozu Yamamoto:	I've been pretty busy lately
[8:17]	Andrew Linden:	Ok, no problem.
[8:17]	Andrew Linden:	I asked Havok what was up with the crash bugs they were trying to fix.
[8:18]	Andrew Linden:	They have not been able to reproduce them yet... using their own demo/test code.
[8:18]	Andrew Linden:	So, I'm not really sure what to do about that yet
[8:18]	Ryozu Yamamoto:	That's pretty frustrating
[8:18]	Andrew Linden:	I see three possibilities (I think it was three). Lemme see if I can remember them now...
[8:19]	Andrew Linden:	(1) We cross our fingers and hope Havok can figure out how to repro, then they fix it.
[8:19]	Seifert Surface:	might work... :)
[8:20]	Andrew Linden:	(2) I figure out how to help them repro it in their offices, perhaps using what we call a "grid in a box"
[8:20]	Andrew Linden:	where one computer runs all of the SL services (including asset server and db)
[8:20]	Ryozu Yamamoto grins, "Grid in a box, wouldn't that be nice to have, heh"
[8:20]	Andrew Linden:	unfortunately I think the GridInABox technology has not been maintained
[8:21]	Ryozu Yamamoto:	Doh
[8:21]	Andrew Linden:	and it would probably be a lot of work to get it functioning again
[8:21]	Andrew Linden:	(3) We modify the RCCS to NOT change shape LevelOfDetail (LOD) and save the physcs engine performance some other way
[8:22]	Sidewinder Linden:	hi guys - sorry - "a bit of a computer problem" but at least it booted once :p
[8:22]	Ryozu Yamamoto:	Ouch, glad to have yah Sidewinder
[8:22]	Andrew Linden:	(4) We totally rip the RCCS system out, move to Havok5, and use ListShapes of convex parts for all concave primitives
[8:22]	Ryozu Yamamoto:	So, the crash problem is in the RCCS?
[8:22]	Seifert Surface:	the havok crash is to do with shape lod
[8:23]	Ryozu Yamamoto:	That would explain my crash
[8:23]	Andrew Linden:	Hrm... I may have a 5th idea, but I can't remember what it was.
[8:23]	Andrew Linden:	Sidewinder, I haven't informed you yet, but I received a response from Havok about their progress on hunting the crash bugs
[8:24]	Andrew Linden:	and they have not been able to reproduce them using their own demo/test code
[8:24]	Sidewinder Linden:	ahh is that separate from the email late yesterday?
[8:24]	Sidewinder Linden:	yes i saw that one...
[8:24]	Sidewinder Linden:	is there any way that we can reproduce it reliably here?
[8:24]	Andrew Linden:	Oh ok, I hadn't noticed that it as CC'd to you.
[8:24]	Sidewinder Linden:	yup np
[8:24]	Andrew Linden:	I was just thinking about what to do about it this morning on the way to work
[8:24]	Andrew Linden:	And was listing the possibilities here.
[8:25]	Sidewinder Linden:	yes, i saw the end of it i think - 4...
[8:25]	Ryozu Yamamoto:	If it's what I'm thinking of, I can repro it pretty reliably
[8:25]	Ryozu Yamamoto:	Well, from an in-world sense
[8:25]	Andrew Linden:	yeah, we have a reliable repro Ryozu
[8:25]	Andrew Linden:	however Havok was trying to repro using their own non-SL test harness
[8:26]	Sidewinder Linden:	so the issue is that we can make it happen here, but they can't? do you think if they step through things in debug here will it tell them anything?
[8:26]	Andrew Linden:	they had copied the RCCS code into a little test world of theirs
[8:26]	Sidewinder Linden:	hmmm
[8:26]	Sidewinder Linden:	so maybe it's not in rccs proper?
[8:26]	Andrew Linden:	Yes, I suspect they could successfully debug here, however they prefer to work in their own office
[8:27]	Andrew Linden:	So idea (2) was to try to help them repro it there... perhaps using "grid in a box"
[8:27]	Andrew Linden:	however, I don't know the current state of "grid in a box" and it is probably non-working at the moment
[8:28]	Sidewinder Linden:	i think q has been working on that... i'll check, but i thik it's not quite ready to go yet
[8:28]	Andrew Linden:	currently the RCCS is instrumented to figure out when the phys engine is loaded
[8:29]	Andrew Linden:	so we might be able to make the RCCS take other actions (instead of dropping collision LOD) to try to stem the phys engine lag
[8:29]	Sidewinder Linden:	ok...
[8:29]	Andrew Linden:	that was idea (3)
[8:29]	Sidewinder Linden:	is there a way to stub out rccs and see if it still crashes? just as a simple check for where the problem is?
[8:30]	ToonjaWatchman Balbozar:	hey guys ssup..
[8:30]	Sidewinder Linden:	hi toonja
[8:30]	ToonjaWatchman Balbozar:	hi you okey?
[8:30]	Andrew Linden:	Idea (4) is to upgrade to Havok5, and lean heavily on their improvements to their hkListShape
[8:30]	Andrew Linden:	and get rid of concave shapes altogether
[8:30]	Andrew Linden:	welcome ToojaWatchman.
[8:30]	Sidewinder Linden:	a possibility... but i would guess that's a significant delay...
[8:31]	ToonjaWatchman Balbozar:	so are you like the owners of secondlife or somin
[8:31]	Ryozu Yamamoto:	I'd imagine so
[8:31]	ToonjaWatchman Balbozar:	you kool
[8:31]	Thickbrick Sleaford:	hello physicists
[8:31]	Sidewinder Linden:	toonja this is a meeting to discuss the status of a beta preview of a new version of the second life simulator
[8:31]	Seifert Surface:	how would you deal with concave stuff then?
[8:31]	Andrew Linden:	Yes well, I'm just listing the ideas I had. Dunno what is the right one yet.
[8:31]	Ryozu Yamamoto:	Hey there Toonja, this is an office hours meeting. We're just talking about the status of Linden Lab's upgrade to Havok4, the physics engine that controls how things move and bounce around in SL
[8:31]	ToonjaWatchman Balbozar:	who me?
[8:31]	Ryozu Yamamoto:	Feel free to hang out, listen in, or throw in ideas if you have them
[8:32]	Andrew Linden:	Concave stuff would collide as a list of convex parts that have the same aggregate collision shape.
[8:32]	ToonjaWatchman Balbozar:	hehe science huh
[8:32]	Ryozu Yamamoto:	Heya Pulse, Zero's hour over?
[8:32]	Thickbrick Sleaford:	is this a good place to ask about a specific bug?
[8:32]	PulseBurst Flow:	No, couldn't get in
[8:32]	PulseBurst Flow:	region is full
[8:32]	Seifert Surface:	sounds like that could involve a large number of parts for certain shapes
[8:32]	Ryozu Yamamoto:	Thickbrick: if it's related to the Beta test or physics engine, perhaps
[8:32]	ToonjaWatchman Balbozar:	skeen
[8:33]	Thickbrick Sleaford:	the llTargetOmega bug, that's supposed to be fixed in hacok4, but I think isn't
[8:33]	Thickbrick Sleaford:	SVC-54
[8:33]	ToonjaWatchman Balbozar:	ble$$ im over thr
[8:33]	Sidewinder Linden:	thickbrick... can you repro the broken behavior?
[8:33]	Andrew Linden:	Yes perhaps seifert. However our research into it in the last Havok2 port project suggests that large numbers of simple shapes work better than concave shapes in Havok.
[8:33]	Seifert Surface:	sounds like a win then
[8:33]	Andrew Linden:	SVC-54 is fixed in Havok4
[8:34]	Andrew Linden:	and I also ported it to a maintenance branch... it should trickle out in the next few weeks
[8:34]	Thickbrick Sleaford:	yes, Andrew. my windmill that sort of works in the main grid doesn't in the beta grid
[8:34]	Seifert Surface:	the worst i can think of for prims would be hollow tori
[8:34]	Seifert Surface:	are there good algorithms for chopping up concave things?
[8:34]	Andrew Linden:	None that would be easy to use for us Seifert, I suspect.
[8:35]	Andrew Linden:	We'd have to roll our own.
[8:35]	Andrew Linden:	However, Don Linden once worked on it, and had it half done.
[8:35]	Seifert Surface:	sounds hard in general
[8:35]	Seifert Surface:	especially with sculpties
[8:35]	Sidewinder Linden:	andrew... do you think we could localize the problem by stubbing rccs out - at least ot see if that is really where the source is? i'm wondering, if it does not repro with havok's test harness, whether the problem might be somewhere around rccs, but not actually in it?
[8:35]	Andrew Linden:	In general yes, however since our shapes (except for sculpties) are all extruded path/profile, it shouldn't be too hard.
[8:36]	Seifert Surface:	hmm, any subset of a hollow sphere is concave, until you get down to individual polygons
[8:36]	Hevenz Vansant:	im here for the free donuts
[8:36]	Hevenz Vansant:	:p
[8:36]	Sorraya Pera:	hahahaha
[8:36]	Ryozu Yamamoto:	Dunno about that. Seifert has some pretty interesting sculpties
[8:36]	ToonjaWatchman Balbozar:	hehe
[8:37]	ToonjaWatchman Balbozar:	go on hevenz lol
[8:37]	Seifert Surface:	oh jeez, those would test it
[8:37]	Ryozu Yamamoto:	Hey Hypatia
[8:37]	Hypatia Callisto:	hi hi :)
[8:37]	Andrew Linden:	I'm ignoring sculpties for now... they don't have correct collidable shapes anyway.
[8:37]	Ryozu Yamamoto:	Weird table =/
[8:37]	Sidewinder Linden:	well here is a high level question...
[8:37]	PulseBurst Flow:	LOL
[8:38]	Hevenz Vansant:	:p
[8:38]	Sidewinder Linden:	if we have one known crash mode
[8:38]	Sidewinder Linden:	do we ship with it or not, assuming that it takes "a while" to figure out a solution
[8:38]	Sidewinder Linden:	my thinking is sort of split
[8:38]	Ryozu Yamamoto:	If you guys can get down to ONE known crash mode
[8:38]	Sidewinder Linden:	on one hand, we could remove a huge number of spurious crashes
[8:38]	ToonjaWatchman Balbozar:	what the hell
[8:38]	Andrew Linden:	We may be able to ship with it in some limited fashion.
[8:38]	Andrew Linden:	As I mentioned, Phillip suggested that we do exactly that.
[8:38]	Sidewinder Linden:	but having one crash mode is enough for griefers to take advantage of the hole whenever they want
[8:38]	Ryozu Yamamoto:	And the other bugs are as close to unnoticable
[8:38]	Ryozu Yamamoto:	I'd say ship
[8:39]	Andrew Linden:	Find some regions (estates) that crash a lot, and ask them if we can bring them up on Havok4 on a trial basis
[8:39]	Sidewinder Linden:	then it seems to me that going to h5 is not the path, since that's a long delay for unknown benefit (not guaranteed to fix the problem)
[8:39]	Andrew Linden:	some estate owners may find that their crash/lag goes down
[8:39]	Sidewinder Linden:	i'm really talking about strategy for how and what to fix, not how we roll-out
[8:39]	Ryozu Yamamoto:	Sidewinder: You'd be suprised how long it may take some of those type to figure how to crash something
[8:39]	Andrew Linden:	and none of the important content in their regions tickles the remaining crash bug
[8:39]	Andrew Linden:	however...
[8:39]	WarKirby Magojiro:	There's a havok 5 now?
[8:40]	Andrew Linden:	some content is already known to tickle it.
[8:40]	Ryozu Yamamoto:	Which is that?
[8:40]	Ryozu Yamamoto:	Aside from that specifically intended to break things
[8:40]	WarKirby Magojiro was about to ask
[8:40]	Andrew Linden:	Havok5 exists, however we are not trying to use it yet. We're on Havok4.6 in the phys upgrade project.
[8:40]	Sidewinder Linden:	btw warkirby h5 is *very* new... (with all that goes with that.. ;)
[8:41]	WarKirby Magojiro:	I see
[8:41]	WarKirby Magojiro:	Will it be easy to add when the time comes?
[8:41]	Gaius Goodliffe:	Well, Havok1 crashes my sim often enough. If Havok4 ships with just one bug, it won't be any worse at any rate...
[8:41]	PulseBurst Flow:	Hmmm.you would have to inform content creators to avoid the bug.
[8:41]	WarKirby Magojiro:	If there's a consistent crash bug,
[8:41]	WarKirby Magojiro:	people will create things to exploit it
[8:41]	Sidewinder Linden:	that's a tricky one pulse, as that also tells anyone elese how to build a purposeful one...
[8:41]	WarKirby Magojiro:	and use it as a weapon
[8:41]	Andrew Linden:	Under the scenario I was just talking about the Haovk4 code would not roll out on all of SL
[8:41]	Sidewinder Linden:	yes war... agreed
[8:41]	PulseBurst Flow:	yes...
[8:41]	Seifert Surface:	are there publically known ways to crash a sim now, that are at least as bad as the sim craqsh in h4?
[8:41]	PulseBurst Flow:	it's a problem...
[8:41]	Andrew Linden:	but would only be deployed in a few places on a trial basis
[8:41]	PulseBurst Flow:	but no good solution..
[8:41]	Sidewinder Linden:	yes seifert...
[8:42]	Andrew Linden:	meanwhile we would be trying to fix the remaining crash bug
[8:42]	WarKirby Magojiro:	I crashed a sim with script overload once
[8:42]	Seifert Surface:	sounds like a pretty clear decision to me
[8:42]	PulseBurst Flow:	well, you are confident it is fixable then?
[8:42]	Andrew Linden:	er... remaining phys engine crash bug
[8:42]	PulseBurst Flow:	not just putting it off, and never fixing it?
[8:43]	Sidewinder Linden:	oh - right - guys - in terms of expectations there are crash bugs in other places
[8:43]	Sidewinder Linden:	but we're trying to resolve all that we find through this project
[8:43]	WarKirby Magojiro:	including non physics issueS?
[8:43]	Sidewinder Linden:	there have been a few lsl-induced crashes fixed during the project
[8:44]	Andrew Linden:	I expect the LL efforts to fix simulator crash bugs will increase greatly when Havok4 is deployed to the entire world of SL
[8:44]	Andrew Linden:	currently there are some bugs in Havok1 that we cannot fix, and they are 99% of all simulator crashes
[8:44]	Gaius Goodliffe nods.
[8:44]	Ryozu Yamamoto:	Oh, Sidwinder: I'm thinking that's a yes. I was on main grid Bug Island the other day when some guys pulled out a sim an instant sim crash onme
[8:44]	Ryozu Yamamoto:	On the known severe sim crash bugs
[8:45]	Gaius Goodliffe:	Physics is definately the number one killer out at Dogfight Atoll.
[8:45]	Sidewinder Linden:	one thing that would be helpful, is if you all could go through the pjira bug list, and email me with suggestions on which ones you think are not showstoppers for shipping to a first set of regions on the main grid
[8:45]	Andrew Linden:	Havok4 fixes most of those, but has at least one (as of yet unfixable) crash bug that is probably less frequent than those of Havok1
[8:45]	WarKirby Magojiro:	There are rumors of a single unscripted object that crashes sims merely by existing
[8:45]	Sidewinder Linden:	warkirby, there is a gun that if you rez it, without even trying to shoot it, will kill an h1 sim
[8:46]	Ryozu Yamamoto raises an eyebrow.
[8:46]	WarKirby Magojiro:	Why is that?
[8:46]	Sidewinder Linden:	now we know what it looks like, and it's pretty umm lightweight in terms of what it does when it works ;)
[8:46]	WarKirby Magojiro:	physics and lots f toruses?
[8:46]	Andrew Linden:	Warkirby, that is probable, however when the crash rate drops it will be easier to focus on the few remaining fixable crash bugs.
[8:46]	Sidewinder Linden:	on h4 it's a non-issue
[8:46]	Gaius Goodliffe:	cool
[8:46]	Sidewinder Linden:	has anyone found new content creation or linkage style bugs - things that would break existing content?
[8:46]	Ryozu Yamamoto:	As long as you can say crash rates go down, I'd say to ship. Even if there's known crash modes, people will be happy that it's at least more stable
[8:47]	Gaius Goodliffe:	Yes
[8:47]	Seifert Surface:	well, go down significantly
[8:47]	Gaius Goodliffe:	I have an airship that works fine on the main grid but won't on the h4 beta.
[8:47]	Seifert Surface:	there will probably be things not working that we havent caught
[8:47]	Sidewinder Linden:	right - as long as we don't "break your stuff", and that's what we're working on now is to make sure that things will rez correctly and not end up broken
[8:47]	Andrew Linden:	Er... what if we ship and it breaks YOUR content Ryozu, but brings the crash rates to a halt?
[8:47]	Seifert Surface:	people will be annoyed that their vehicles and games act differently
[8:47]	Ryozu Yamamoto:	I'll cry ;_;
[8:47]	WarKirby Magojiro:	Needs to be sure all legitimate content still works, or can adapt to work
[8:47]	Sidewinder Linden:	@andrew - that's *the* question :)
[8:47]	Gaius Goodliffe:	OTOH, it could be argued it should never have worked to begin with... :p
[8:48]	Sidewinder Linden:	gaius - which airship is that?
[8:48]	Ryozu Yamamoto:	Heh, I meant to imply that most things should still work alright
[8:48]	PulseBurst Flow:	Well, not knowing what it is exactly, but just knowing there is such a bug, is worrisome if planning content...will it be compatible.
[8:48]	Gaius Goodliffe:	That one. :)
[8:48]	Andrew Linden:	I don't expet zero content breakage under Havok4. In fact, I've intentionally broken some "content". Blitz guns for example.
[8:48]	Sidewinder Linden:	pulse - you'll know pretty quickly if you hit it ;)
[8:48]	Gaius Goodliffe:	You can see the problem already I'm sure....
[8:48]	Sidewinder Linden:	gaius?
[8:49]	PulseBurst Flow:	Well, I can't even Beta..
[8:49]	Sidewinder Linden:	ahh
[8:49]	PulseBurst Flow:	updating the preview database anytime soon?
[8:49]	Ryozu Yamamoto:	Nice shiny blimp O_o
[8:49]	WarKirby Magojiro:	Is this crash reproducible?
[8:49]	Andrew Linden:	The database is updateing now, or is done.
[8:49]	Seifert Surface:	any feeling on the percentage of legitimate content that would be affected?
[8:49]	Andrew Linden:	Yes, Warkirby the crash is reproducable, but crashes deep in Havok4.
[8:50]	PulseBurst Flow:	So Andrew, I can beta now?
[8:50]	Sidewinder Linden:	QA is testing the preview, and as soon as it looks like the db updates went cleanly we'll open it
[8:50]	Andrew Linden:	However, it is related to the RCCS (runtime collision control system) that is non-Havok4 code we wrote to cap the lag.
[8:50]	WarKirby Magojiro:	If there's a solid repro, what's stopping it being fixed?
[8:50]	Andrew Linden:	PulseBurst, you can beta by the end of today I expect.
[8:50]	Ryozu Yamamoto:	Sweet,
[8:50]	PulseBurst Flow:	thanks guys
[8:51]	Ryozu Yamamoto:	WarKirby: Repro only goes so far, you still have to get feedback from the code on just -what- it's doing
[8:51]	WarKirby Magojiro:	So setup lots of debug messages
[8:51]	Andrew Linden:	WarKirby, the Havok engineers do not yet have a fix, AND they have not yet been able to repro the crash in their own test code.
[8:51]	WarKirby Magojiro:	make it happen, and see what the logs say
[8:51]	WarKirby Magojiro:	hm
[8:51]	WarKirby Magojiro:	Seems like it'd be best to wait for it to be fixed, then
[8:52]	WarKirby Magojiro:	Can't hide forever...
[8:52]	Ryozu Yamamoto:	Heh
[8:52]	WarKirby Magojiro:	I'll have to test my ki blasts in havok 4 later
[8:52]	Sidewinder Linden:	i have another strategy question... the one i've been asking each of the past couple of weeks... are we to the point that i should do the "big push" for product vendors to try all their wares on h4?
[8:52]	Andrew Linden:	Well, since we've got such a big crowd here at the end of today, I should mention...
[8:53]	ToonjaWatchman Balbozar shouts: monobrow
[8:53]	Seifert Surface:	sidewinder: do you expect most things to work now, or not?
[8:53]	WarKirby Magojiro:	Sidewinder. I'd like to see the Omniohaze functional again before that, personally
[8:53]	WarKirby Magojiro:	Omniphaze*
[8:53]	Sidewinder Linden:	heh... well.... not quite sure seifert
[8:53]	Seifert Surface:	if 80% of vehicles are going to be broken, seems too early
[8:53]	Andrew Linden:	that the Havok4 preview now has some bugs fixed (in particular a center of mass problem) that should fix a large variety of odd problems
[8:53]	Ryozu Yamamoto:	XD
[8:53]	Ryozu Yamamoto:	I'll test the Phaze out soon as it's back up
[8:54]	Sidewinder Linden:	well i guess why don't we see how the experience goes with this new build, and then go from there...
[8:54]	Ryozu Yamamoto:	I have a feeling it'll work a bit better
[8:54]	Andrew Linden:	so I'd appreciate it if anyone who has already filed a bug about broken content could come in and re-test if any of those bugs are still open in jira.secondlife.com
[8:54]	Sidewinder Linden:	the hope is that we've got some key things straightened out that will help a variety of problems
[8:54]	WarKirby Magojiro:	I don't see a need for urgency, personally
[8:54]	Thickbrick Sleaford gave you Smooth Windmill Turret (still broken after SVC-54 ?).
[8:54]	WarKirby Magojiro:	we've lived with havok 1 for years
[8:54]	WarKirby Magojiro:	a few more weeks or even months wouldn't hurt
[8:54]	Ryozu Yamamoto:	That's true.
[8:54]	Gaius Goodliffe:	ow
[8:54]	Sidewinder Linden:	warkirby - the urge is that people lose content when a sim rollback has to be done... so as we wait, more people lose content
[8:54]	Andrew Linden:	Yeah...
[8:55]	Ryozu Yamamoto:	Ironic
[8:55]	Ryozu Yamamoto:	Some people "gain" content
[8:55]	Andrew Linden:	let me list the reasons for urgency...
[8:55]	WarKirby Magojiro:	Yes, there is that
[8:55]	Sidewinder Linden:	so we are trying to balance getting rid of sim crashes, and reducing content loss through sim crashes - against not having things work after the update
[8:55]	Andrew Linden:	(1) my own sanity. I've been working on this project for far longer than we expected it to last.
[8:55]	Thickbrick Sleaford:	I'm off. Thanks for doing this in the open, Lindens!
[8:56]	WarKirby Magojiro knows how that feels
[8:56]	Sidewinder Linden:	yw thickbrick
[8:56]	Andrew Linden:	(2) Some influential lindens spend their time watching the crash rates, and they are eager to see that number go down.
[8:56]	WarKirby Magojiro:	I see
[8:56]	Andrew Linden:	Meh... I'm partially joking.
[8:57]	PulseBurst Flow:	only partially. LOL
[8:57]	Andrew Linden:	Yes, a few more weeks or months won't be the end of SL
[8:57]	Ryozu Yamamoto:	Heh
[8:57]	Hypatia Callisto:	to be fair, things are way better now regarding sim crashing than it was say, spring 2006 with the massive grid crashing.
[8:57]	Andrew Linden:	I'm just motivated to get Havok4 done.
[8:57]	Sidewinder Linden thanks andrew for being so solid through a very long path... :)
[8:57]	WarKirby Magojiro:	: )
[8:57]	Hypatia Callisto:	indeed :)
[8:57]	Gaius Goodliffe:	mmph... I remember that.
[8:57]	Ryozu Yamamoto:	While I have this many people sitting at the table at once, can I ask a favor of you all?
[8:57]	Seifert Surface:	oh, do you think we could get a followup blog post from michael on megaprims?
[8:57]	WarKirby Magojiro:	Has Cubey Terra ever attended this meeting?
[8:58]	PulseBurst Flow:	Hmmm...less sim crashing, more viewer crashing?
[8:58]	WarKirby Magojiro:	This will affect a lot of his prodcts radically
[8:58]	Andrew Linden:	What is the favor Ryozu?
[8:58]	Ryozu Yamamoto:	I need a screenshot with a relativly loaded table for, ooh, bye table
[8:58]	Ryozu Yamamoto:	Nevermind, hehe
[8:58]	Hypatia Callisto seconds Seifert on that :)
[8:58]	Gaius Goodliffe:	Cubey's filed some JIRA stuff.
[8:58]	WarKirby Magojiro:	what happened to the table?
[8:58]	Andrew Linden:	Yeah, I set the autoreturn of this parcel to 60 minutes
[8:58]	Andrew Linden:	60 minutes is now up
[8:59]	Ryozu Yamamoto chuckles
[8:59]	Sidewinder Linden:	that's one way to time a meeting ;)
[8:59]	Ryozu Yamamoto:	Cue end of the meeting ;)
[8:59]	Andrew Linden:	Ryozu, shall we rez a new table and all sit for a screenshot?
[8:59]	WarKirby Magojiro:	Can you rez another quickly for a picture?
[8:59]	Ryozu Yamamoto:	Nah, it's all good =)
[8:59]	PulseBurst Flow:	The table disappears evrery time Ryozu coughs.;-D
[8:59]	Ryozu Yamamoto:	I'm not in any hurry
[8:59]	trombone del ostia whispers: Man you sure you can handle the Third Toke!
[8:59]	trombone del ostia whispers: Playing...
[8:59]	Andrew Linden:	sure, but we all have to stand up then resit
[8:59]	Gaius Goodliffe:	Oh dear, time flies.
[8:59]	WarKirby Magojiro:	We can do it next week, I suppose
[8:59]	Ryozu Yamamoto:	That's fine by me ;)
[9:00]	Ryozu Yamamoto:	Thanks for coming guys
[9:00]	WarKirby Magojiro:	I'll hard to tell which charis are for the table
[9:00]	Ryozu Yamamoto:	Guess I really should do something about those silly chairs
[9:00]	WarKirby Magojiro:	the one beside hypatia
[9:01]	Chair: Press Page Up to move chair up, or Page Down to move chair down
[9:01]	Ryozu Yamamoto:	Also, anyone who wants a copy of the table, just let me know
[9:01]	PulseBurst Flow:	When the music stops, everyone grab a chair.
[9:01]	WarKirby Magojiro:	heh
[9:01]	Andrew Linden:	Everybody have a seat.
[9:01]	Hevenz Vansant:	sorry side didnt mean to sit on ya head :p
[9:02]	Sidewinder Linden:	hi cubey
[9:02]	Cubey Terra:	hi
[9:02]	Andrew Linden:	somebody else take a screenshot
[9:02]	Ryozu Yamamoto:	Woah, heya cubey!
[9:02]	Gaius Goodliffe:	Speak of the devil...
[9:02]	WarKirby Magojiro:	Speak of the devil :)
[9:02]	WarKirby Magojiro:	jinx
[9:02]	Gaius Goodliffe:	lol
[9:03]	Cubey Terra:	I didn't know about the havok office hours :)
[9:03]	Ryozu Yamamoto:	Whoever wants a copy, just let me know
[9:03]	Ryozu Yamamoto:	Thanks guys
[9:03]	Seifert Surface pinged cubey
[9:03]	Andrew Linden:	alas, today's hour is now up Cubey
[9:03]	Hypatia Callisto:	Hee
[9:03]	WarKirby Magojiro:	I was just asking if you had ever been here
[9:03]	Cubey Terra:	doh!
[9:03]	Sidewinder Linden:	ok guys, so as soon as the preview is open i'll post a blog about it - should be by later today
[9:03]	WarKirby Magojiro:	since you're one of those most strongly affected
[9:03]	Sidewinder Linden:	let's go and retest all of our known open bugs, ok?
[9:03]	Ryozu Yamamoto:	Yep!
[9:03]	Sidewinder Linden:	because we are hoping that a large proportion of them are now resolved
[9:03]	PulseBurst Flow:	Thanks Sidewinder, Andrew
[9:03]	Andrew Linden:	Thanks everyone.
[9:04]	ToonjaWatchman Balbozar:	yoyoyo
[9:04]	Hypatia Callisto:	looking to get in as soon as it reopens