User:Andrew Linden/Office Hours/2009 03 24

Transcript of Andrew Linden's office hours:

[11:00]	Wolfhaven Teleportation HUD v3.0 is Ready
[11:00]	Arawn Spitteler:	Just made it? I see andrew standing.
[11:00]	Liisa Runo:	me and my neighbors have sent AR once a day now, for 2 days, but it is still there, and sim is useless, TD and FPS is 50% and script takes 1.5 to 3 seconds to llSleep(1.0);
[11:00]	Kitto Flora:	Hello hello
[11:01]	Moon Metty:	hi Andrew
[11:01]	Arawn Spitteler:	I was doing a few tests on SVC-3945 and I'm wondering, is it possible for an invisible phantom prim to block sitting?
[11:01]	Twisted Laws:	bummer, maybe try the support track.
[11:01]	Susie Chaffe:	ooops sorry Kitto
[11:01]	Arawn Spitteler:	Kitto found a nice seat
[11:01]	Simon Linden:	They could ... the sit location is based on trying to figure out what you clicked, so that might get in the way
[11:01]	Kitto Flora:	It was comfly! :)
[11:02]	Liisa Runo:	yea, that will be the next thing
[11:02]	Kitto Flora:	Phantoms intecepting clicks - a long time problem/bug/griefing tool
[11:02]	Arawn Spitteler:	So, sitting around a shrubbery, could fail, if we don't see the shrub in the way.
[11:02]	Kitto Flora:	Er - Alphas not phantoms
[11:02]	Eata Kitty:	Sounds like Z buffer fun
[11:02]	Andrew Linden:	Deep breath... I must stop working on that bug because OfficeHours are about to start.
[11:03]	Stev Spiegel:	may I?
[11:03]	Arawn Spitteler:	I didn't sheck whether the shrub was actually phantom, but I couldn't sit on the bush either
[11:03]	Moon Metty:	:)
[11:03]	Kitto Flora:	Andrew needs a dual core.
[11:03]	Eata Kitty:	Can I have a booster seat
[11:04]	Andrew Linden:	Eata, the chair moves. Use PageUp
[11:04]	Simon Linden:	I keep meaning to edit this table/chair set to they can raise up higher
[11:04]	Eata Kitty:	Woah
[11:04]	Eata Kitty:	Technology!
[11:04]	Kitto Flora:	Eata is a Petster?
[11:04]	Andrew Linden:	Simon, got any announcements? I don't have much.
[11:05]	Simon Linden:	No, nothing new I can think of
[11:05]	Andrew Linden:	I forget what I was working on since the last meeting, in fact. But I know what I'm working on today (some simulator shutdown bug in main-server-7 branch).
[11:05]	Kitto Flora:	Shuts down too much? OR not enough?
[11:06]	Andrew Linden:	Does not properly shutdown... never saves simstate on shutdown.
[11:06]	Kitto Flora:	Oh
[11:06]	Andrew Linden:	I don't think we'll want to deploy that bug.
[11:06]	Arawn Spitteler:	Isn't that a neat new feature, Abend on standard restart?
[11:07]	Stev Spiegel:	:)
[11:07]	Andrew Linden:	It could be. I should add an [Extra Content Loss] button in the Estate management tools.
[11:07]	Susie Chaffe:	thank you but the current content loss tools are working just fine
[11:07]	Andrew Linden:	Ok anyway, I don't have anything to contribute to the agenda today (at least not yet).
[11:08]	Andrew Linden:	Someone will have to make some suggestions.
[11:08]	Kitto Flora:	Whats likely to get fixed soon?
[11:08]	Helena Lycia:	Hello
[11:08]	Simon Linden:	Ah, not sure if we discussed this last time, but our next server will still be a 32 bit binary. Moving to 64bit caused memory usage to grow, and we're not totally sure if that's a problem or not
[11:09]	Arawn Spitteler's found a posible new feature, and might jira as such, for all it might compeat with SLim: I got lag-orbitted to a point that I had to relog, when several TPs sent fialed to help. Appearently, there's a half crash state, where we don't exist on the grid, but can still communicate.
[11:09]	Andrew Linden:	Well, I was working on the solution to the prim-encroachment problem, but maint-server-7 is now blocked and that is my #1 top priority. There are some pending fixes in maint-server-7, and -8.
[11:09]	Andrew Linden:	Those fixes are the things that would be in server-1.27
[11:09]	Andrew Linden:	anything fixed tomorrow would probably go into server-1.28, whenever that is
[11:10]	Arawn Spitteler:	1.27 is coming to an Aditi?
[11:10]	Andrew Linden:	so I guess I'm wondering... are you interested in what might be in server-1.28, or what I think will be included in server-1.27?
[11:10]	Helena Lycia:	Hiya Rex
[11:10]	Rex Cronon:	hello everybody
[11:10]	Rex Cronon:	hi helena
[11:10]	Kitto Flora:	Finally resolving the Megaprim situation would remove a lot of urban myth thats going around
[11:10]	Andrew Linden:	No, server-1.26 is on aditi. Hrm... I wonder what the schedule is on that.
[11:11]	Andrew Linden:	server-1.27 hasn't branched yet, but I've got to get maint-server-7 in shape before it branches.
[11:11]	Andrew Linden:	What are the ubran myths you speak of Kitto?
[11:11]	Andrew Linden:	s/ubran/urban/
[11:12]	Kitto Flora:	Latest thing seems to be - people re-packaging the old megaprims into new folders then passing them out claiming that all kinds of theings 'have been fixed'
[11:12]	Kitto Flora:	Implying some kind of new exploit. But I cant find eny hard evidence for such
[11:12]	Arawn Spitteler:	What was broken?
[11:12]	Kitto Flora:	Oh - and theres a 'new hacked client' that can make megaprims.
[11:13]	Andrew Linden:	Oh hrm... a new exploit would be interesting. I haven't heard of one.
[11:13]	Kitto Flora:	Well essentialy - in SL one can no longer create megaprims.
[11:13]	Eata Kitty:	I haven't heard of anything but there are already loads of types around
[11:13]	Eata Kitty:	I mean, from before
[11:13]	Kitto Flora:	Those who want megaprims desciibe that as 'broken'
[11:13]	Arawn Spitteler:	Hacked Clients can no longer, according to history, but 1.25.6 might be buggy
[11:14]	Andrew Linden:	Interesting. So you'd have to special-order some strange megaprim size, and if they can deliver it then it would be true.
[11:14]	Helena Lycia:	Could megaprims be introduced gradually? slowly increase the max size of createdable prims and monitor how they behave before moving to a larger creatable size?
[11:14]	Kitto Flora:	ANdrew: Yes
[11:14]	Andrew Linden:	No Helena, a gradual increase of prim size shouldn't work.
[11:14]	Helena Lycia:	Ah OK
[11:15]	Arawn Spitteler:	Meaning Prim Limiots can't be slid?
[11:15]	Andrew Linden:	Ok, well if anyone can verify a new megaprim exploit I'd like to hear about it (so I could close it, of course).
[11:15]	Eata Kitty:	I think mega prims should be a special limited case. Like, purely for sim builders visual use. So you can't script them or anything strange to avoid weirdness
[11:15]	Andrew Linden:	Oh Helena, I thought you were asking about the ability for a Resident to gradually increase prim size to megaprim range.
[11:15]	Arawn Spitteler:	Prim Size Limits as an estate setting; that migh be a server setting?
[11:16]	Helena Lycia:	No
[11:16]	Rex Cronon:	and i was thinking of making a megaprim tv screen, but without scripting it won't be possible:(
[11:16]	Helena Lycia:	I mean LL upping the max size to 20m and seeing how the world behaves
[11:16]	Kitto Flora:	I think what Andrew was headed for is stil the best solution
[11:16]	Andrew Linden:	I had mentioned in the past that it might be possible to expand the supported prim size limit "gradually" in the sense of not boosting it immediately to something like 256m, but to 20m, then 40m a few months later.
[11:16]	Helena Lycia:	and if that's good going to 30 or 50m and so on
[11:16]	Andrew Linden:	However, such a policy change must wait for some prerequisites (IMO).
[11:17]	Helena Lycia:	Oh
[11:17]	Andrew Linden:	Yes, that would be possible.
[11:17]	Kitto Flora:	YEah - get the encroachment fixed first
[11:17]	Arawn Spitteler:	Prim Encroachment would be a problem, even at pr3esent. I've gone through ban-lines using Sit Target
[11:17]	Helena Lycia:	I could soo really use larger prim sizes, even if it was only upto 20m
[11:18]	Eata Kitty:	Well sit target you aren't technically in the ban area so I don't think that matters as such
[11:18]	Arawn Spitteler:	20 meters would allow a circular bench at the chat range limit
[11:18]	Helena Lycia:	You can get 300m with a sit target regardless of the prim size
[11:18]	Andrew Linden:	The two main prerequisites in my mind are: (1) a solution to prim-encroachment and (2) static objects collide across region boundaries
[11:18]	Helena Lycia:	You can make one with a 10m tube and some clever twisting
[11:18]	Kitto Flora:	1) is easy - just needs doing?
[11:18]	Helena Lycia:	(a 20m disk) ((well 19.5m dick)
[11:18]	Helena Lycia:	Err disk
[11:18]	Kitto Flora:	2) Is a bit more complicated - more to add?
[11:18]	Andrew Linden:	(1) is mostly done
[11:19]	Andrew Linden:	Yes, I think (2) is harder.
[11:19]	Arawn Spitteler just turns it upside down: What's the possibility of negative profile cuts?
[11:19]	Eata Kitty:	Is prim enroachment just the old thing of the fact they overlap because it judges it by the center point
[11:19]	Kitto Flora:	2) already fails with existing prims. A known problem
[11:19]	Yann Dufaux:	hello:)
[11:19]	Moon Metty:	hi Yann :)
[11:20]	Andrew Linden:	Currently the notion of whether an object is on a parcel pivots entirely on the world-position of the root-prim's geometrical center
[11:20]	Arawn Spitteler:	Static Objects colliding across region boundaries? I thought they were supposed to be Phantom
[11:20]	Andrew Linden:	(geometrical center != center of mass)
[11:21]	Arawn Spitteler:	Private Builds often reach over protected land
[11:21]	Andrew Linden:	Arawn, I would like them to cease to be phantom (collisionless) across region boundaries.
[11:21]	Kitto Flora:	The Encroachment problem is that your neighbor can rez an object with center on his land but it extends into your land. Encroaching.
[11:21]	Simon Linden:	We're not thinking about killing all encroachment, but giving the tools for the parcel owner to deal with neighboring prims that overlap into their land
[11:22]	Andrew Linden:	Hrm... which reminds me... prim-encroachment across region boundaries would still be possible under the initial state of that project. I'll have to work on that
[11:22]	Arawn Spitteler:	Roadways, such as The Wall, should have redundant prims
[11:22]	Andrew Linden:	I might have to work on (2) sooner than I thought.
[11:22]	Kitto Flora:	Andrew: Yes .. but Encroachment within a region is stil the major problem
[11:22]	Andrew Linden:	yeah, I don't mind delivering a partial solution. Full speed ahead.
[11:23]	Kitto Flora:	I'd say 1) fixed would be good enough to deploy
[11:23]	Moon Metty:	the only time i was bothered by encroachment was over the sim border :)
[11:23]	Andrew Linden:	One interesting thing about the prim-encroachment solution project is getting the rest of LL on board.
[11:23]	Kitto Flora:	Moon: What happened to resolve the probem?
[11:23]	Eata Kitty:	Is it not thought of as a big issue?
[11:24]	Moon Metty:	well, i talked to the object owneres a few times
[11:24]	Eata Kitty:	I thought it would be the kind of thing people who own land love to moan about
[11:24]	Moon Metty:	then i rezzed some plywood inside their building, that got them thinking
[11:24]	Andrew Linden:	It is a big issue, but some Lindens would be against it if it broke too much existing content, or didn't have enough override options.
[11:24]	Kitto Flora:	Hahah a :)
[11:24]	Eata Kitty:	Oh I see. More of a legacy support thing
[11:25]	Andrew Linden:	For example, I already know that some estate owners would want to disable the "return on parcel overlap" feature.
[11:25]	Helena Lycia:	Oops
[11:25]	Andrew Linden:	Mostly because they have some estate content that overlaps parcels they lease to others.
[11:25]	Andrew Linden:	Similarly, it turns out, there is some use of that overlap feature on the Mainland.
[11:26]	Andrew Linden:	Luna region was mentioned as an example.
[11:26]	Kitto Flora:	Well it only becomes a problem IF the encroached party does the return
[11:26]	Kitto Flora:	Its like an Easement option
[11:26]	Kitto Flora:	You give the power company an Easement for their power pole
[11:27]	Andrew Linden:	Right, but some estate owners will prefer to deal with parcel owner complaints about prim-encroachment rather than deal with the possiblility of lost content for return.
[11:27]	Kitto Flora:	If you chansaw down the power pole, from your land, the power company cuts you off :)
[11:27]	Eata Kitty:	So it's an estate option
[11:27]	Eata Kitty:	Or parcel option
[11:27]	Kitto Flora:	Andrew: Yes, thats a better option.
[11:27]	Andrew Linden:	yes, it will be a per-region option
[11:27]	Rex Cronon:	u go "green" kitto:)
[11:28]	Andrew Linden:	I haven't yet implemented an estate-wide switch. I was hoping that per-region would suffice.
[11:28]	Kitto Flora:	Green costs a lotta green.
[11:29]	Andrew Linden:	so anyway, my next step in that project is to write copy for a "blog post" that will circulate internally
[11:29]	Kitto Flora:	An estate wide switch would save linden content on Maintenance land? Thats all the roads...
[11:29]	Andrew Linden:	and give Linden's an opportunity to debate my solution to prim-encroachment
[11:29]	Andrew Linden:	I just need to sit down and write something up.
[11:30]	Arawn Spitteler recalls there's abit of Linden Road Encroachment
[11:30]	Andrew Linden:	yes, that is an issue that will be up for debate
[11:30]	Arawn Spitteler:	Should estate holders, including Linden, have a right of Easement?
[11:30]	Andrew Linden:	whether Linden Content will be susceptible to the prim-encroachment returnability logic
[11:30]	Andrew Linden:	Exactly Arawn.
[11:31]	Kitto Flora can see 'french riots' happening :)
[11:31]	Arawn Spitteler doesn't know a riot fromk a usual french dialog: I thought that came up with Steam Sky City?
[11:31]	Helena Lycia:	I think estate owners and managers should be able to
[11:31]	Eata Kitty:	Well if they don't already complain about enroachment then it must not be that big of an issue
[11:32]	Stev Spiegel:	is not there a deeper philosophical issue here: the amount of fuzz that is tolerated?
[11:32]	Kitto Flora:	Far from it - mostly
[11:32]	Kitto Flora:	Try roadside residents offering free land to put more LL content on?
[11:32]	Arawn Spitteler:	Fuzz? I don't think you mean Cops, G-Team or other griefers
[11:33]	Stev Spiegel:	hehe
[11:33]	Andrew Linden:	Stev. Amount of "fuzz"? Are you refering to the "slop tolerance" of what a prim-encroachment would entail? Or are you bringing up a completely different topic?
[11:33]	Stev Spiegel:	well, yesh slop tolerance
[11:33]	Stev Spiegel:	but in general we pretend discreteness
[11:34]	Andrew Linden:	Some slop will be tolerated, actually. Probably about 0.1m of overlap will be allowed.
[11:34]	Yann Dufaux:	Neck
[11:34]	Yann Dufaux:	EXTEND!
[11:34]	Stev Spiegel:	overlap is a frame of reference issue
[11:34]	Kitto Flora:	Its going to be the Havok collision tolerance I'd guess
[11:34]	Stev Spiegel:	a scale issue
[11:34]	Andrew Linden:	Mostly because the "check for overlap" will actually involve using the physics engine collision query code, and it has some 0.1m slop.
[11:35]	Stev Spiegel:	but that does not have to be invaraint
[11:35]	Stev Spiegel:	can change, region, estate, etc
[11:35]	Arawn Spitteler:	At what pointdoes n Access Line orbit us off our carpets? Is it right where root location enters the sim?
[11:36]	Kitto Flora:	Any idea why thats there Andrew - do they just use low-definition math in Havok?
[11:36]	Andrew Linden:	Well, I could imagine increasing the accuracy of the overlap check later, but I'm not too interested in making the slop a per-region setting.
[11:36]	Andrew Linden:	Kitto, the slop in the collision is used by Havok to really really try to prevent collision penetration.
[11:37]	Arawn Spitteler:	I think Zoning requires fences to have a few inches set-back
[11:37]	Andrew Linden:	It used to be that Havok1 would have performance problems (and general bad physics) when two objects interepenetrated, so they tried very hard to prevent it.
[11:37]	Andrew Linden:	In Havok2 or 3 they introduced their own "penetration resolution code" that tries to push objects apart when the penetrate.
[11:37]	Stev Spiegel:	physics at rest is still physics
[11:38]	Kitto Flora thinks about that and wonders what ODE does.
[11:38]	Andrew Linden:	It still has performance problems, but it doesn't necessarily give totally useless collision info (as it did in Havok1)
[11:38]	Andrew Linden:	yeah, I don't know what ODE does.
[11:38]	Arawn Spitteler:	ODE?
[11:39]	Kitto Flora:	I think I may find out.. .soon :/
[11:39]	Andrew Linden:	From what I understand, Havok uses some "GJK" algorithm to compute closest aproach between two objects.
[11:39]	Kitto Flora:	ODE is the physics engine used by OpenSim
[11:39]	Andrew Linden:	The "GJK" would basically return "I don't know" if they happened to overlap at all.
[11:39]	Andrew Linden:	ODE = Open Dynamics Engine (I think)
[11:39]	Kitto Flora:	Something like that
[11:40]	Andrew Linden:	There are a few free physics engines. Natually one popular one uses the GNU license and the other uses a BSD licens
[11:40]	Andrew Linden:	I forget which one ODE is.
[11:41]	Andrew Linden:	Then there are a few "free as in beer" physics engines.
[11:41]	Kitto Flora:	I forget. Dont matter ... OpenSim is using ODE by default. Someone higher up has decided its 'ok'
[11:41]	Andrew Linden:	And some "free as in beer, if you aren't a corporation" licensed engines. I think Havok now falls into that categoryl
[11:41]	Kitto Flora:	OH?
[11:42]	Kitto Flora:	Havok has changed its licencing?
[11:42]	Andrew Linden:	Yeah, OpenSim is BSD license, I think. So ODE is probably BSD.
[11:42]	Andrew Linden:	Havok was purchased by Intel. Shortly thereafter the Havok physics libs became "free for non-commercial use"
[11:43]	Eata Kitty:	Guess they felt there was competition from the open source
[11:43]	Kitto Flora:	Wonder if they consider OpenSim users 'commercial'
[11:43]	Andrew Linden:	Meanwhile, PhysX was purchased by nVidia , and was already "free for non-commercial use, and maybe even for commercial use".
[11:44]	Andrew Linden:	That is a good question Kitto -- are commercial OpenSim deployments "commercial"? Probably.
[11:44]	Eata Kitty:	Speaking of... would it be possible to put a PhysX card, or maybe a GPU that does CUDA in simulators and then turn on continuous collision detection?
[11:45]	Kitto Flora:	Open Sim itself is not commercial - products are free to use
[11:45]	Kitto Flora:	But some grids using OpenSim are obviously commercial
[11:45]	Andrew Linden:	Eata, in theory it would be possible to boost the computing power of a simulator using a GPU
[11:45]	Kitto Flora:	Legend City for one...
[11:45]	Andrew Linden:	however in practice it probably wouldn't be a viable hardwar strategy for LL
[11:45]	Andrew Linden:	we currently put multiple simulators on a single host
[11:46]	Andrew Linden:	it isn't clear how we would share the GPU resources between the various simulators
[11:46]	Andrew Linden:	and also, the hosts we buy might not support PCI-e which is used for most GPU cards.
[11:47]	Andrew Linden:	any PCI-e card would have to fit into an L-bracket to fit into the server case
[11:47]	Andrew Linden:	and I wonder whether a PCI-e card would like that
[11:47]	Eata Kitty:	Thats a shame. It seems like it would free up the CPU a lot if a dedicated card can pick up the load
[11:47]	Andrew Linden:	the bandwidth is so high on a PCI-e bus an L bracket might introduce too much noise
[11:48]	Andrew Linden:	I've seen AGP cards fail on L-brackets in the past. Some worked, but others didn't.
[11:48]	Stev Spiegel:	heat issues?
[11:48]	Andrew Linden:	yes, heat would also be an issue
[11:48]	Kitto Flora:	Is the L-bracket and extender that allows PCI etc cards to be mounted parallel to the MoBo?
[11:48]	Arawn Spitteler:	Doesn't LL doe enough business to design its own servers?
[11:49]	Andrew Linden:	yeah, the card would have to sit parallel to the motherboard
[11:49]	Andrew Linden:	No LL doesn't do enough business to design its own servers yet.
[11:49]	Kitto Flora:	Yeah - that'd be unusual for PCIe
[11:49]	Eata Kitty:	When you have 7000 servers you aren't going to overhaul them very often :o
[11:49]	Eata Kitty:	Thats a guess as there are about 24k sims :p
[11:50]	Kitto Flora:	And I dont think theres any PCIe non-right-abgle (vertical) sockets
[11:50]	Arawn Spitteler:	3,650 servers a year is ten a day. Get some high school kids to put them togeather every week
[11:50]	Stev Spiegel:	mom look what we built today in shoppe class
[11:50]	Andrew Linden:	Actually, I asked #ops to count the servers the other day... I think the simulator host count is just under 6k right now.
[11:51]	Kitto Flora:	Thats 6k pizza boxes? Or cores?
[11:51]	Andrew Linden:	6k pizza boxen
[11:51]	Eata Kitty:	You should get some kind of big control center like they have in NORAD
[11:51]	Helena Lycia:	Don't mention pizza, I haven't had dinner yet
[11:51]	Stev Spiegel:	mmmmmm
[11:52]	Andrew Linden:	the boxen are split up into multiple colocation facilities. I think we are split between three different facilities right now.
[11:52]	Moon Metty gave you temp rez limiter.
[11:52]	Kitto Flora:	3 at last count
[11:52]	Kitto Flora:	Hows the LLnet doing?
[11:52]	Eata Kitty:	Oooh yes, LLnet
[11:52]	Andrew Linden:	LLNet is complete. All VPN's are dead.
[11:52]	Eata Kitty:	Sweet
[11:52]	Kitto Flora:	So whens llNet going to Europe?
[11:53]	Eata Kitty:	is there even a datacenter in europe
[11:53]	Helena Lycia:	All the weird sim crossing events seem to have gone away - leaving with the traditional bad sim crossings
[11:53]	Eata Kitty:	There should have been an opening ceremony with some kind of oversized pair of scissors
[11:53]	Simon Linden:	We did notice a definite improvement with the TP success rate upon switching to LLNet
[11:53]	Andrew Linden:	LLNet was a project to move some internal SL traffic off of some hardware accelerated VPN routers to some dedicated lines that link our different colocation facilities to each other directly.
[11:53]	Arawn Spitteler:	I crashed Sunday, and should file a new feature jira on it
[11:54]	Eata Kitty:	Doesn't LLNet enable the HTTP texture plan
[11:54]	Kitto Flora:	So whens llNet going to Europe? - enabling a colo use in Europe?
[11:54]	Andrew Linden:	No, the HTTP texture plan is mostly between the simulators and the viewers (SL clients).
[11:54]	Andrew Linden:	We already move textures about via HTTP internally.
[11:55]	Arawn Spitteler:	It seems odd, that the simulators would ever load the actual textures
[11:55]	Simon Linden:	They load them so they can be sent to the viewer
[11:55]	Simon Linden:	... and that's exactly what we want to remove
[11:55]	Andrew Linden:	Originally the simulators were the main gateway for all info. They would take the textures and encode them into UDP packets.
[11:56]	Moon Metty:	how long ago was that?
[11:56]	Andrew Linden:	We started development in 2000. I don't think we started pulling functionality out of the simulator until... 2005 or 2006. We opened SL in summer 2003.
[11:56]	Simon Linden:	That's they way it still works ... eventually we'll switch to http instead of our messaging
[11:57]	Eata Kitty:	Ah, UDP. Unreliable Datagram Protocol. Or perhaps, Undelivered Datagram Protocol
[11:57]	Moon Metty:	ok
[11:57]	Andrew Linden:	I've got another meeting right after this one.
[11:57]	Stev Spiegel:	ty Andrew
[11:58]	Andrew Linden:	So if anyone has been waiting until the end of the meeting to raise an issue, now is the time.
[11:58]	Andrew Linden:	(I have to say that, or sometimes I get people pinging me in IM right as the meeting ends.)
[11:58]	Helena Lycia:	Is there any technical reason why a llSetGlow function couldn't be implemented that didn't have a 0.2 sec delay?
[11:58]	Arawn Spitteler:	I half crashed, the other day, and I was wondering about recoverability.
[11:58]	Andrew Linden:	0.2 sec delay on llGetGlow()?
[11:58]	Liisa Runo:	how LL choose what will be blacklisted and what not, why we still have tools that create lag, and dialog spam, they been on market for ages, blacklist?
[11:59]	Moon Metty:	well, have a look at the algorithm for temprez limiting when you have time, it's crude but effective
[11:59]	Eata Kitty:	Yes! Glooow. I totally hate changing glow. The delay is a real big pain
[11:59]	Helena Lycia:	Glow can only be scripted via teh SetPrimParms type functions and thereefore have a delay
[11:59]	Arawn Spitteler:	Set Glow would be part of llSetPrimitiveParams() I'd guess, with a delay for physics purposes.
[11:59]	Andrew Linden:	Thanks Moon. I'll try to look it over.
[11:59]	Moon Metty:	:)
[11:59]	Helena Lycia:	Be nice to have delay free functions for glow, like there are for alpha and colour
[12:00]	Andrew Linden:	Hrm... not all llSetPrimitiveParams() changes are expensive, but some are.
[12:00]	Andrew Linden:	I suspect the delay is built to assume the worst case
[12:00]	Helena Lycia:	http://jira.secondlife.com/browse/SVC-4011?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=103647#action_103647
[12:00]	Andrew Linden:	however... the place where the delay is implemented is in a spot where it can't know about whether the operation is expensive or not.
[12:01]	Andrew Linden:	It would be hard to switch based on expensive/cheap operations.
[12:01]	Helena Lycia:	Any called to the Prim Parms functions incurs a delay, regardless of what you do
[12:01]	Eata Kitty:	A dedicated call would be good though
[12:01]	Andrew Linden:	Right. I've felt for a while that some of the built-in delays in script calls should be revisited.
[12:01]	Helena Lycia:	JI've created a JIRA requesting them
[12:02]	Helena Lycia:	(see link in chat earlier)
[12:02]	Andrew Linden:	I wonder now that Mono is working it might be time to do that. I'll have to ask Babbage Linden to see what he thinks.
[12:02]	Helena Lycia:	Cool
[12:02]	Andrew Linden:	Ok thanks for the link Helena.
[12:02]	Andrew Linden:	Alright, I've got to go.
[12:02]	Andrew Linden:	Thanks for coming everyone.
[12:02]	Moon Metty:	thank you Andrew
[12:02]	Eata Kitty:	Byebye
[12:02]	Moon Metty:	see you soon!
[12:02]	Helena Lycia:	THanks ANdrew
[12:02]	Simon Linden:	Thanks everyone for coming
[12:02]	Stev Spiegel:	mmmmmm
[12:02]	Moon Metty:	thank you Simon
[12:03]	Twisted Laws:	bye, thanks
[12:03]	Kitto Flora:	Lunchtime! Byebye all