Chat log from the meeting on 2020-04-21

From OpenSimulator

Jump to: navigation, search

[11:00] Gavin.Hird Hi Andrew
[11:00] Misterblue Waves: not me... I ping him every now and then but he's been missing for a day or two
[11:00] Andrew Hellershanks: Hello, everyone.
[11:00] Selby.Evans Hi Andrew
[11:00] Andrew Hellershanks: Who has been missing?
[11:01] Kayaker Magic: Andrew, do you have any news about Bill Blight?
[11:01] Selby.Evans Bill
[11:01] Andrew Hellershanks nods
[11:01] Andrew Hellershanks: I don't have any informaiton about BIll. I also don't have any contact info for him.
[11:02] Gavin.Hird Hi Arielle
[11:02] Selby.Evans Hi Arielle
[11:02] Arielle Popstar: Hi Gavin
[11:02] Arielle Popstar: HI Selby
[11:02] Kayaker Magic: Since you are here Mister Blue and I've been looking at physics recently, do you know the status of my old mantus 8033?
[11:02] Arielle Popstar: hows everyone
[11:02] Andrew Hellershanks: Hello again, Misterblue. You got here early today. :)
[11:03] Kayaker Magic: When you fire a cannonball on SL or ubODE at 30m/s it flies at 30m/s. But on Bullet, it flew at 60m/s
[11:03] Misterblue Waves: figured I couldn't miss the meeting if I got here early ;-)
[11:03] Andrew Hellershanks: :)
[11:04] Kayaker Magic: You may have fixed that ages ago, but the mantis was never closed.
[11:04] Arielle Popstar: because at 60 m/s he gets here early Kay :)
[11:04] Kayaker Magic: hehe
[11:06] Kayaker Magic: And Ubit, thanks for looking at that new mantis of mine so soon! If you look at the output from that, it seems that cnnonballs slow down as they fly. Does ubODE have air resistance in its physics?
[11:07] Kayaker Magic: I tried changing the friction of my cannonballs in the build dialog, but it has no effect on the slowdown while flying.
[11:08] Arielle Popstar: la la la
[11:08] Andrew Hellershanks: brb
[11:08] Kayaker Magic: <crickets>
[11:09] Misterblue Waves whispers: chirp, chirp, chirp
[11:09] Kayaker Magic: I hope all this silence is caused by people engrossed in looking up the answers to my questions.
[11:10] Gavin.Hird I think Ubit might be afk
[11:11] Misterblue Waves: what viewer should I be using these days?
[11:11] Arielle Popstar: FS 6.02
[11:11] Kayaker Magic: I have another question for Ubit today, should I keep asking them or wait?
[11:12] Arielle Popstar: or Dayturn
[11:12] Gavin.Hird what OS do you use MB?
[11:12] Arielle Popstar: weell still waiting for an answer from Mister B on the 60 ms bullets
[11:12] Kayaker Magic: I would use Dayturn all day long every day, but it doesn't have the pre-processor that I am addicted to in FS :(
[11:13] Gavin.Hird :-))
[11:13] Arielle Popstar: whats the pre processor?
[11:13] Kayaker Magic: Lets you put #include and #ifdef and #define lines in your scripts.
[11:13] Misterblue Waves: I haven't touched BulletSim in quite a while...
[11:14] Misterblue Waves: I'm afraid that if I open that box I'll be lost for quite a while rewriting all my old code
[11:14] Kayaker Magic: To include code off you hard drive, to define new constants, and to select code to build based on those constants.
[11:14] Misterblue Waves: and UBit has done a great job with ubODE
[11:14] Arielle Popstar: lot still use Bullet
[11:15] Arielle Popstar: some things Ubode or its meshmerizer doesnt do well
[11:15] Kayaker Magic: It would be fair of you Blue to respond "Well test it and see if it was fixed and if so mark the mantis closed yourself!"
[11:16] Ubit Umarov: hi all sorry was a bit afk
[11:16] Kayaker Magic: A bit? Is that like being a bit pregnant?
[11:16] Gavin.Hird a bit?
[11:16] Misterblue Waves: actually 8033 looks like a scaling problem so it might not be too hard to fix...
[11:16] Gavin.Hird like only touching the KB with one hand?
[11:16] Ubit Umarov: bill is lost at RL, also afk at irc, last time we talked, a few days ago, he was ok
[11:16] Kayaker Magic: Well, a few days ago is good news!
[11:17] Arielle Popstar: oh good he is ok at least
[11:17] Ubit Umarov: ubode adds a small constant drag, to hep stabilze things
[11:17] Misterblue Waves: I'll have to recreate my build environment... is anyone still using 32 bit OpenSimulator?
[11:17] Arielle Popstar: he still has his grid?
[11:17] Ubit Umarov: only a issue on deep space simulations ( if any notices )
[11:17] Gavin.Hird the Raspberry PI is running 32 bit still, but bullet does not work on it
[11:18] Kayaker Magic: I would like to predict the range of a cannonball, if I use Newton I get a number that is too big, how can I include this drag in my calculation?
[11:18] Gavin.Hird it collides and stuff, but impossible to place object in world from inventory - opegl error
[11:18] Ubit Umarov: think he still has his grid yes
[11:18] Misterblue Waves: you can use the C# Bullet port to have physics on the Raspberry Pi... something Nebadon did
[11:18] Andrew Hellershanks: I'm back.
[11:18] Misterblue Waves: and I'm now running most of my PIs in 64 bit anyway
[11:19] Gavin.Hird well, Raspbian is not 64-bit
[11:19] Kayaker Magic: Is there a 64bit OS for the PI 4 now?
[11:19] Ubit Umarov: gezz EEP is now on sl main viewer
[11:19] Gavin.Hird unless you do a lot of hackaround
[11:19] Arielle Popstar: yes Ubit
[11:19] Misterblue Waves: yes, Raspian is keeping the old PIs alive :)
[11:20] Misterblue Waves: I'd like to build for only 64bit Windows/Linux (Mono) and for IOS
[11:20] Gavin.Hird I am testing a render change form EEP right now
[11:20] Arielle Popstar: wouldnt that mean the regions can do either?
[11:20] Gavin.Hird from*
[11:20] Gavin.Hird iOS - there is no viewer running iOS anytime soon
[11:20] Ubit Umarov: well my eep code is old now, and i can't still move on without a tvp with it ( other than cool vl)
[11:21] Gavin.Hird unless you write a Metal renderer of course
[11:21] Arielle Popstar: doesnt the FS beta have it?
[11:22] Ubit Umarov: no
[11:22] Gavin.Hird have what Arielle?
[11:22] Arielle Popstar: EEP
[11:22] Ubit Umarov: and next release will not have it (eep) of course
[11:23] Arielle Popstar: yes Beq mentioned that last week
[11:23] Gavin.Hird EEP is such a beast to implement
[11:23] Ubit Umarov: i still dont have a updated bin of the release candidate
[11:24] Ubit Umarov: (fs)
[11:24] Ubit Umarov: we should be able to activate httppipeline on it ( and die :p )
[11:25] Ubit Umarov: a damm old thing.. well details
[11:25] Gavin.Hird described as a very unreliable old thing too
[11:25] Gavin.Hird so people have largely abandoned it
[11:25] Ubit Umarov: yeah but now on the limited uses of viewers
[11:26] Gavin.Hird I don't have support for it since there is no support in curl
[11:26] Ubit Umarov: its a problem on http servers.. some suported other didn't and ecah own way
[11:26] Gavin.Hird and I will not add it anyway else
[11:26] Ubit Umarov: same for browsers
[11:26] Ubit Umarov: curl did removed it on newer versions
[11:27] Gavin.Hird because it is unreliable
[11:27] Ubit Umarov: guess ll is stuck on a older one now, bc it needs pipeline
[11:27] Gavin.Hird http/2 is the recommended direction
[11:27] Ubit Umarov: yeah and more useless crap added
[11:28] Ubit Umarov: even more delays..
[11:28] Gavin.Hird ll has http/2 in their viewer, but do they actually use it
[11:28] Ubit Umarov: conns start http1.x then negociate 2.0
[11:28] Ubit Umarov: thats several 100's ms added
[11:28] Ubit Umarov: stupid
[11:29] Ubit Umarov: viewers use http as small messages thing.. those delays matter
[11:29] Gavin.Hird my ping time to ll servers are usually in the 350+ms order
[11:29] Ubit Umarov: many things should had stayed lludp
[11:30] Ubit Umarov: but lots of ppl hate udp, and block it
[11:30] Gavin.Hird you don't HAVE to implement the same stack as LL :-)
[11:30] Ubit Umarov: yeah ? how many viewer devs wiil follow a stack i could do?
[11:31] Ubit Umarov: when they lose udp things we need all the time
[11:31] Misterblue Waves: if they are going to start over, they should look into HTTP3... that's all built on UDP with focus on low startup latency
[11:31] Gavin.Hird how many does now?
[11:31] Ubit Umarov: well many universities and industry, totally block udp
[11:32] Ubit Umarov: non sense... but its how it is
[11:32] Gavin.Hird if they implement opensim inside their FW, they should be albe to handle also UDP
[11:33] Misterblue Waves: we either need a new viewer or to have someone working on multiple protocols for one of the viewers
[11:33] Ubit Umarov: well http keepalive and so some extent, pipeline is critical for this to work
[11:34] Ubit Umarov: but a stress on open files..
[11:34] Ubit Umarov: without pipeline each viewer opens a ton of connections
[11:34] Ubit Umarov: 24 or more
[11:35] Ubit Umarov: and with tcp rules of closing, those stay ocupied for ages since no longer needed
[11:36] Ubit Umarov: well on code changes last week, think nothing ??
[11:36] Ubit Umarov: just more changes on the http low level code
[11:37] Ubit Umarov: i did help fixing that money module, for this changes
[11:37] Ubit Umarov: hope author gets my test code
[11:37] Kayaker Magic: ?What did you fix in the money module?
[11:37] Ubit Umarov: its on mantis
[11:38] Ubit Umarov: it did depend on the dll i removed directly
[11:38] Gavin.Hird did not Melanie write that module?
[11:38] Ubit Umarov: no
[11:38] Ubit Umarov: DTL something
[11:38] Ubit Umarov: mel has her own thing.. somewhere
[11:38] Ubit Umarov: private code i think
[11:39] Andrew Hellershanks: DTL/NSL
[11:39] Ubit Umarov: yes that one
[11:40] Ubit Umarov: since it where my changes, i did look on fixing it
[11:40] Ubit Umarov: in case i would need to add compatibilty code
[11:40] Ubit Umarov: no need
[11:40] Ubit Umarov: its mostly a config change
[11:40] Ubit Umarov: and one line of code change
[11:41] Andrew Hellershanks: Yes, I noticed that. I was looking at the actual code change.
[11:41] Ubit Umarov: was mainly removig references to the dll
[11:42] Ubit Umarov: i did add 2 lines of code
[11:43] Ubit Umarov: to kill the 100-continue need .net adds by default
[11:43] Ubit Umarov: the http server handles it gain.. but its another useless roundtrip delay
[11:44] Kayaker Magic: Here is another question: llSetRegionPos works most of the time, but fails and does nothing at 9000m altitude. llSetPos does work at that altitude. What is going on?
[11:44] Ubit Umarov: my code to handle it was not working bc seems -net expects exact test on the status message.. yeack
[11:45] Ubit Umarov: no idea
[11:45] Andrew Hellershanks: Kayaker, interesting problem.
[11:46] Gavin.Hird at exact 9000 and over?
[11:46] Misterblue Waves: 9000m is a bit high... don't know why one would work and no the other
[11:46] Kayaker Magic: I didn't know ANYTHING worked above 4000m, but had someone call me and say "Your script stopped working!" and when I went to check, I found myself gasping for air!
[11:46] Ubit Umarov: return VecDist(pos, llGetRootPosition()) <= 0.1 ? 1 : 0; yeack ugly
[11:47] Gavin.Hird I was about to say 4096
[11:48] Andrew Hellershanks: llSetRegionPos does some checks on the coordinates supplied. It doesn't like Z to be more than Constants.RegionHeight.
[11:48] Kayaker Magic: That looks like C-- code, is that syntas supported in C#?
[11:48] Ubit Umarov: well more than 4km is suicidal
[11:48] Andrew Hellershanks: The comment says that is 4096.
[11:48] Ubit Umarov: errors do precision do pileup badly
[11:48] Andrew Hellershanks: llSetPos just calls SetPos with out checking the parameters.
[11:49] Ubit Umarov: but see no checks about z
[11:50] Ubit Umarov: thats old code
[11:51] Andrew Hellershanks: Yes. Those limit checks in llSetRegionPos have been there since August of 2015.
[11:51] Kayaker Magic: Uh oh, does that mean if I complain, you are going to 'fix' llSetPos to have the check for Z, instead of fixing llSetRegionPos to work at altitude?
[11:51] Andrew Hellershanks: Kayaker, I was just thinking about that. :)
[11:51] Kayaker Magic: LOL
[11:51] Ubit Umarov: that was var size code changes i guess
[11:53] Kayaker Magic: is Constants.RegionHeight settable in the ini files?
[11:55] Andrew Hellershanks: Hm... it is odd. The llSetRegionPos has a "workaround" if statement that checks for out of bound X and Y values yet the if just below allows X and Y to be as far as 10m outside the region bounds.
[11:56] Kayaker Magic: Probably for crossing borders?
[11:56] Andrew Hellershanks: That first if statement may (or may not) be obsolete as it relates to region crossings.
[11:57] Andrew Hellershanks: After various checks it calls SetPos() the same as in llSetPos.
[11:58] Andrew Hellershanks: There used to be a number of problems related to region crossings. IIRC, they have been fixed now. That first if in llSetRegionPos may no longer be required.
[11:58] Misterblue Waves: hopefully it wasn't a work-around for DSG... there were overlapping regions that sync'ed so area crossings worked
[11:59] Kayaker Magic: Since it is almost the hour, I would like clarification on a previous question: How can I predict that small constant drag in ubODE? Is there a K constant? Is it proportional to v or v^2?
[12:00] Andrew Hellershanks: It would make sense for it to be proportional.
[12:00] Kayaker Magic: In RL, air resistance is proportional to velocity squared.
[12:01] Andrew Hellershanks: Sounds like a question for both Ubit and Misterblue.
[12:01] Kayaker Magic: Well, I'm probably only going to do this in ubODE.
[12:02] Arielle Popstar: back to la la la
[12:03] Andrew Hellershanks: Perhaps the drag coefficient should be 0 when Z > 4k (ie. "outer space").
[12:03] Arielle Popstar: SecondLife unaccesable
[12:04] Ubit Umarov: unaccesable?
[12:04] Gavin.Hird EEP killed the cat
[12:04] Arielle Popstar: There are currently internet-level issues causing problems across Second Life, including logins.
[12:04] Arielle Popstar:
[12:04] Ubit Umarov: better there than here :)
[12:04] Selby.Evans I could not log in either.
[12:05] Misterblue Waves: Pretty sure BulletSim doesn't have any air resistance computation
[12:05] Kayaker Magic: Does this effect OpenSim grids?
[12:05] Ubit Umarov: well issues happen all the time
[12:05] Misterblue Waves: there must be issues all over the Internet... Github and other places have been slow this morning
[12:05] Arielle Popstar: if it an internet level issue then would suppose it might affect some of us here too
[12:05] Ubit Umarov: i was using the eep release canditate wne it worked
[12:06] Arielle Popstar: got in that time
[12:07] Gavin.Hird my ping time to is time=1.202 ms now
[12:08] Andrew Hellershanks: My ping time is averaging 104mS to
[12:09] Andrew Hellershanks: Ubit, did you see the question about whether the drag coefficient is proportional to velocity or v^2?
[12:10] Kayaker Magic: <crickets>
[12:11] Andrew Hellershanks: Kayaker, You will have to wait for another time to get an answer to that question.
[12:11] Andrew Hellershanks: Any other questions or comments for today?
[12:11] Ubit Umarov: i don't remember how that drag is
[12:12] Andrew Hellershanks: Ubit, np. That will be your homework assignment for next week. ;)
[12:12] Kayaker Magic: LOL
[12:12] Ubit Umarov: just that it is there to help things to stop
[12:13] Ubit Umarov: and drain out energy from the simulation
[12:13] Andrew Hellershanks: Depending on the projectile, there is gravity to bring the items down to the ground which will help stop them.
[12:13] Ubit Umarov: gravity is a conservative force
[12:14] Ubit Umarov: groud, depends on object friction settings
[12:14] Andrew Hellershanks: Unless you have a very large var I don't think the drag force would be enough to stop some fast moving items before they reached the far side region boundary.
[12:15] Kayaker Magic: I'm having trouble with the sideways velocity, not the gravity. The sideways velocity decreases during flight so the ball lands before my predicted time.
[12:15] Ubit Umarov: i don't remember its strengh either :)
[12:15] Kayaker Magic: Actually, the time predicted is correct, it is the distance traveled that is less than my prediction
[12:16] Ubit Umarov: behavior on vehicles can be confusing
[12:16] Kayaker Magic: Not a vehicle, just a tossed prim.
[12:16] Andrew Hellershanks: The drag force should also be proportional to the cross-sectional area of the projectile in the direction of movement but that could be hard to calcualte.
[12:16] Ubit Umarov: no, no object shape taken in consideration ever
[12:17] Selby.Evans MUst go -- bye all
[12:17] Andrew Hellershanks: ok, bye Selby.
[12:17] Gavin.Hird be safe Selby
[12:17] Andrew Hellershanks: Ubit, I didn't think it would be but it is a factor in RL.
[12:18] Ubit Umarov: complex thing in rl
[12:18] Kayaker Magic: Yeah, I don't need RL accurate physics, I just want to be able to predict what the in-world physics will do, and just predict the distance a ball will fly.
[12:18] Andrew Hellershanks: Yes, very. The simplest way to handle it in OS is to vary the drag based on the size of the bounding box of the projectile.
[12:18] Ubit Umarov: even with competers and good models, ppl stil use wind tunnels to be sure
[12:18] Ubit Umarov: computers..
[12:18] Andrew Hellershanks nods
[12:19] Ubit Umarov: or water things for boats and subs :)
[12:19] Ubit Umarov: tanks
[12:20] Andrew Hellershanks: Right. The drag for a boat would be a lot higher than something moving through air.
[12:20] Andrew Hellershanks: It is now 20 past. Anything else for today?
[12:21] Andrew Hellershanks: Not seeing anyone typing, I'll take that as no.
[12:21] Kayaker Magic: I'm good for today.
[12:21] Andrew Hellershanks: That will do it for another week. Thank you all for coming. See you again next week.
[12:22] Gavin.Hird be safe all
[12:22] Kayaker Magic: Thanks to the devs, and buy all!
[12:22] Arielle Popstar: waves

Personal tools
About This Wiki