Chat log from the meeting on 2024-07-09

From OpenSimulator

Jump to: navigation, search
[11:06 AM PT]  Andrew Hellershanks: Hello, everyone.
[11:06 AM PT]  Jagga Meredith: If I use a cabint with a short SCSI cable, it boots fine with multiple partitions coming up.  If I use a longer cable it doesn't.  SCSI terination is on the drive
[11:06 AM PT]  Jagga Meredith: \
[11:06 AM PT]  Gavin.Hird @grid.xmir.org:8002: Hi Andrew
[11:06 AM PT]  Andrew Hellershanks: Wow. Time was moving slowly today. Took a lot longer than usual to get logged in.
[11:06 AM PT]  Ubit Umarov: and it is set on? :)
[11:07 AM PT]  Gavin.Hird @grid.xmir.org:8002: cable too long. usual issues
[11:07 AM PT]  Jagga Meredith: termination on
[11:07 AM PT]  Jagga Meredith: crap
[11:07 AM PT]  Cuga.Rajal @rajal.org:9000: yes you have to keep the scsi cable short, or buy a very expensive one. Same issue as getting 40G/sec from a USB4, you need a special cable
[11:07 AM PT]  Andrew Hellershanks: Jagga, SCSI 1 or 2?
[11:07 AM PT]  Jagga Meredith: probably 2
[11:07 AM PT]  Gavin.Hird @grid.xmir.org:8002: whihc Mac is it?
[11:08 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Good news. This weeks commits include a fix for permissions regarding shared objects and contents :)
[11:08 AM PT]  Jagga Meredith: classic
[11:08 AM PT]  Cuga.Rajal @rajal.org:9000: what OS does that run?
[11:08 AM PT]  Jagga Meredith: 7.5.5
[11:08 AM PT]  Gavin.Hird @grid.xmir.org:8002: it inly has SWCSI 1
[11:08 AM PT]  Gavin.Hird @grid.xmir.org:8002: SCSI 1
[11:08 AM PT]  Cuga.Rajal @rajal.org:9000: why?
[11:09 AM PT]  Gavin.Hird @grid.xmir.org:8002: short cable needed
[11:09 AM PT]  Jagga Meredith: ok
[11:09 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: GetNotecardLineSync got merged thanks to a pull request from Sue Cripter
[11:09 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Some throttling for viewers being a bit aggressive with agent updates
[11:09 AM PT]  Ubit Umarov: ppl where still talking about scsi
[11:10 AM PT]  Jagga Meredith: I got my answer
[11:10 AM PT]  Ubit Umarov: but yeah, accept that conttibution with GetNotecardLineSync code
[11:11 AM PT]  Ubit Umarov: and made a few changes to it
[11:11 AM PT]  Ubit Umarov: seen a fs commit to allow for 125 agent updates per second
[11:12 AM PT]  Ubit Umarov: so added code to ignore more than  -45 per second
[11:12 AM PT]  Ubit Umarov: ofc can't stop the wasted udp
[11:12 AM PT]  Joe Magarac: Wow. That must be for combat regions.
[11:12 AM PT]  Ubit Umarov: sadly seems viwers are sending a lot more
[11:13 AM PT]  Ubit Umarov: depending on their gpu fps
[11:13 AM PT]  Ubit Umarov: if doing 400fps, they send like 300 updates per sec or so
[11:14 AM PT]  Cuga.Rajal @rajal.org:9000: thats crazy
[11:14 AM PT]  Ubit Umarov: ofc that can just force drops of lludp
[11:14 AM PT]  Andrew Hellershanks: Hello, Web
[11:14 AM PT]  Ubit Umarov: ll did reduce to 10/s
[11:14 AM PT]  Ubit Umarov: then got complains from users
[11:15 AM PT]  Gavin.Hird @grid.xmir.org:8002: what did they change it to?
[11:15 AM PT]  Ubit Umarov: their cars and bikes depend on 45 updates/c
[11:15 AM PT]  Ubit Umarov: ll changed to 10/s
[11:15 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: All the while round trip can be as much as 200ms so you get dozens of packets that are potentially no longer relevant on either end
[11:15 AM PT]  Ubit Umarov: max
[11:15 AM PT]  Web Rain: Hi there :)
[11:15 AM PT]  Ubit Umarov: it is not round trip.. is rate
[11:15 AM PT]  Joe Magarac: Right. No point in having more than one status in flight.
[11:16 AM PT]  Ubit Umarov: well their scripts do need 45/s
[11:16 AM PT]  Gavin.Hird @grid.xmir.org:8002: they don't want to be DDOS-ed by update requests
[11:16 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I know, but the rate over the time means you send stuff that might already conflict with information the server got from another source
[11:16 AM PT]  Ubit Umarov: they use rate to control turn speed etc
[11:16 AM PT]  Ubit Umarov: ofc not good scripts..
[11:16 AM PT]  Ubit Umarov: but than SL is not to drive cars and bikes
[11:16 AM PT]  Ubit Umarov: it just 2 bad for that
[11:17 AM PT]  Ubit Umarov: only action at sl is s.. errrr pose balls action :P
[11:18 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I'm all for more capabilities with scripts etc, but combat? Original Doom has better combat features than you can achieve given the restrictions we have to work with. Trying to make elephants fly feels like
[11:18 AM PT]  Ubit Umarov: i told beq that 125 was 2 much.. and.. well got another "glitch in the matrix"
[11:18 AM PT]  Ubit Umarov: so i added limit in out code
[11:19 AM PT]  Ubit Umarov: at least we will drop the packets faster not letting it flood scripts etc
[11:19 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: 125 is an update every 8ms, which has to be processed and turned into data for the next frame, which you only have like 16ms for realistically
[11:20 AM PT]  Ubit Umarov: ofc with current fs with pbr, most ppl dont' have 25 fps..
[11:20 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Pbr, the powerpoint experience
[11:20 AM PT]  Ubit Umarov: bew gave a "lecture about nyquist" etc...
[11:21 AM PT]  Ubit Umarov: that ofc does not apply to this..
[11:21 AM PT]  Ubit Umarov: bahh details..
[11:21 AM PT]  Ubit Umarov: beq..
[11:21 AM PT]  Ubit Umarov: they also ask ll to do the same..  ll did not decide
[11:21 AM PT]  Ubit Umarov: at least last time i looked
[11:22 AM PT]  Ubit Umarov: sl servers ignore all events faster then 45/s
[11:22 AM PT]  Ubit Umarov: so..  125 is...  waste..  but well...
[11:23 AM PT]  Ubit Umarov: in fact that issue kinda shows how bad are thoise viewers keyboard and mouse handlers
[11:23 AM PT]  Cuga.Rajal @rajal.org:9000: seems there should be a slider for that
[11:24 AM PT]  Ubit Umarov: nahh no slider jsut reasonable code
[11:24 AM PT]  Joe Magarac: No, no, fix the problem, don't delegate it to the user.
[11:24 AM PT]  Ubit Umarov: yeah that is not a user option
[11:24 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Sending more than can be processed also means you drop potentially important information randomly, granted those updates are not that important, but still
[11:25 AM PT]  Ubit Umarov: that is crap keyboard and mouse handling, etc
[11:25 AM PT]  Ubit Umarov: actually old code..
[11:25 AM PT]  Ubit Umarov: but i don't remember that flood..
[11:26 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: There are enough potential downfalls from async protocol and now they want to send more updates than can be handled... nah that actually tracks with LL stupidity if you think about it
[11:26 AM PT]  Andrew Hellershanks: There are some people who think we are supposed to do what LL does. ;)
[11:26 AM PT]  Ubit Umarov: well ll did found the issue and did limit to 10 :)
[11:27 AM PT]  Lyr Lobo: It helps to have consistency in features to stimulate feature advancement, but the products are also different
[11:28 AM PT]  Lyr Lobo: Thanks, Ubit
[11:28 AM PT]  Ubit Umarov: ( you can see the talk at https://jira.firestormviewer.org/browse/FIRE-34171)
[11:28 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Free to the mantra losing weight by removing both kidneys and wondering why you are not feeling good
[11:28 AM PT]  Ubit Umarov: ( and https://feedback.secondlife.com/bug-reports/p/directional-input-delays-with-latest-pbr-capable-viewers )
[11:30 AM PT]  Ubit Umarov: thin i didn't update this region
[11:30 AM PT]  Cuga.Rajal @rajal.org:9000: So this could account for sluggishness with the new viewer
[11:31 AM PT]  Ubit Umarov: ohh this is the viewer with the ll limit
[11:31 AM PT]  Ubit Umarov: nahh opensim only sends movement updates also at 10/s
[11:31 AM PT]  Ubit Umarov: with the defauilt settings
[11:32 AM PT]  Ubit Umarov: only good for social regiopns
[11:32 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Speaking of movement...
[11:32 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Anyone played with NPCs recently?
[11:32 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Specifically them walking around
[11:32 AM PT]  Lyr Lobo: No, not even for dancing
[11:32 AM PT]  Joe Magarac: I've had that working in SL for years.
[11:32 AM PT]  Gavin.Hird @grid.xmir.org:8002: nope
[11:32 AM PT]  Ubit Umarov: wiath
[11:33 AM PT]  Andrew Hellershanks: Not recently
[11:33 AM PT]  Ubit Umarov: on code changes also fixed a issue on permissions
[11:33 AM PT]  Joe Magarac: Animesh, though, not OS NPCs.
[11:33 AM PT]  Ubit Umarov: in case of a prim shared to group
[11:33 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I noticed the other day they appear to be walking ever so slightly slower than normal avatars
[11:33 AM PT]  Web Rain: not yet but i will soon play around an NPC teacher... so, if you need testing around let me know
[11:34 AM PT]  Gavin.Hird @grid.xmir.org:8002: they must be getting older
[11:34 AM PT]  Andrew Hellershanks: hehe
[11:34 AM PT]  Ubit Umarov: i meant on permissions on a group shared object inventory :)
[11:34 AM PT]  Cuga.Rajal @rajal.org:9000: I have an NPC cat
[11:34 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Not a big deal, just a bit annoying when you try to follow them around and I can't quite work out why they are doing that
[11:35 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: In most respects they are just avatars, though their movement is controlled by code
[11:35 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I made some tests with async command rates and simfps, but that made no difference
[11:35 AM PT]  Ubit Umarov: they use moveto code
[11:35 AM PT]  Ubit Umarov: same as avatar moveto
[11:36 AM PT]  Ubit Umarov: but not same as keyboard remote ( that also has run )
[11:36 AM PT]  Ubit Umarov: controled by  but does not depend on the rate of the agent updates
[11:37 AM PT]  Gavin.Hird @grid.xmir.org:8002: different walk animations between the two of you?
[11:37 AM PT]  Ubit Umarov: those are viewer side
[11:37 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: The animation should not have bearing on things. I thought maybe it was down to the way they are moved too, but neither keyboard nor mouse move are slow enough
[11:37 AM PT]  Ubit Umarov: well they do not have viewer side overrides etc
[11:38 AM PT]  Ubit Umarov: well npc movement has no changes in years?
[11:38 AM PT]  Gavin.Hird @grid.xmir.org:8002: does the NPC wear a scripted AO?
[11:39 AM PT]  Web Rain: it is possible to turn left/right while walking ?
[11:39 AM PT]  Ubit Umarov: don 't remember..  but server side ao may impact it
[11:39 AM PT]  Ubit Umarov: well script one may also...
[11:39 AM PT]  Ubit Umarov: but that changes the animations..  vincent.Sylvester
[11:40 AM PT]  Ubit Umarov: is talking about speed
[11:40 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I'm thinking this might be related to network as Ubit suspects so I have to make some local tests and with farther away machines
[11:40 AM PT]  Ubit Umarov: guess he did let his npcs eat 2 much :P
[11:40 AM PT]  Lyr Lobo chuckles
[11:41 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: NPCs not walking the same speed is like a cardinal sin in gamedev cause it's rather annoying for long periods of time
[11:41 AM PT]  Ubit Umarov: he should know the can't store them on same inventory as cookies
[11:41 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Question is are NPCs too slow or avatars too fast
[11:43 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I guess I can always try to give them some applyimpulse and push them along heh
[11:43 AM PT]  Ubit Umarov: not sure if any did compare, since the move mechanics is diferent
[11:43 AM PT]  Andrew Hellershanks grins at Vincent
[11:44 AM PT]  Ubit Umarov: ( except if moving the avatar with autopilot etc )
[11:44 AM PT]  Ubit Umarov: hmm servers side autopilot.. there is also a viewer side autopilot
[11:45 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: The movement code has no idea what it moves along so making that aware of NPCs and telling it to go faster seems like a waste of code
[11:46 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: If anything need to find a better place to inject some speed into them
[11:47 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Or perhaps the problem is avatars actually asking for too much movement and the NPCs are behaving properly
[11:47 AM PT]  Ubit Umarov: hmm that may also depend on physics engine..  yeack
[11:47 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Good point, haven't tested that, probably should
[11:47 AM PT]  Cuga.Rajal @rajal.org:9000: I have been seeing more physics sluggishness in recent weeks but not sure why
[11:47 AM PT]  Andrew Hellershanks: NPCs and avatars should be running off the same physics engine.
[11:48 AM PT]  Ubit Umarov: yeasteady or the day bf ppl noticed bad http
[11:48 AM PT]  Ubit Umarov: at osgrid at least..
[11:48 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Internet hayfever
[11:48 AM PT]  Ubit Umarov: guess internet weathere
[11:48 AM PT]  Ubit Umarov: guess internet weather
[11:49 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Pull the plug and drain the excess lol
[11:50 AM PT]  Ubit Umarov: well good thing on pbr viewers is that they do less flood of regions with agent updates :P
[11:50 AM PT]  Web Rain: http is bas... use https instead XD
[11:50 AM PT]  Ubit Umarov: oops the PR waiting :P
[11:51 AM PT]  Web Rain: i did not test PBR yet, i think my old i5 will not follow :(
[11:51 AM PT]  Ubit Umarov: its more gpu
[11:51 AM PT]  Ubit Umarov: that did move some work from cpu to gpu
[11:51 AM PT]  Web Rain: poor nvidia 920m
[11:51 AM PT]  Ubit Umarov: guess you will be fine..  if you like 2fps
[11:52 AM PT]  Ubit Umarov: :p
[11:52 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: It's also unoptimized mess so even on modern cards it fails to utilize the whole thing
[11:52 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: So far couldn't get more than 70% out of it
[11:52 AM PT]  Ubit Umarov: well don't play tihe the fs vram limit
[11:52 AM PT]  Ubit Umarov: think you will get more crashes using it
[11:53 AM PT]  Ubit Umarov: ofc it will use all you gpu ram..
[11:53 AM PT]  Andrew Hellershanks: IIRC, Some years back I had better performance using an Nvidia 8600GT than I got with the 9500GT. Only changed due to the 8600 went bad.
[11:54 AM PT]  Ubit Umarov: gpu performance is a marketing thing
[11:54 AM PT]  Ubit Umarov: well cpus also
[11:54 AM PT]  Ubit Umarov: you need to test for you use case..
[11:54 AM PT]  Andrew Hellershanks nods
[11:55 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Multicore benchmarks when so many things can't utilize more than 4 due to how code has to work to build frames, you can only multithread so much
[11:56 AM PT]  Ubit Umarov: well the multicore benchmarks i have do fill all cores
[11:57 AM PT]  Ubit Umarov: P and E  yeack
[11:57 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: I been pondering about that for a while actually trying to find a solution for maptile zoom levels, something that could do them in parallel, but haven't found something that doesn't cause locking issues
[11:58 AM PT]  Joe Magarac: Talk to me about that offline. I'm rewriting that part of Sharpview now.
[11:58 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: Best lead I have so far is just generating each level on its own thread, but that means loads more memory usage caching the tiles
[11:58 AM PT]  Ubit Umarov: that is bad
[11:58 AM PT]  Ubit Umarov: a level depends on the previus one
[11:59 AM PT]  Joe Magarac: You mean generating tiles, or viewing tiles?
[11:59 AM PT]  Vincent.Sylvester @hg.zetaworlds.com: It generates through. Adding a single tile it then adds it to the next above and so on. They are not stitched together
[11:59 AM PT]  Joe Magarac: Here's a viewer: http://animats.com/sl/map/
[12:00 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: I mean you can generate them in multiple ways technically so long as what comes out is the same
[12:00 PM PT]  Joe Magarac: Hold down CTRL and left mouse to tilt ant rotate.
[12:00 PM PT]  Ubit Umarov: adn why do oyu want many cores on that? the machine only does map??
[12:00 PM PT]  Ubit Umarov: jezz
[12:00 PM PT]  Ubit Umarov: opensim is not like a pc game
[12:00 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: It takes quite a while to generate the levels of 13000 region tiles
[12:01 PM PT]  Ubit Umarov: it is legal for a pc game to take over all the machine resources
[12:01 PM PT]  Ubit Umarov: opensim CAN NOT
[12:01 PM PT]  Ubit Umarov: :)
[12:01 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: I mean yeah, like I said, going too far with multithreading makes little sense
[12:01 PM PT]  Ubit Umarov: so hold your horses on using all the cores on opensim
[12:01 PM PT]  Gavin.Hird @grid.xmir.org:8002: are you generating all on one machine?
[12:02 PM PT]  Ubit Umarov: gavinm it is grid side map
[12:02 PM PT]  Ubit Umarov: it is..
[12:02 PM PT]  Gavin.Hird @grid.xmir.org:8002: yeah, I know
[12:02 PM PT]  Gavin.Hird @grid.xmir.org:8002: but each simulator generates it part of the map
[12:02 PM PT]  Ubit Umarov: actually SL map around sl21b is a total mess
[12:03 PM PT]  Ubit Umarov: old regions there on some zoom levels etc
[12:03 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Yeah I nuke them off disk and then tell robust to redo them. It starts a render thread and after a while they are all back, just takes a bit long for my tastes
[12:03 PM PT]  Ubit Umarov: it is not something one needs to update fast..
[12:03 PM PT]  Gavin.Hird @grid.xmir.org:8002: so you let robust generate all times?
[12:03 PM PT]  Gavin.Hird @grid.xmir.org:8002: tiles?
[12:03 PM PT]  Ubit Umarov: we only need it to finish withing ones lifetime..
[12:04 PM PT]  Ubit Umarov: ;
[12:04 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Yeah it only takes like ten minutes
[12:04 PM PT]  Andrew Hellershanks: hehe
[12:04 PM PT]  Andrew Hellershanks: Vincent, for how many regions?
[12:04 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: 13000
[12:04 PM PT]  Andrew Hellershanks: Doesn't sound too bad.
[12:04 PM PT]  Gavin.Hird @grid.xmir.org:8002: agreed
[12:04 PM PT]  Joe Magarac: Gotta go, next meeting.
[12:04 PM PT]  Andrew Hellershanks: 130/minute
[12:05 PM PT]  Andrew Hellershanks: ok, Joe. Thanks for dropping by.
[12:05 PM PT]  Ubit Umarov: but on that, remember that one level is previus one with half(?) resolution
[12:05 PM PT]  Ubit Umarov: so,.. does depend on it
[12:05 PM PT]  Andrew Hellershanks: We are just past the hour mark. Does anyone have a topic they want to raise before we end todays meeting?
[12:06 PM PT]  Web Rain: yep :)
[12:06 PM PT]  Andrew Hellershanks: ok, go ahead Web.
[12:06 PM PT]  Cuga.Rajal @rajal.org:9000: I have one quick question too when Web os done
[12:06 PM PT]  Web Rain: i published my modified version of NSL/DTL money server with BTCPay support
[12:06 PM PT]  Andrew Hellershanks: ok, you'll be next, Cuga.
[12:07 PM PT]  Web Rain: https://github.com/AdilElFarissi/nsl-opensim-currency-for-btcpay
[12:07 PM PT]  Andrew Hellershanks: BTCPay? Is that something related to Bitcoin?
[12:07 PM PT]  Web Rain: so, if someone need it or need help around let me know
[12:07 PM PT]  Ubit Umarov: is that updated for dotnet 8 ?
[12:08 PM PT]  Web Rain: yes andrew, is a self hosted crypto processor for payments and many
[12:08 PM PT]  Cuga.Rajal @rajal.org:9000: requires PHP7, wont run on 8?
[12:08 PM PT]  Web Rain: yep Ubit >= to 0.9.3
[12:08 PM PT]  Andrew Hellershanks: web, ok
[12:08 PM PT]  Ubit Umarov: ok, ty
[12:09 PM PT]  Web Rain: PHP7 due to the removal of the xmlrpc stuff from 8
[12:09 PM PT]  Cuga.Rajal @rajal.org:9000: hmm ok
[12:09 PM PT]  Andrew Hellershanks: Web, PHP8 isn't able to use phpxmlrpc?
[12:10 PM PT]  Web Rain: no, removed but there is other xml stuff that we can use
[12:10 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Strange I just used that the other day
[12:10 PM PT]  Web Rain: the next on that is converting the server/php part to a BTCPay plugin
[12:11 PM PT]  Andrew Hellershanks: I'll have to to run some tests with the PHP8 I have installed on my machine. I have some modules that use the phpxmlrpc package/module.
[12:12 PM PT]  Ubit Umarov: we do not touch on money modules much less pyramid schemes.. err i mean cryto things
[12:12 PM PT]  Web Rain: functions like xmlrpc_encode/decode are not in 8
[12:12 PM PT]  Web Rain: i know that Ubit ^^
[12:13 PM PT]  Ubit Umarov: just  a reminder for the log :)
[12:13 PM PT]  Gavin.Hird @grid.xmir.org:8002: I need to get going
[12:13 PM PT]  Gavin.Hird @grid.xmir.org:8002: TC all
[12:13 PM PT]  Ubit Umarov: cya
[12:13 PM PT]  Ubit Umarov: cuga ?
[12:13 PM PT]  Andrew Hellershanks: ok, Gavin.
[12:13 PM PT]  Ubit Umarov: had a issue?
[12:13 PM PT]  Cuga.Rajal @rajal.org:9000: regarding floating point jitter on large varregions,
[12:13 PM PT]  Web Rain: may be this is why the opensin cashier ends empty in the end of the year hihihihi
[12:14 PM PT]  Cuga.Rajal @rajal.org:9000: could that jitter be causing physics engine sluggishness?
[12:14 PM PT]  Andrew Hellershanks: Web, anything else you wanted to say?
[12:14 PM PT]  Ubit Umarov: yeap empty but not in jail :P
[12:14 PM PT]  Cuga.Rajal @rajal.org:9000: could there be a link between floating point jitter and physics problems
[12:14 PM PT]  orbert tatham: Ok, guys, I need to bail out also
[12:14 PM PT]  orbert tatham: HAve a good week
[12:14 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: How large a var?
[12:14 PM PT]  Andrew Hellershanks: you too, orbert.
[12:14 PM PT]  Web Rain: no thank you andrew... just arian 6 1st fly was successful ^^
[12:14 PM PT]  Cuga.Rajal @rajal.org:9000: well, 2 cases - 8x8 and 16x16
[12:14 PM PT]  Ubit Umarov: not sure if jitter is the right name
[12:14 PM PT]  Andrew Hellershanks: ok, Floor is yours, Cuga
[12:14 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: oof
[12:15 PM PT]  Cuga.Rajal @rajal.org:9000: sorry if I talked over you Web
[12:15 PM PT]  Ubit Umarov: guess that needs proper debug cuga.Rajal
[12:15 PM PT]  Web Rain: no problem ^^
[12:16 PM PT]  Motoko.Karu @login.digiworldz.com:8002: have a good week everyone....
[12:16 PM PT]  Andrew Hellershanks: you too, Motoko
[12:16 PM PT]  Ubit Umarov: more math means less precision...  but how bad.. depends
[12:16 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: 8x8 is kinda at the edge of what you want for physics stuff. That's a lot of variables going round at that scale, 16x16 is yeah I'd be surprised if that didn't cause some funky collisions
[12:16 PM PT]  Andrew Hellershanks: How much movement at a time, Cuga?
[12:16 PM PT]  Cuga.Rajal @rajal.org:9000: Ubit what is proper debug?
[12:16 PM PT]  Ubit Umarov: ( less precision due to accumulated errors )
[12:16 PM PT]  Ubit Umarov: like repeat the movement situation closer to 0
[12:17 PM PT]  Ubit Umarov: ie withing a standard size
[12:17 PM PT]  Cuga.Rajal @rajal.org:9000: there are a couple different types of problems
[12:17 PM PT]  Cuga.Rajal @rajal.org:9000: one is, in both sizes even the 8x8. after the sim is up for some time I get zones where the engine thinks there is a physical object there, but nothing there
[12:17 PM PT]  Ubit Umarov: er on the 256 square at left/bottom corner,  versus similar corner at right-top corner
[12:17 PM PT]  Cuga.Rajal @rajal.org:9000: nothiong shows with invisibles on or with Area Search
[12:18 PM PT]  Cuga.Rajal @rajal.org:9000: restarting sim fixes
[12:18 PM PT]  Ubit Umarov: that seems unrelated to precision issues
[12:18 PM PT]  Cuga.Rajal @rajal.org:9000: Ubit, I tried testing near 0,0 and near the opposite corner, with similar problems
[12:19 PM PT]  Cuga.Rajal @rajal.org:9000: so not FP accuracy dorectly, but wondering still if size of sim can cause
[12:19 PM PT]  Ubit Umarov: so more unlike to be related to precision issues..
[12:19 PM PT]  Cuga.Rajal @rajal.org:9000: possible to have other physics issues with large varregion, not related to hjitter?
[12:19 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: You are seeing an active physics thread or whatever even though there is nothing physical on the region?
[12:19 PM PT]  Ubit Umarov: did you remember that objects delete is asunc and can take time?
[12:20 PM PT]  Andrew Hellershanks: Should be possible to calculate the resolution you can get using floating point.
[12:20 PM PT]  Ubit Umarov: .. is async..
[12:20 PM PT]  Cuga.Rajal @rajal.org:9000: a few separate issues 1) vehicle movement is very jittery/stuttery for maybe 30 secs
[12:20 PM PT]  Web Rain: in cryptos, the decimals numbers are * 100m before the math and /100m to output
[12:21 PM PT]  Cuga.Rajal @rajal.org:9000: trying to sort ot source of issues
[12:21 PM PT]  Cuga.Rajal @rajal.org:9000: its on someone else's grid using 0.9.3, so not bleeding edge server but pretty new
[12:22 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: The physics enable bounding box is 1024x1024.. well in Bullet at least, though I think they are all the same size.. so regions larger than that when you login to it it creates one, but never deletes it. That's normally fine, it just stays there consuming resources
[12:22 PM PT]  Cuga.Rajal @rajal.org:9000: other physics issues but that was the main one
[12:22 PM PT]  Ubit Umarov: can't guess :)
[12:22 PM PT]  Cuga.Rajal @rajal.org:9000: ok then.. thanks for letting me run it by
[12:23 PM PT]  Cuga.Rajal @rajal.org:9000: Vincent that is likely the issue
[12:23 PM PT]  Cuga.Rajal @rajal.org:9000: I have to restart OS to fix
[12:24 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: It shouldn't cause it to behave slow though. When a physical object comes in, it should just spring back to life
[12:24 PM PT]  Cuga.Rajal @rajal.org:9000: Yeah this is a different issue with physics, not sluggish, jusyt phantom "objectsw" that arenty there
[12:24 PM PT]  Lyr Lobo: but different than their original bounding boxes, right?
[12:25 PM PT]  Cuga.Rajal @rajal.org:9000: Lyr the proble is that the object has been moved or deleted, but the physics engine is acting as if it were still there
[12:25 PM PT]  Cuga.Rajal @rajal.org:9000: you can stand on "it"
[12:26 PM PT]  Cuga.Rajal @rajal.org:9000: but nothing there. restarting sim fixes
[12:26 PM PT]  Lyr Lobo: aw a ghost *teases8
[12:26 PM PT]  Lyr Lobo: the phantom
[12:26 PM PT]  Lyr Lobo: not famliar with that
[12:26 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: This happens with specific object or any prim?
[12:26 PM PT]  Cuga.Rajal @rajal.org:9000: yes but the invisible object blocks moven=ment of vehicles
[12:26 PM PT]  Cuga.Rajal @rajal.org:9000: is a real problem
[12:26 PM PT]  Cuga.Rajal @rajal.org:9000: It happens with any object that "breaks" the physics engine
[12:27 PM PT]  Cuga.Rajal @rajal.org:9000: usually happens when switching a vehicle from Physics on to Physics off
[12:27 PM PT]  Cuga.Rajal @rajal.org:9000: and that switch fails in the engine
[12:27 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: So a specific mesh then
[12:27 PM PT]  Ubit Umarov: wel we do have 2 engines :P
[12:27 PM PT]  Cuga.Rajal @rajal.org:9000: yes
[12:27 PM PT]  Cuga.Rajal @rajal.org:9000: It's likely a Bullet-specific fail
[12:28 PM PT]  Cuga.Rajal @rajal.org:9000: but not predictable when it will happen
[12:28 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Bad mesh can cause a ton of things including that I suppose
[12:28 PM PT]  Lyr Lobo: well let us know how your testing goes. I must dash, my apologies
[12:28 PM PT]  Cuga.Rajal @rajal.org:9000: np, take care Lyr
[12:29 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: You can check via the console whether it actually "sees" an object still there by the way
[12:29 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Area search looks at what the viewer "sees", which is different from the region
[12:29 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Region might have more data the viewer doesn't get
[12:29 PM PT]  Cuga.Rajal @rajal.org:9000: no way to "see" the physics meshes in the region
[12:29 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Only objects
[12:30 PM PT]  Cuga.Rajal @rajal.org:9000: It's my own region that is almost empty, I am the only builder, so I know there is nothing there
[12:30 PM PT]  Ubit Umarov: rl calls
[12:30 PM PT]  Cuga.Rajal @rajal.org:9000: thats a good guuess but this is definitely an object that was there then moved or deleted
[12:31 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: I mean you can, via console, list all objects the region knows about, which should be similar to what's in the database as well, so you can check if it leaves behind a ghost object or if it is further down the pipe in the physics engine
[12:31 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Physics engine might be harder to debug, but I suspect there is a way to add a command for listing the objects it currently has in "cache"
[12:32 PM PT]  Cuga.Rajal @rajal.org:9000: I am pretty sure it is a physics mesh management issue
[12:32 PM PT]  Ubit Umarov: ok need go rl
[12:32 PM PT]  Ubit Umarov: cya nect week, if not before, here or there
[12:32 PM PT]  Vincent.Sylvester @hg.zetaworlds.com: Would be good to know if it was something specific to one of the engines
[12:32 PM PT]  Andrew Hellershanks: As do I.
[12:32 PM PT]  Ubit Umarov: cya next..
[12:32 PM PT]  Cuga.Rajal @rajal.org:9000: I had an object there. Stood on it. I deleted it. Should have fallen down into the water
[12:33 PM PT]  Andrew Hellershanks: On that note I will call the meeting to a close.
[12:33 PM PT]  Andrew Hellershanks: Thank you all for coming. See you again next week.
Personal tools
General
About This Wiki